Home

Content - KU Leuven

1. discussion of the results Figure 4 3 4 4 and 4 5 represent the relationship between serial communication telnet and SNMP left graphs It also shows the influence of the command length and the duration of the execution time right graphs GET operation SNMP is the best communication method to get information of the switch Telnet can be used as well when the commands are short It is recommended to avoid serial communication The first step taken to explain these differences is to take a glance at the overhead The speed of serial communication is 9600 bps and has 2 bit overhead to 8 bits data This is the start and stop bit A parity bit is not used in this test Telnet packets flow at a higher speed 100Mbps in this situation The speed gain is less than 100 000 000 9 600 because telnet has more overhead To send 1 frame telnet needs 90 bytes Another difference is the protocol being used Telnet uses TCP while SNMP uses UDP That s why SNMP has to deal with less overhead 66 bytes frame Every command is small enough to fit in just one frame So the overhead is not the main reason for these speed differences The fact that TCP is connection ori nted and UDP is connection less should be a better explanation TCP takes care of acknowledging every octet This is done by seq and ack flags which slows down the communication Concerning the length of a command we expect that serial communica
2. rijen efficientie SGESV Core2Duo E6300 o 1 86 GHz 100 5 90 80 5 70 MATLAB 60 50 4 GotoBlas2 40 Ref LAPACK 30 4 MKL 20 4 10 4 0 2000 4000 6000 8000 10000 12000 rijen efficientie DGESV Xeon E5506 2 13 GHz 100 5 90 5 80 5 70 MATLAB 60 50 GotoBlas2 40 Ref LAPACK 30 4 MKL 20 10 5 0 T T T r i i 0 2000 4000 6000 8000 10000 12000 rijen efficientie 100 90 80 70 60 50 4 40 4 30 20 4 10 4 SGESV Xeon E5506 2 13 GHz MATLAB GotoBlas2 Ref LAPACK MKL 0 2000 4000 6000 8000 10000 12000 rijen Improving audio quality for hearing aids P Verlinden Katholieke Hogeschool Kempen peter verlind gmail com S Daenen NXP Semiconductors steven daenen nxp com P Leroux Katholieke Hogeschool Kempen paul leroux khk be Abstract Since hearing problems are becoming more frequent these days the necessity for high quality hearing aids will grow In order to achieve high audio quality it is necessary to use a good audio codec Nowadays there are a lot of high quality audio codecs but because the target application is a hearing aid some limitations need to be taken into consideration such as delay and hardware limitations This is the reason why a low com
3. NxH2180 Line out SBC Dec 12S FIGURE VI BLOCK DIAGRAM DEVELOPMENT BOARD In this diagram three important components can be distinguished The Codec is an ADC DAC it s is used to convert the analog signal to a digital signal and vice versa The NxH1210 will encode the audio the NxH2180 will decode the audio So the audio comes from the line in and goes to the codec Then it goes through the NxH1210 to be encoded After that the encoded signal is sent to the NxH2180 and is decoded In the final stage it is sent back to the codec then to the line out The Philips Subband Encoder is programmed such that it s easy to test different configurations of the Philips Subband Coder A number of different parameters can be set the number of subbands 4 or 8 the block size 4 8 12 16 and the bitpool size Other than that it is also possible to select in which way the audio is encoded Four choices are available Mono only the left or right channel is encoded Dual or stereo these modes are quite similar both the left and nght channel are encoded Joint stereo when this is selected left and right channel are encoded But information that is the same in both channels is encoded only once so this should get the best results In this setup with one development board the bitrate is limited to the bitrate of PS this is the bus used to transfer the audio sam
4. T T da gt y 16 32 64 128 Fig 8 NLMS influence of the factor a on the SNR gain SNR gain dB 40 35 30 25 7 20 4 15 8 16 32 64 128 Fig 9 RLS SNR gain After choosing the adaptive algorithm the goal of the last experiment is to decide which beamformer sum and delay beamformer or GSC is suitable to suppress noise and reverberation and to see what is the effect of adding more microphones and increasing the distance d between 2 microphones in a microphone array We achieved this by simulating the following microphone arrays e Array with 2 hypercardioid PZMs and a distance of 0 024 m between 2 adjacent microphones e Array with 4 hypercardioid PZMs and a distance of 0 024 m between 2 adjacent microphones e Array with 6 hypercardioid PZMs and a distance of 0 024 m between 2 adjacent microphones e Array with 2 hypercardioid PZMs and a distance of 0 072 m between 2 adjacent microphones For a microphone array with 2 microphones we have to generate 2 input signals To obtain the simulated signals of the microphone array we record a reference signal with the close talk microphone in the following scenario reverberant room veranda with raised curtains ambient noise from a nearby fan of a laptop sample frequency of 48 kHz a 16 bit resolution test subjects with a normal voice and no functional constraints speaker in front of the array Next we simula
5. x Ya Ze a J n n end When a blob moves only the value of the moving blob has to change in equation 1 This results in a movement of the midpoint Therefore the object has to move equal to the movement of the midpoint In Silverlight 3 0 we can use the code below to calculate the midpoint of all points foreach KeyValuePair lt int Point gt origPoint in origPoints totalOrigxXPosition origPoint Value X total rigYPosition OrigPoint Value Y double commonOriginalXPosition totalOrigXPosition origPoints Count double commonOriginalYPosition totalOrigYPosition rig oints Count Point commonOrigPoint new Point commonOrigXPosition commonorigyPosition e O Figure 6 Pan drag hand 4 Enlarge Shrink When we speak about multi touch most people think about the resizing or enlarging and shrinking of an object see Figures 7 and 8 by using two points moving from or towards each other If there are only two blobs in the object we can measure the distance of the two points by eguation 2 dist 4 x2 x ya Yi D If there are more than two blobs in the objects we first need to calculate the midpoint by equation 1 We then have to determine the sum of the distances of all the points to the midpoint So by every movement of a blob we only need to calculate the distance of the blob to the midpoint and change it with his previous value in the sum In Silverlight 3 0 we
6. D Limitations The blocking matrix in the GSC gives several limitations These limitations are e Reduction of noise in the steer direction Due to the spatial 0 the noise coming from the same direction as the speech is not suppressed e Signal Leakage Through reverberation the speech can come from a direction other than the steer direction In this case the speech will be suppressed Voice activity detection 10 11 is required IV EXPERIMENTS AND RESULTS The goal for the first experiment 1s to find the most suitable microphone for speech recognition by handicapped persons For this experiment we consider two different recording scenarios The first set of recordings were made in a laboratory setting and have the following characteristics a reverberant room ambient noise from a nearby fan of a laptop and test subjects with a normal voice and no functional constraints The test subjects receive a list with 72 commands which must be spoken out The recordings were made with a sample freguency of 48 kHz and a resolution of 16 bit To pick up the speech we use different microphones 4 hypercardioid PZMs at the corners of the room 1 omnidirectional lavalier 1 cardioid handheld at a distance of 80 cm 1 close talk and a commercial microphone array at lm in front of the speaker The setup for the first set of recordings can be seen in figure 4 The second set of recordings figure 5 were made in a real life setting the livin
7. DRIVE who gave me the opportunity to work and learn on their new and old server infrastructure I also wish to acknowledge Ward Vleegen and Jan Stroobants for their support in my research to the different applications that had to be migrated in the Flanders DRIVE company and especially the application conducted in this paper ActiTime Thanks are also placed for Tom Croonenborghs who coached me through the whole process and gave help and advice to write this paper REFERENCES 1 J J Cuadrado Gallego Implementing software measurement programs in non mature small setting Software process and product measurement 2008 pp 162 2 http Tomcat Apache org 3 V Vaswani Maintenance backup and recovery The complete reference of mysql 2004 pp 365 4 http www actitime com 5 M Bond D Law Installing Jakarta Tomcat Tomcat kick start 2002 pp 25 6 M Kofler Microsoft office openoffice staroffice The definite guide to mysql 5 pp 120 121 Determine server W install application to 7 M Devos Onderzoek naar een nieuwe IT infrastructuur 2010 8 Apache Tomcat 6 startup error available at http www iisadmin co uk p 22 9 J Kelbly M Sterling A Stewart Introduction to hyper V Windows server 2008 Insiders guide to Microsoft s Hypervisor 2009 pp 1 4 10 T Cerling J Buller C Enstall R Ruiz Management Mastering Microsoft Virtualization 2009 pp 69 11 A Velte J A Kap
8. ensures that the bulk of the investments in the SaaS infrastructure isn t totally exploited In these circumstances a maximum of 25 users can use the SaaS application without a user experiencing errors Optimum use of the SaaS application IOS Mapper would allow even less than 5 users The generation of a report takes an average of 50 seconds and the opening of a wizard lasts 11 seconds as shown in figure 9 As explained above in II RESPONSE TIME a user will shut down the SaaS application IOS Mapper and won t make use of it anymore leading to commercial loss Response time ms 5 user 50266 Fig 9 Response time 5 simultaneous users A real life multi users profile of IOS International is shown in figure 10 At the moment there are 10 editor users and 90 viewers Fig 10 A real life multi users profile After bringing the thinking times into account we get the following response times as shown in figure 11 Response times ms 10 editor users and 90 viewers 40000 36176 35000 30000 26061 25000 20000 15000 9758 10000 6640 5905 _ 5000 943 942 5 28 NY KU xe NN AG AN A RO ae RA re x x x EN KS RN amp we 9 jd amp amp y Y 9 A XY N 2 Ey amp Y o amp lt amp Fig 11 Response times ms 10 editor users and 90 viewers We see that the response time of the report template t
9. P lt port number ActiTime lt actitime_data sql A short explanation what to fill in e lt username gt amp lt password gt see previous section e lt port number gt The port number that is used to access the SQL database e The variable before the lt sign specifies wich database is used e The variable after the lt sign specifies the database file in our case this is the file we ve created in the previous section To execute this command the Windows command prompt 1s used en C AWINDOWS system32 cmd exe Microsoft Windows XP versie 5 1 2600 lt C Copyright 1985 2661 Microsoft Corp C Documents and Settings administrator gt cd C Program Files MySQL MySQL Server 5 1 bin C Program Files MySQL MySQL Server 5 1 bin gt mysql u P 3306 actiti me lt actitime_data sq n U Figure 2 3 command prompt example of how to insert the userdata into the new ActiTime database The last step is to restart the Tomcat server When the Tomcat server is reset the ActiTime application can be used A test to see if the ActiTime application is working like before is made and we check to see if no data is lost All tests turn out positive and the installation of the application in the business network can start E ActiTime installation on the business network The next step is to implement the ActiTime application in the operational network A simple Windows xp machine is used to test the application in the n
10. The ACP Azimuth Change Pulse bit is set when the radar has rotated a given angle Every time the ACP bit has been set the ACP counter is incremented The value of this counter is used to check where the radar is pointing at The number of ACP pulses per rotation determines radar precision A common value is 4096 which gives a radar precision of 0 087 per impulse The ARP Azimuth Reference Pulse bit is set when the radar has reached a reference point e g North This pulse resets the ACP counter 3 We can use this byte stream to display the raw video in an intensity graph Fig 1 where the intensity represents target clutter or noise power Nm 25 99 Z e 25 90 25 85 25 80 25 75 Bei x 25 65 25 60 2e 25 50 25 45 25 40 25 36 7 U U U U U U U U U U U U I 44 79 44 00 43 00 42 00 41 00 40 00 39 00 38 00 37 00 36 00 35 00 34 00 33 00 32 27 Azimuth deg Fig 1 Intensity graph of PSR Raw Video single target B Secondary Radar Digital Data Digital data is stored in proprietary RASS S6 data fields consisting of 128 bytes where each byte or set of bytes represents a property of the target An example of a RASS S6 data field is given in Figure 2 3335345269 7422 Timestamp 6 231371 X Nm 0 FL STAT 14 rargetiD 39 343330 Y Nm 0 000000 Elev 14 Track Nr 300 000000 Velocity P Res 2 1 Scan Nr 0 000000 Heading P Res 3 el 1 Code 0 PSR Refl 0 Res 4 el 2
11. The specifications are not fully listed in this paper It is important to note that you choose the virtual network adapter by the network preferences A network adapter is highly recommended because we want full network access for our employees to reach the server through a web browser intranet With the virtual network adapter it is possible to register the virtual machine in the business network With the use of a virtual network adapter the virtual machine act as a real machine that is connected to the business network C Conflicts between Hyper V and Trend micro When the new virtual machine is created and we turn the machine on a little problem comes up The machine turns itself off after a while with an unknown reason After some search was done a possible solution for this behavior can be found The explanation of the problem can be found by the Trend Micro real time scan which is in use in the whole company Trend Micro is configured to scan the whole hard disk of the Windows server machine The directory of the virtual hard disk file needed for hyper V where our virtual OS is stored is scanned by Trend Micro s real time scanning Since this directory is scanned by trend micro the vhd virtual hard disk file is also scanned When the vhd file is scanned hyper V prevents us to create or start new virtual machines 12 Hyper V stops all the virtual machines that are created and are scanned through the trend micro real time scan
12. the bandwidth needed for VoIP will increase as well Depending on what security measures or the kind of tunneling used the overhead will also increase For example Using a virtual private network IP security will add 50 to 57 bytes of overhead Considering the small size of a voice packet this amount of overhead is a significant amount All these factors codec choice data link overhead sample size have positive and negative impacts on the total bandwidth To calculate the total bandwidth that is needed we must consider these contributing factors as part of the equation 2 More bandwidth required for the codec requires more total bandwidth More overhead associated with the data link requires more total bandwidth Larger sample size requires less total bandwidth RTP header compression requires significantly less total bandwidth RTP defines a standardized packet format for delivering audio and video over the internet It includes a data portion and a header portion The header portion is much larger than the data portion since it contains an IP segment UDP segment and a RTP segment Standard 40 bytes of overhead uncompressed and 2 to 4 bytes compressed Considering these factors the calculation to calculate the total bandwidth reguired per call is done with the following formula 2 Total_Bandwidth Layer2_overhead IP_UDP_RTP_overhead Sample_Size Sample_Size Codec_Speed Meaning if we use a G
13. y A TCP IP Application Virtual NIC Physical NIC Virtual network Switch u Company network T A uane NIC Figure 4 2 Scheme of a virtualized computer B Installation of the Hyper V terminal Installation of the Hyper V terminal on the Windows server 2008 product is very straightforward The installation of the hyper V terminal can be found in the roles section of the Windows server 2008 product 10 First open server manager and click on the roles option Click on the add roles link and an installation wizard is shown Mark the hyper V role and click next An illustration of where to find this role 1s given in figure 4 3 Add Roles Wizard x 5 Select Server Roles Before You Begin Select one or more roles to instal on this server Roles Description Hyper V provides the services that ar ona este nana Progress vetual machines and ther resources Fie Action View Heb A Active Directory Federation Services Each virtual machine i a vetualzed la of Gi ae Active Directory Lightweight Directory Services computer system that operates in an Active Directory Rights Management Services oled execution environment This Server Manager IMPREZA E Appication Server ddr operating 5 En k More about server roles eem cms Figure 4 3 installation of the hyper V role Next you can specify the virtual machine specifications 1 1
14. 08 68 16 20 5 10 5 efficientie oa o Table 1 Intel microprocessor export compliance metrics 0 2000 4000 6000 8000 10000 12000 rijen The value of A ppy 18 an estimation of the number of units Figure 2 Efficiency results of LAPACK By the use of SIMD Single Instruction Multipe Data instruction has a processor the ability to do processing in parallel and do not have real FPU s anymore Depending on the architecture some constant values that are more or less correct are agreed upon When using floating point precision 4 bytes in stead of double precision 8 bytes the processor can handle twice as many datainstructions because of the bytesize Concerning the efficiency results lets have a look at Figure 2 The conclusion of Figure 1 is definitely confirmed and now we see more clearly that the MKL library has a better performance than GotoBlas2 There is also a remarkable conclusion about GotoBlas2 when you look at all the figures Appendix A On older architectures GotoBlas2 is better than MKL on newer architectures with more cores and larger caches GotoBlas2 is less performant but also it is degrading when We will test the performance of the mentioned solvers on Aa the matrix size rises different CPU architectures of Intel as these are a good representative of the x86 family CPUs on the market today Chosen architectures are For the C implementation of FS LSSVM we will use Net
15. 1952 Wilkes and Stringer wrote In some cases it may be possible for two or more micro operations to take place at the same time 3 nowadays parallel processing is being used to create more processing power By using parallel processing one could highly increase the speed of execution The idea behind this technology is the fact that a code and commands can be written in such a way that processes can be divided into smaller parts which then can be run simultaneously on different processors It is as plain as day that as long as the 2 SCK CEN Mol Boerentang 200 B 2400 MOL process can be segmented into more and smaller pieces without creating too much overhead and interconnections it will be better to use as many resources as possible Because supercomputers were often quite expensive only large companies or institutes were able to acquire such processing power The solution to this problem was clustering many low end cpu s together By doing so one could create a system with processing power that could be compared to supercomputers but the cost was considerable less The introduction of clusters within a workplace not only results in a gain of performance and memory but also in more complexity Once multiple users started using the same resources there had to be some sort of way of organizing the distribution of those shared resources by using algorithms Those algorithms are comparable with the scheduling algorithms used in the ker
16. 32 890 01 34 574 01 36 782 01 33 938 94438ms o 1434ms s l 00 02 781 00 02 719 00 03 297 00 02 641 00 02 735 00 02 828 00 02 984 00 03 000 00 03 188 00 02 563 2874ms o 226ms s s 00 02 922 00 02 312 00 03 203 00 02 328 00 02 234 00 02 391 00 02 172 00 02 297 00 02 407 00 02 297 2456 o 316ms SNMP SET wait o 143ms o 143ms l 1 01 37 859 01 33 858 01 33 878 01 35 053 01 34 490 01 33 251 01 32 273 01 34 693 01 33 755 01 33 189 94230ms o 1431ms Is 01 35 425 01 35 374 01 34 577 01 34 375 01 35 519 01 35 594 01 35 955 01 36 250 01 34 688 01 34 780 95254ms o 590ms s 1_ 00 01 64 00 01 516 00 01 797 00 01 954 00 01 687 00 01 735 00 02 031 00 01 703 00 01 797 00 01 703 1756ms o 141ms s s 00 01 484 00 01 390 00 01 532 00 01 594 00 01 672 00 01 625 00 01 563 00 01 609 00 01 578 00 01 578 1562ms o 75ms Serial SET no wait I 01 01 985 01 02 563 01 02 110 01 01 735 01 02 750 01 02 735 01 02 703 01 02 360 01 02 016 01 02 485 62344ms o 343ms l s 00 57 828 00 56 905 00 57 388 00 58 719 00 57 587 00 57 063 00 56 987 00 57 468 00 57 785 00 56 938 57467ms o 530ms s l 00 51 924 00 50 157 00 51 748 00 50 447 00 50 563 00 50 453 00 50 3
17. 329 00 33 375 00 33 641 00 33 032 00 33 016 33393ms o 556ms Telnet GET I 00 11 140 00 11 531 00 11 093 00 11 078 00 11 171 00 10 906 00 15 046 00 10 812 00 11 000 00 10 906 11468ms o 1207ms s s 00 03 844 00 03 266 00 03 578 00 03 469 00 03 359 00 03 469 00 03 438 00 03 390 00 03 297 00 03 406 3452ms o 156ms SNMP GET 1 00 02 562 00 02 312 00 02 062 00 02 187 00 02 277 00 02 043 00 02 168 00 02 183 00 02 355 00 02 248 2240ms s s 00 02 656 00 02 890 00 02 641 00 02 719 00 02 766 00 03 109 00 02 875 00 02 593 00 02 812 00 02 812 2787ms Serial SET wait Ll 01 36 860 01 36 728 01 36 681 01 37 075 01 35 920 01 38 108 01 34 218 01 37 672 01 38 891 01 36 282 96844ms o 1210ms Is 01 36 110 01 36 343 01 36 374 01 37 611 01 36 788 01 38 656 01 37 131 01 36 625 01 36 335 01 36 140 96811ms o 758ms sl 00 07 469 00 07 016 00 07 266 00 07 563 00 07 017 00 07 313 00 07 391 00 07 157 00 07 017 00 07 220 7243ms o 185ms s s 00 05 641 00 06 375 00 05 860 00 06 688 00 05 922 00 06 000 00 06 063 00 05 906 00 06 062 00 05 922 33393ms o 278ms Telnet SET wait Ll 01 35 048 01 38 954 01 33 673 01 33 298 01 32 967 01 33 827 01 33 717 01 33 171 01 33 546 01 33 406 94161ms o 1687ms Is 01 34 375 01 33 780 01 35 955 01 34 547 01 34 201 01 33 335 01
18. 3V voltage signal This resembles the input range for the analog input of the MAC data logger which is 0 to 3V with a precision of 0 01V This setup was calibrated to give a 3V output voltage for an input voltage of 118mV 118mV would resemble the maximum output of the sensor at 1500W m when we confirm that the sensor is linear This would also give that precision of 0 01 V complies to 5W m C Results To determine the linearity of the sensor we need to calculate the correlation coefficient of the correlation between the data from the spectrometer and the sensor We downsampled the data from the spectrometer the sample rate of the sensor data being 1 sample per minute The plot of the 2 signals can be seen in Figure 1 Comparison Avantes spectrometer blue Cellsol 200 red 600 500 Re DD Oo LI Oo Oo 200 100 8 5 9 9 5 10 105 11 115 12 125 B 135 absolute time T hour Figure 1 The correlation coefficient between the 2 signals was calculated to be 92 8 which indicates that there is a large linearity between the 2 signals However is the sensor signal on average 25 larger than the spectrometer signal This is probably the result of a calibration error However this is less important because the calibration for every sensor is different just as the efficiency of every PV setup will be different And so they all need to be calibrated after installation Secondly we compared the sensor data with the instanta
19. 729 codec 40 byte sample size using Frame Relay with Compressed RTP it would result in Total_Bandwidth 6 2 40 40 8 000 9 600 bps If we would have no RTP compression it becomes Total_Bandwidth 6 40 40 40 8 000 17 200 bps When we take the utilization of VAD into account on both examples Total_Bandwidth 9 600 35 6 240 bps Total_Bandwidth 17 200 35 11 180 bps This shows us the great advantage of using the G 729 codec that supports VAD B Configuring Analog Ports For a long time analog ports were used for many different voice applications such as local calls PBX to PBX calls on net off calls etc Now that we only work with digital phones we only connect our fax machines to the analog ports Faxes are something completely different as to making a simple telephone call Fax transmissions operate across a 64 kbps pulse code modulation PCM encoded voice circuit In packet networks on the other hand the 64 kbps stream is in most cases compressed to a much smaller data rate This is done by using a codec that is designed to compress and decompress human speech Fax tones deviate from this procedure and therefore a sort of relay or pass through mechanism is needed There are three available options to operate fax machines in a VoIP network 2 1 Fax relay The fax bits are demodulated at the local gateway the information is send across the voice network using the fax relay protocol
20. A more comfortable microphone to wear is the lavalier microphone This microphone is clipped on the clothes Other microphones which are not attached to the human body are a handheld microphone and PZM The handheld microphone can be brought close to the mouth but in this case we have to take the microphone in hand This isn t suitable for handicapped persons So we can place the handheld microphone on a stand but this might results in a larger distance between speaker and microphone The PZMs are placed on the four walls of a room For the commercial microphone array We make a similar remark regarding the distance Finally every microphone has a polar pattern This pattern can be omnidirectional cardioids hypercardioid or bidirectional While an omnidirectional pattern records every sound 360 the other patterns record the sound in a narrower band The acquisition system is also composed out of a recorder This recorder must have the following requirements e A sample frequency of 8 kHz or more e A resolution of 16 bit or higher e Able to record more than 4 channels synchronous e Able to record the picked up speech of each microphone on a separate track Due to this last requirement we can analyze the data for each microphone individually Il GENERALIZED SIDELOBE CANCELLER The GSC is used to reduce the noise in a speech signal It consists of 3 parts a sum and delay beamformer a blocking matrix and an adaptive algorithm
21. B C D Periodic 0 21 1 08 2 61 2 04 Random 4 01 6 88 8 75 2 61 Table 2 Additional SNR gain in dB for the different microphone arrays tested on the GSC algorithm under the presence of periodic noise array with 2 microphones and d 0 024 A array with 4 microphones and d 0 024 m B array with 6 microphones and d 0 024 m C array with 2 microphones and d 0 072 m D Column L gives the used filter length for LMS with a convergence factor equal to 0 01 L A B C D 2 252 17 18 8 75 11 48 4 321 36 14 15 11 24 01 8 6 41 39 29 28 77 37 49 16 12 76 37 00 36 55 36 82 52 24 68 34 36 34 21 34 26 64 31 41 31 55 31 47 31 50 Table 3 Additional SNR gain in dB for the different microphone arrays tested on the GSC algorithm under the presence of random noise array with 2 microphones and d 0 024 A array with 4 microphones and d 0 024 m B array with 6 microphones and d 0 024 m C array with 2 microphones and d 0 072 m D Column L gives the used filter length for LMS with a convergence factor equal to 0 01 L A B C D 2 0 01 0 18 0 26 0 01 4 0 02 0 19 0 28 0 01 8 0 02 0 19 0 28 0 01 16 0 02 0 19 0 28 0 01 32 0 02 0 19 0 28 0 01 64 0 01 0 19 0 27 0 01 of the sum and beamformer from the gain of the GSC Where Table 2 shows the results for periodic noise Table 3 visualizes the results for random noise The last experiment showed that the sum and delay beamformer might offer a good solution to reduce random noise This ca
22. De Boeck for his scientific assistance REFERENCES 1 Net SNMP v6 0 0 Available at http search cpan org dist Net SNMP 2 Net Ping 2 36 Available at http search cpan org smpeters Net Ping 2 36 lib Net Ping pm 3 Net Telnet 3 03 Available at http search cpan org jrogers Net Telnet 3 03 lib Net Telnet pm 4 libnet 1 22 Available at http search cpan org gbarr libnet 1 22 Net FTP pm 5 Regular expression operations Available at http docs python org library re html module re 6 pysnmp 0 2 8a Available at http pysnmp sourceforge net 7 ping py Available at http www g loaded eu 2009 10 30 python ping 8 telnetlib Available at http docs python org library telnetlib html 9 ftplib Available at http docs python org library ftplib html 10 SNMP library 1 0 1 Available at http snmplib rubyforge org doc index html 11 Net Ping 1 3 1 Available at http raa ruby lang org project net ping 12 Net Telnet Available at http ruby doc org stdlib libdoc net telnet rdoc classes Net Telnet html 13 Net FTP Available at http ruby doc org stdlib libdoc net ftp rdoc index html 14 SNMP4j vl v2c Available at http www snmp4j org doc index html 15 telnet package Available at http www jscape com sshfactory docs javadoc com jscape summary html 16 SunFtpWrapper Available at http www nsftools com tips SunFtp Wrapper java 17 ASocket h ASocket_i c ASocketConstants h Available at ftp
23. In figure 1 we see a scheme of the GSC where the inputs y will be the signals picked up by the microphones and the output is the enhanced speech signal Each of the 3 parts is explained next A Sum and Delay beamformer A beamformer is a system which receives sound waves with a number of microphones All these sensor signals are processed to a single output signal for achieving a spatial directionality Due to the directionality a beamformer can be used for 1 limiting reverberation 1 11 reducing the noise coming from other directions than the speech An example of such beamformer is the sum and delay beamformer Fig 1 Generalized Sidelobe Canceller 5 This beamformer must be steered in the direction of speech So a steering angle is obtained Figure 2 visualizes this angle Because of this steering angle the microphone signals are delayed against each other The delay can be calculated using the following manner 9 d COS v 1 Here d and v are respectively the distance between two adjacent microphones and the speed of sound 343 7 S To get the microphone signals in phase the sum and delay beamformer must add a delay Expression 1 1s used to decide this delay Afterwards these signals are added together Finally the result of the summation is divided through the total numbers of microphones 10 1 maa ZA 2 Some limitations of the sum and delay beamformer are 2 3 41 e Lim
24. OF MONITORING TOOLS APRIL 2010 PAPER BY PHILIP VAN DEN EYNDE 43 THE IMPLEMENTATION OF WIRELESS VOICE THROUGH PICOCELLS OR WIRELESS ACCESS POINTS PAPER BY JO VAN LOOGR CE A seateunsns sindenkesseangngsseseananinons eananinseseduasian ekain E E 49 USAGE SENSITIVITY OF THE SAAS APPLICATION OF IOS INTERNATIONAL PAPER BY LUC VAN ROEY sssssssssssssses 55 FIXED SIZE LEAST SQUARES SUPPORT VECTOR MACHINES STUDY AND VALIDATION OF A C IMPLEMENTATION PAPER BY STEFAN VANDEPUTTE sasstasiusanesie tisinaisee segtinnsinnesetuianyngisine sunginnescuieiaess velinintenpesegnnnyneequnushas sebisnadeesegunnanneseuenenes 60 IMPROVING AUDIO QUALITY FOR HEARING AIDS PAPER BY PETER VERLINDEN ccssscssccceccsececcccccsscesseeees 66 PERFORMANCE AND CAPACITY TESTING ON A WINDOWS SERVER 2003 TERMINAL SERVER PAPER BY ROBBY WVIELOCKA E E setainiivnn ses undeueyseounininnins A E 72 SILVERLIGHT 3 0 APPLICATION WITH A MODEL VIEW CONTROLLER DESIGNPATTERN AND MULTI TOUCH CAPABILITIES PAPER BY GEERT WOUTERS csnssccsssssnosssannsastossavacunensoantansessusanecossannnaensesnnsusenstanniapiessuacunoseannenenes 78 COMPARATIVE STUDY OF PROGRAMMING LANGUAGES AND COMMUNICATION METHODS FOR HARDWARE TESTING OF CISCO AND JUNIPER SWITCHES PAPER BY ROBIN WUYTS osn ooo onsenseoesensensereeensenn 83 Introduction We are proud to present you this first edition 2009 10 of the Proceedings of M Sc thesis papers from our Master students in E
25. R Oglesby Terminal Services for Microsoft Windows Server 2003 Advanced Technical Design Guide st ed Washington DC USA BrianMadden com Publishing 2004 2 E Sheesley SolutionBase Working with Microsoft Windows Server 2003 s Performance Monitor TechRepublic com 2004 3 A Silberschatz P B Galvin G Gagne Operating System Con cepts 8th ed Asia John Wiley amp Sons Pte Ltd 2008 4 R Morimoto A Abbate E Kovach and E Roberts Microsoft Windows Server 2003 Insider Solutions 1st ed USA Sams Pub lishing 2004 5 D Bird Keep Tabs on Your Network Traffic Available at http www enterprisenetworkingplanet com netsysm article php 10954_ 3328281_1 February 2010 6 Terminal Server Capacity Planning Available at http technet microsoft com en us librarycc75 1284 aspx February 2010 7 What is that Page File for anyway Available at http blogs technet com askperf archive 2007 12 1 4 what is the page file for anyway aspx February 2010 8 Autolt Documentation Available at http www autoitscript com autoit3 docs February 2010 Silverlight 3 0 application with a Model View Controller destenpattern and multi touch capabilities Geert Wouters geert wouters edu khk be IBW K H Kempen Associatie KULeuven Kleinhoefstraat 4 B 2440 Geel Belgium Abstract The technology and the availability of multi touch devices is rapidly growing Not only the industry is making th
26. and finally the bits are remodulated back into tones at the far gateway The fax machines are unaware that a demodulation modulation fax relay is occurring Mostly the packetizing and encapsulating of data is done by the ITU T T 38 standard and is available for H 323 MGCP and SIP gateway control protocol 2 Fax pass through The modulated fax information from the PSTN is passed in band with an end to end connection over a voice speech path in an IP network There are two pass through techniques a The configured codec is used for voice and fax transmission This is only possible using the G 711 codec with no VAD en Echo cancellation EC or when a clear channel codec is used like G 726 32 In this case the gateways make no difference between voice and fax calls Two fax machines communicate with each other completely in band over a voice call b Codec up speed or fax pass through with up speed method This means that the codec configured for voice is dynamically changed to the G 711 codec by the gateway The gateways are to some extent aware that a fax call is made by recognizing a fax tone automatically changing through the use of Named Signaling Event NSE messaging the voice codec to G 711 and turn off EC and VAD for the duration of the call Fax pass through is supported by H 323 MGCP and SIP gateway control protocol 3 Fax store and forward This method breaks up the fax process in sending and receiving processes For
27. are several ways to connect these components The most suitable wiring can be found in figure This design provides a universal solution to test a standalone Cisco or Juniper switch and a Cisco chassis with supervisor installed It is possible to eliminate the external FTP server by using flash memory of the switch as a directory for an FTP transfer Note this implies some disadvantages Enough space on the flash is required and this solution is not that universal for Cisco and Juniper switches As you see critical connections are attached directly to the MasterSwitch Critical connections are connections from which you have to be 100 sure they are operational In this case it s the link between MasterSwitch PC and MasterSwitch FTP The other connections are for testing purpose This increases the reliability of the test On the other hand programming becomes more complex The programmer has to deal with vlan s to redirect icmp and tcp packets to the SlaveSwitch C Test operations The purpose of the script can be summarized into one sentence Testing each interface on errors to make sure you can install the switch in an operational environment It is possible to test the interfaces at different levels It would be possible to check if the bit error rate for a given operational time does not exceed the treshhold To accomplish this it is necessary to send a huge amount of data If you send 1 kB it is not sufficient to observ
28. bits are assigned to the lower subband This is because G 722 is focused on speech Since there are almost no bits available for higher frequencies this codec will not perform well for high frequency signals The Philips Subband coder doesn t have this problem because bits are assigned using the SFI So every subband can get enough bits even the subbands which contain the higher frequencies A second major difference is that G 722 uses ADPCM encoders while the Philips subband coder uses an APCM encoder Here the G 722 codec has an advantage because it uses prediction However this makes the codec slightly more complex But in our application this isn t a problem because the G 722 codec is implemented in hardware If we combine these facts than in theory the Philips Subband Coder should perform better than the G 722 codec for music signals VI PHILIPS SUBBAND CODER IMPLEMENTATION To test the Philips Subband Coder one development board with two DSPs is used One DSP is a CoolFlux DSP NxH1210 the other chip is an NxH2180 The NxH2180 can also be used to connect two development boards wirelessly via magnetic induction In this setup each development board represents a hearing aid Since only the quality of the Philips Subband Coder needs to be examined only one development board is used Figure 6 shows the block diagram of the test setup SBC Codec 12S NxH1210 Enc Line in
29. choice for a virtual machine is the best option because buying a new server would cost the company additional money to simply run the ActiTime application There are a lot of virtualization solutions 13 A few programs that accomplish the task to create and manage virtual machines are vmware xen virtual box Hyper V etc Xen and virtual box are both Open source programs and vmware is a program you have to pay for There is not much of a difference between the different virtualizations programs Since there is little difference between the applications we opt for Microsoft hyper v We choose the hyper v solution because of its ease of use and because hyper v is already included in the Microsoft Windows server product Just add the role of the hyper v application and a virtual machine program is up and running V CONCLUSION In this paper we discussed a way to migrate an application Because a lot of applications needed to be moved to a new server as explained in the short situation scheme the steps described in this paper are not the same for every application that has to be migrated This paper treats a few problems that could possibly come up during the migration process It is not likely that for other applications the same problems come up With a short explanation what time tracking contains the migration of a software tool for time tracking 1s treated in this paper The process of migrating an application and its user data is in mos
30. complexity and low delay 6 7 For the analysis filter the modulation function is given by 2 C n cos n 5 k 1 k E 0 7 n E 0 L 1 In this function M is the number of subbands and L represents the filter length The synthesis filter has a similar function C n cos E n x 5 E 0 7 n E f 0 L 1 B APCM adaptive pulse code modulation After the audio signal is split in several subbands the samples are encoded using APCM The first step in this encoding process is calculating scale factors To this end the subbands are divided in block of length 4 8 12 or 16 For example 128 input samples are transformed in 8 16 subband samples which are then processed as a block The first step is to determine the maximum value for each subband in the block The maximum values are quantized on a logarithmic scale with 16 levels Thus the scale factor needs 4 bits to be coded as a scale factor index The scale factor index can be found by Algorithmic delay without bit reservoir SFI log max After the scale factors are calculated all the samples of that block are divided by the scale factor Such that all samples are in the interval 1 1 Number of available bits a lin Normalizing Subband Samples Samples Index FIGURE I APCM ENCODER Then adaptive bit allocation is used to distribute the available bits over the different subbands The number of bit
31. how it is possible to improve audio quality for a hearing aid This hearing aid was using a speech codec G 722 To improve quality the Philips Subband Coder is proposed After looking at the structure of both codecs it can be concluded that the Philips Subband Coder performs better for music signals than G 722 But at the moment there is a limitation to a bitrate of 166 kpbs For this reason artifacts are heared when using the Philips Subband Coder although when compared with G 722 the sound itself is better With G 722 the higher frequencies don t really come through the Philips Subband Coder solves this problem When new hardware is available which allows higher bitrates the Philips Subband Coder is a possible choice for this application Most important reasons for this are its low complexity thus low memory and MIPS requirements Also this codec has a low delay making it ideal for hearing aids ACKNOWLEDGEMENTS I thank Steven Daenen for giving me the chance for doing this research at NXP also I would like to thank Koen Derom for his help at NXP Further I want to thank Paul Leroux guiding me through this project REFERENCES 1 Hawley M L Litovsky R Y and Culling J F The benefit of binaural hearing in a cocktail party Effect of location and type of interferer J Acoust Soc Am 115 2004 pp 833 843 2 F de Bont M Groenewegen and W Oomen A High Quality Audio Coding System at 128kb s in
32. identify the target device followed by the source 1 MAC address to identify the sending device At the end of the MAC header there is a 4 byte Etherlype field This identifies the used protocol for IPv4 it s value is 0x0800 Since we re creating a new protocol it is suitable to adjust the EtherType field We have chosen the 2 byte value OxFFFF to identify the management packets In this way a possible mix up with other protocols is avoided and the management packets complies with IEEE standards 3 1 2 Payload Payload is the content of the packet and contains follow ing fields e target MAC 6 bytes e target name 32 bytes e source IP 4 bytes e action 1 byte The target MAC is also found inside the MAC header but are not always identical When using broadcast messages all devices within that subnet will receive the broadcast packet In this case it should only be processed by the simulation device it was destined to The target name is a unigue name for every simulation device and is well suited for identifying the device Since Layer 2 packets are used the IP protocol is omitted and no IP addresses are used The IP source field is included for logging purposes The action field defines what command the operating system must execute this gives the possibility to further expand the use of these management packets 3 2 Processing All incoming packets are examined by the network interface All broadcast and
33. in other codecs There are a lot of codecs available these days Since high quality for music needs to be achieved the delay of speech codecs isn t discussed since they perform poorly for music signals In table 1 several codecs are listed with their delays 3 Notice that the lowest delay is still 20ms at a sampling rate of 48 kHz for a sampling rate of 32 kHz this becomes even higher For use in hearing aids this is unacceptable The reason for this high delay is that these codecs use a psycho acoustic model which introduces higher complexity and therefore more delay This higher complexity means that these codecs use bigger block sizes for the encoding process which introduces more delay These codecs also use an MDCT filter bank This type of filter bank also has a longer delay than a OMF filter bank TABLE I OVERALL DELAYS OF VARIOUS AUDIO CODECS SAMPLING RATE 48 KHZ MPEG 1 Layer 2 192 kpbs MPEG 1 Layer 3 128 kpbs 34 ms 54 ms MPEG 4 AAC 96 kpbs MPEG 4 HE AAC 56 kpbs MPEG 4 AAC LD 128 kpbs 55 ms 129 ms UI PHILIPS SUBBAND CODER A Subband splitting In the first step the audio signal has to be split in several subbands The Philips Subband Coder uses 4 or 8 subbands To split the signal into subbands an analysis filter is used at the decoder side a synthesis filter is used to recombine the subbands Cosine modulated filter banks are used Both are polyphase filter banks These type of filters have low
34. include robustness speed and universality This will be discussed in topic In the first stage the most appropriate language has to be chosen After defining the programming language the real programming work can be done While thinking about some useful methods it became immediately clear that there isn t just one suitable solution Getting and setting data from and to the switch can be realised with different communication methods In this paper a comparison between serial communication telnet and SNMP can be found Afterwards another benchmark is set up to decide whether it is useful to re mplement the script in another more efficient language II PROGRAMMING LANGUAGE Determining the most suitable programming language is the first step taken to realise the script In the early days you were restricted to choose between fortran cobol or lisp At the moment the amount of programming languages exceeds the number of thousand The need to select some languages to compare is inevitable This selection can be found below and will be discussed very shortly e Java e C e Perl o Python e Ruby e PHP PHP PHP is a server side scripting language In some applications it is used to monitor network traffic and display the results in a webbrowser PHP needs a local or external server to run PHP scripts Java Nortel Device Manager is a GUI tool to configure Nortel switches which is fully written in Java That
35. independent and have a common distribution The second characteristic B specifies the nature of the service time They are also assumed to be independent and identically distributed and are independent of the interarrival times The third characteristic s is the number of parallel servers The fourth characteristic q describes the queue discipline Depending on whether the customers are being treated FCFS LCFS etc The fifth characteristic specifies the maximum allowable number of customers in the system including customers who are waiting and customers who are in service i e system capacity The last characteristic gives the size of the population from which customers are drawn Unless the number of potential customers is of the same order of magnitude as the number of servers the population size is considered to be infinite After analyzing the jobs who where submitted by the users the past few months we could conclude that interarrival times and services times were independent identically distributed random variables having an exponential distribution Also the queue discipline used by SCKeCEN is non preemptive Prioritizing thus the system we are working with can be described as M M c Pnp co Before the upgrade from the cluster c would be 192 48 nodes X 4 logical processors Systems like these are called Markovian systems which stems from the fact that the exponential distribution is the only continuous distribution with the markov pro
36. interference of home appliances At the Warande we opted to use all three standards This way we are sure that there will always be enough open connections for clients This is of no inconvenience to the client since the present technology of wireless network adapters will search for a connection regardless of the standard being used when supported The result is shown in the image below The yellow areas in the image represent areas where there is no need for coverage or areas where we do not care if there is coverage or not Using this method I was able to conclude that there are 16 WAP s needed in the first building to provide the areas with enough coverage for wireless internet connection and 3 extra WAP s to ensure the needed coverage for voice traffic The second building needed 13 WAP s to get enough coverage for the wireless internet connections and an additional 14 WAP s for the necessary coverage for voice traffic HI THE CONFIGURATION Since the need for security in the sector is very high I will explain this section by means of a few examples because I can not share the actual configuration method and commands with the public The configuration needed must allow a person to call internally to other IP phones or to analog phones externally Also we must foresee usage of faxes This means that a configuration of analog ports for the faxes and digital ports for the actual calls is necessary Next to these two diff
37. is executed during the script For example if an SNMP request is done a counter ISNMP is added by 1 The next step taken is to eliminate some negligible operations such as split functions They were only executed 5 times The remaining results can be found in figure 12 1 Then all these operations needs to be programmed in Java C Perl Python Ruby and PHP Each operation is executed as many times as seen in Quantity of executions To accomplish operations like SNMP requests sometimes external modules packages are used A list of all used packages can be found in table 12 3 Note that implementation inefficiency is dealt with Here is the explanation using an example During a telnet connection it is necessary to wait for the Type Amount Percent Quantity of executions 500000 measurements Regex 332872 Variable changes 144477 Function calls 2848 10107 SNMP 1687 0 0119732 5987 If functions 3900 Push array 1397 Ping 97 FTP transfer 85 Telnet operations 0 000177433 89 TABLE IV USED OPERATIONS E Regex E Variable changes E Function calls U m SNMP E if functions S mm Push array Ping m FTP transfer Telnetoperations Fig 10 Used operations System specifications PC HP Compaq NC 6120 1 86GHz 2GB RAM Platform Windows XP 32 bit interpreter compiler Perl ActivePerl 5 10 1 1007 Python Python 2 6 4 Ruby Ruby 1 9 1 p376 Java JDK 6u19 and NetBeans 6 8 C Visual C 2008 Express E
38. is very useful to test the functionality of the switch Preferably this is done by a fully automatic program which needs minimal user interaction In this paper the design and testoperations are discussed shortly The implementation of a script or program can be done in several ways and in different languages In this work a basic implementation has been made using Peral script showing the required functionality Afterwards a custom benchmark shows if it is useful to implement the same functionality using other more efficient languages Several communication methodes like serial communication telnet and SNMP are examined This paper will prove which communication method is the most effective in a specific situation focussing on getting and setting switch parameters I INTRODUCTION EFORE configuring and installing new switches at companies it is recommended to make sure every single ethernet or gigabitport is working properly Companies are free to sign a staging contract which covers this additional quality test At Telindus the staging process is executed manually Not only may this be an extremely lengthy and uninspiring job but more important automating these processes also allows to deliver a higher quality service at a lower cost Concerning these important issues we wrote a fully automatic script to test Cisco and Juniper switches To solve the first issue the script must ensure minimal user interaction Other requirements
39. it is easy to understand that lines 2 and 3 will have a better fit than lines 1 and 4 If we then compare these calculated maxima either line 2 or 3 will have the maximum fit suppose line 3 Next we will calculate the line for which the maximum correlation is above half the correlation of line 3 This is done bottom up The first line that meets this condition will be line 2 The range that corresponds to line 2 will be taken as the starting range of the target After calculating the starting range of the target we will use a polynomial fit to calculate the target s azimuth location power and an error number mean squared error between the target s slow time echo and the parabola that fits best as is shown in Figure 4 The x value and y value of the calculated best fitting parabola s vertex will be used to represent respectively the target s azimuth location and the amplitude in Volts of the target s reflected signal that is received by the antenna B RCS SKOLNIK 5 provides the following definition The radar cross section of a target is the fictional area intercepting that amount of power which when scattered equally in all directions produces an echo at the radar equal to that from the target RCS is used to assess radar sensitivity It is used to measure the ability to detect a target at any given range Targets with a low RCS like a Cessna might not be spotted at long range while the new A380 which has a higher RC
40. network load during the 30 minutes a N scan procedure it s clear that MonitorMagic has the lowest Di S k use of bandwidth IO maps Bandwidth 14 000 Mb 12 000 10 000 100 00 90 00 8 000 80 00 6 000 4 000 NIE ni 60 00 0 000 30 00 T 6 o 2 V L2 EL 6 B 20 00 a z GH ELT i ge Lik PEPER EGE 0 00 an el PD a O mo oO Cc ep c VY O w VY o Hn s o Hn HW He wp DR pP Vv 2 Q v o AZ m O o Le 0O O amp Y o gt DV v 2 GF Ff dd Z 2 u O EMME DOE c EF a a 2 2 6 nT z x oo S 8 _ Z oO u c O Z lt oO u FS 2 os os lt f E a mReads sec wm Writes sec Transfers sec J he Fig 3 Disk results E Total Mb E Mb tool gt server E Mb server gt tool With the details listed in the following table Reads Writes Transfers Fig 2 Bandwidth results monitor sec sec sec Ipsentry 0 000 1 146 1 146 WebWatchBot 0 013 1 816 1 829 With the details listed in the following table l ook TO Tembria server monitor Total Mb server tool 0 549 1 522 2 071 a ete ar NK 3 233 0 736 2 497 ManageEngine 0 826 1 752 2 578 ManageEngine 3 324 0 550 2 774 Active perts mS IO U SolarWinds SE 5839 12671 Web WatchBot 12 205 0 591 11 614 Ipsentry 12 776 6 992 5 784 l ServerAssist 94 827 18 805 76 021 salu Table 4 Bandwidth results detail The price is a parameter that may not be under
41. one or more requirements each of which requests a number of resources of a given type e Nodes A node is a collection of resources with a particular set of associated attributes e Policies Policies are generally specified via a config file and serve to control how and when jobs start e Quality of Service The term quality of service QoS refers to resource reservation control mechanisms Quality of service is the ability to provide different priority to different applications users or data flows or to guarantee a certain level of performance The Maui Scheduler allows administrators fine grain control over QoS levels on a per user group and account basis e Reservations A reservation can be made to ensure that a job with high priority will run as soon as possible The manner in which all these objects are managed can be described in several core algorithms 7 First of all there is a priority algorithm which will calculate the overall priority of every job that has been submitted The basis of the priority algorithm is a First In First Out FIFO algorithm but this is enhanced through the combination of weighted factors Priority serviceWeight serviceFactor resourceWeight resourceFactor fairshareWeight fairshareFactor directspecWeight directspecFactor target Weight targetFactor a bypass Weight bypassFactor Priority Evaluation Metrics Component Service Current queue time and ex Allows fav
42. programs or load all documents at the same time When studying two real users working at the Terminal Server during their job memory usage for both employees ranges from 90 MB to 160 MB This means that the real users use less memory than the simulation script Therefore the Terminal server can support more users than the calculated number of 7 2 Processor The output from the Processor Time counter indicates that there isn t a sustained value of 100 utilization which should mean that the processors aren t too busy However when we look at figure 7 which shows the output from the Processor Queue Length counter we can see that there is a sustained value of around 10 with peaks up to 20 The Queue Length counter indicates the number of requests which are backup up as they wait for the processors If the processors are too busy the queue will start to fill up quickly which indicates that the processors aren t fast enough The queue shouldn t have a sustained value of 2 which is the threshold Figure 7 show that the counter has a sustained value significantly greater than 2 so the processors of the Terminal Server aren t fast enough This will probably result in a decrease of performance when more users are using the server This can be resolved by upgrading the processors 3 Network Network usage can be a limiting factor when it comes to Terminal Server environments It 1s the interface between the Terminal Server and the ne
43. s the reason why this language became a promising solution Many network applications require multithreading where Java is the ultimate language to handle these multithreaded operations However as we will see later multithreading was not of any interest in our particular situation C Normally applications written in C are very fast It s interesting to check if this statement is true regarding network applications Perl Ruby Python Unlike Java and C these alternatives are scripting languages Object ori nted programming is possible especially with Ruby but it is not it s main purpose The syntax of these three languages differ Ruby and python don t use braces but take care of clarity with tabs Perl on the other hand uses braces like the most languages do Some sites ensure that python is the fastest http data perl it shootout On the other hand Perl is the fastest along other websites http xodian net serendipity index php archives 27 Benchmark PHP vs Python vs Perl vs Ruby html The reason for these different results can be easily explained Based on one specific benchmark it would be unfair to conclude that Perl 1s the fastest in every respect It is only possible to compare these languages with a specific purpose in mind Our purpose 1s to write a script which automatically tests hardware of a Cisco or Juniper Switch In this case it would be useless to benchmark the graphic processing skil
44. scores for the different configurations of the different modes These values are the average score of 11 different audio signals TABLE III SBC CONFIGURATION PARAMETERS confi cont2 eonf3 Org anchorl anchor2 B Phase 2 Table 4 gives the results from the listening test with the different modes these results are also the average of 1 Laudio signals TABLE IV SBC CONFIGURATION PARAMETERS IX DISCUSSION OF RESULTS After examining the results of phase the conclusion can be made that the configurations with 8 subbands and block length 16 always give the best results at a limited bitrate of 166 kpbs In phase 2 the results show that the joint stereo mode is best But the audio quality isn t very high Artifacts can be heard which is due to the limited bandwidth of 166 kpbs The artifacts aren t audible when the frequency band is limited In modern music though the frequency band is very wide and this causes more artifacts In 4 the G 722 codec is evaluated From these results it was concluded that for music signals a number of audible distortions were revealed that do not occur for speech signals Also the perceived bandwidth of the coded music was less than 7 kHz This is something that wasn t noticed during the listening tests of the Philips Subband Coder The evaluation of G 722 also showed that more noise presented itself in the higher subband X CONCLUSION The main question in this paper was if and
45. system with one small queue 1 e few nodes per slot One consideration was made about the usage of serial jobs on the new nodes Because the AMD Opteron nodes haven t got hyper threading enabled they are more suited for running serial jobs This results from the fact that internode communication is much faster than intranode communication Thus when mapping the queue to the nodes it seems best to reserve the AMD Opteron nodes for serial processing This means that those departments that often submit serial jobs will get more AMD Opteron nodes and thus less Intel Core I7 nodes To remain fair to all departments we ran a benchmark using a MCMP program which will be used by most users to create their model After finding the difference in speed between the AMD Opteron processor and a Intel Core 17 node we rearranged the amount of nodes each queue receives thus making sure every institute receives processing power according to their financial input IV SCHEDULING ALGORITHM For scheduling purposes Maui is being used At the moment the scheduling decisions are made by taking into account 3 parameters e Group configuration by enabling the group priority the system can implement the company policy Resources should be allocated according to the share in the investment e User configuration Although this parameter is set every user within one department has the same target and therefore everyone has equal priority When a use
46. the real user behavior This behavior is imitated by building an interaction script with the user requests A load generator like Jmeter then passes through the interaction script adapted with test parameters based on a real life environment on the SaaS application IOS Mapper The load generator imitates the behavior of the web browser it sends continuous requests to the SaaS application waits a certain time after the SaaS application has answered to the request this 1s the thinking time which real users also have and then sends a new reguest The load generator can simulate thousands of concurrent users at the same time to test the SaaS application We ll use Jmeter as the load generator It s a completely free Java desktop application With Jmeter we ll record the behavior of the users of the SaaS application IOS Mapper Afterwards we ll make a load model from the recordings We can introduce this load model into Jmeter and subsequently we are able to simulate our load model Each simulated web browser is a virtual user A load test will only be valid if the behavior of the virtual users resembles the behavior of the effective users For this reason the behavior of the virtual users must e follow patterns resembling real users e use realistic thinking times e have an asynchronous behavior between each user Figure shows a load model of a virtual user using the SaaS application IOS Mapper based on the patterns of a real l
47. therefore easier to find the center by adding half of the height to the vertical coordinate of the position vector and submitting this new vector to the same rotation matrix Finding the radius is just a matter of applying Pythagoras to the known radius of the bottom circle and half of the height Similar techniques can be used for all the other primitives A ray is determined by its starting and ending point Let P be the starting point and P the ending point The direction Rg is defined as the normalized vector pointing from P to Pe P t is a point along the ray PH P t R 1 The intersection test is explained in the figure Fig 1 Intersection of a ray and a sphere First vector O pointing from P to the sphere center C is constructed OSCE 2 Next we find the length along the ray between P and C by using the dot product of Q and Rg P C Q R 3 Substituting the t in equation 1 with this length we can find C which is the orthogonal projection of the center point C on the ray C P P C R 4 The bounding sphere is intersected if the distance between C and C is less than the radius r C pz and C 9 2 d C C x x y H 2 5 d C C lt r 6 One thing we ve overlooked so far is that a ray is of infinite length while we re interested in a ray segment bounded by the source and the studied point Imagine the studied point lies between two
48. thick concrete walls of the building When we send data through connecting to the WAP s we will use the 2 4 GHz or 5 GHz frequency ranges The 2 4 GHz range is used by 802 11b and 802 11g IEEE standards and is probably the most widely used frequency range In this range we have 11 channels each 22MHz wide This means that we can only use channel 1 6 and 11 because the other channels will overlap with others and cause interference This is One more factor we need to include when we make our actual survey The 5 0 GHz frequency range contains the IEEE standard 802 1la Because 802 11a uses this range and not the 2 4 GHz range it is incompatible with 802 11 b or g 802 lla is mostly found in business networks due to the higher cost Each standard has its pros and cons 5 802 11a pros o Fast maximum speed up to 54 Mbps o Regulated frequencies prevent interference from other devices signal 802 11la cons o Highest cost o Shorter signal range that is easily obstructed 802 11b pros o Lowest cost o Good range that is not easily obstructed 802 11b cons o Slowest maximum speed up to 11 Mbps o Possibility of interference of home appliances 802 11g pros o Fast maximum speed llMbps using DSSS and up to 54 Mbps using OFDM o Good signal range that is not easily obstructed o Uses OFDM to gain bigger data rates o Backward compatible with 802 11b 802 11g cons o More expensive then 802 11b o Possibility of
49. transmitted over the WAN RiOS transmits the segments of a file or e mail to the next Steelhead before the client has requested this file or e mail Therefore a user can access this file or e mail faster Transaction prediction TP optimizes the network latency The Steelhead appliances intercept and compare every transaction with a database that contains all previous transactions Next the Steelhead appliances make decisions about the probability of future events If there is a great likelihood of a future transaction occurring the Steelhead appliance performs the transaction rather than waiting for the response from the server to propagate back to the client and then back to the server RiOS has a CIFS optimization feature that improves windows file sharing and maintains the appropriate file locking CIFS or Common Internet File System is a public variation of the Server Message Block SMB protocol E Management Streamlining RiOS was designed to simplify deployment and management of Steelhead appliances There mustn t be made any changes to servers clients or routers The management of a Steelhead appliance can be done through Secure Shell SSH command line or a HTTP S graphical user interface The management of a complete network of Steelhead appliances can be done through the Central Management Console CMC The CMC is an appliance that provides centralized enterprise management configuration and reporting Il IPANEMA TECHN
50. unicast packages that match are accepted and passed on At kernel level all incoming packets are processed At an early stage the EtherType of every MAC header is examined to match OXxFFFF If no match is detected e g other protocol it is left untouched If the packet matches a subroutine is exe cuted and the entire package MAC header payload is passed using pointers This function further validates the incoming packet and executes the desired command based on the payload s action field 3 3 Summary A layer 2 packet layout is designed and can be used execute tasks remotely One of these task is to initiate a Soft Off command using the information found with the ACPI framework Combing both the ACPI frame work and layer 2 management packets it is possible to remotely power off a router simulation device We can hereby conclude that remote power off is possible and can be successfully implemented in an operating system with no power management extensions 4 REMOTE CONTROL POWER ON The last step is to power on the simulation device When powering off the entire device is placed into the ACPI G2 S5 Soft Off state Meaning that all devices are shut down completely This is a problem since an inactive network device cannot receive network packets or even process them 4 1 Remote Wake Up Remote wake up is a technology to wake up a sleeping device using a special coded Magic Packet Most network devices s
51. walls while the source lies outside of these walls The ray will intersect both walls but the path between the source and the studied point intersects just one wall In the above test an intersection is found even if the ray ends before reaching the bounding sphere To counter this we ll use an extra test if equation 6 is satisfied Fig 2 Halved chord length r vr L 7 d P P lt d P C r 8 If equation 8 is satisfied we can ignore the intersection we found earlier Note that is the distance calculated in 5 The effectiveness of the bounding sphere depends on how close the sphere fits the original object While this certainly is not perfect for long thin objects the proposed method provides a considerable increase in performance while inducing reasonable precalculations and programming complexity HI NARROW PHASE The broad phase calculations before allow us to eliminate most of the none intersected volumes from the calculations The remaining volumes are used in ray triangle intersections tests Each volume s triangle list is iterated and each triangle on the list submitted to a test The test is divided into three stages In a first stage the intersection point of the ray with the plane of the triangle is calculated This requires determining the plane equation which is a time consuming calculation Then we check if the intersection is located within or on the borders of the triangle Finally we ll use ano
52. we took when implementing this in C It has been shown that real time data compression can be a very useful tool not only for disk and memory usage but also to reduce the time spend on reading data for offline analysis afterwards REFERENCES 1 A Kruger and W F Krajewski Efficient Storage Of Weather Radar Data Iowa University Iowa 1995 2 Intersoft Electronics 2009 Radar Interface Module RIM782 Available at http www intersoft electronics com 3 C Wolff 2009 Azimuth Change Pulses 16 February 2010 at http www radartutorial eu 17 bauteile bt04 en html 4 Intersoft Electronics 2009 Radar Target Generator RTG698 Available at http www intersoft electronics com 5 M I Skolnik Introduction to radar systems 2002 Vol 3 pp 49 64 6 E F Knott Radar Cross Section Measurements 2004 pp 14 18 7 J C Toomay and P J Hannen Radar Principles for the Non Specialist Vol 3 2004 pp 79 82 8 I Falcounbridge Radar fundamentals 2002 ch 14 9 M I Skolnik Introduction to radar systems 2002 Vol 3 ch 7 10 Maxim Integrated Products 2000 Application Note 641 ADC and DAC glossary Available at http www maxim ic com 11 K Hughes and T Hughes Parallel and distributed programming using C 2004 ch 4 Migration of a time tracking software application ActiTime Maarten Devos Ward Vleegen Tom Croonenborghs BW K H Kempen Associatie KULeuven B 2440 Geel Belgiu
53. will not be a problem Memory comitted bytes 100 000 90 000 80 000 70 000 60 000 50 000 40 000 30 000 20 000 10 000 0 000 MonitorMagic lpsentry Tembria server monitor ActiveXperts WebWatchBot ServerAssist PA Server Monitor Spiceworks ManageEngine SolarWinds m Memory Fig 6 Memory results With the details listed in the following table monitor Memory ServerAssist 9 526 PA Server Monitor 10 170 Spiceworks 15 171 ManageEngine 19 970 SolarWinds Table 8 Memory results detail 75 218 IV CONCLUSION After excessive testing in a standardized environment we have come up with the best tool that competes with the requirements Conclusions can be taken in several departments e Network load e Disk e Price e CPU e Memory The summarization consists of mean values of all measured results classified by importance in decreasing order and listed from best to worst All of this gives us the best suitable tool for the company As you can see in the benchmark section there is a great difference concerning network load DISK CPU memory and the price that comes with the tool The most important factors were discussed earlier that brings us to the overall comparison of the tools and_ their performance The following graph arranged according to best performance to worst will give us the best suitable tool for the company A small remark concerning the graph the price will not be
54. 000 20000 40000 10000 T Me 30000 0 T 20000 longexec longexec shortexec shortexec 10000 long short long short on command command command command Serial Telnet SNMP Fig 5 SET wait 70000 70000 60000 50000 60000 50000 40000 30000 40000 20000 30000 zoni 20000 0 I I I 10000 longexec longexec shortexec shortexec long short long short 0 1 command command command command Serial Telnet SNMP Fig 6 SET no wait IV SCRIPT As previously mentioned a script would be very useful to test a Cisco or Juniper switch automatically Some conditions must be met The script must be fast robust universal and needs minimal user interaction This section describes the operation of the script A Purpose Before a switch will be installed at a company this script will prove that every interface is able to send and receive data If no errors are detected the switch has passed the test which can be verified in a HTML report showing every detected error The possibility to add some configuration automatically is an extra useful feature The switch can be tested and configured at the same time B Design The script will need an FTP server a PC from which the script has run a MasterSwitch and a SlaveSwitch The ee e 2950 24 MasterS witch 2950 24 SlaveSwitch Server PT FTP Fig 7 Design SlaveSwitch is the switch being tested There
55. 088 02 18 831 02 15 408 02 33 437 140828 ms o 5070ms Python 02 10 171 02 10 296 02 09 467 02 04 624 02 11 671 02 07 874 02 22 264 02 10 780 02 14 608 02 08 186 130994 ms o 4497ms PHP 03 00 454 02 58 308 02 57 960 03 03 256 03 01 936 02 59 375 03 02 162 03 10 087 03 03 672 02 58 644 181585 ms o 13209ms Java 01 32 326 01 31 530 01 38 607 01 30 510 01 33 558 01 34 546 01 32 474 01 33 643 01 41 844 01 36 428 94547 ms C 01 43 425 01 41 096 01 39 315 01 38 329 01 39 565 01 41 426 01 38 567 01 37 238 01 38 642 01 37 939 99554 ms o 1794ms TABLE VI RESULTS CFR PDF 200000 180000 160000 140000 120000 100000 80000 60000 40000 20000 Java C Perl Python Ruby PHP Fig 11 benchmark results As you can see in figure 12 2 it can be easily seen that Perl 1s the fastest among all scripting languages As mentioned before it s a good idea to wonder whether it is useful to rewrite the script in Java or C Let s take a look at the results Perl needs 104182ms to handle the script C and Java are respectively 4 442 and 9 248 faster Because all operations are executed approximately 3 56 times the original value these percentages will be strongly reduced We can conclude that rewriting the script doens t give a remarkable additional value VI CONCLUSION To test a switch manually it
56. 1_5_n23 i Sn eam mg met Tj EEE DEN setup exe mg pe gewijzigde Monitoring 24548 57880 1211056 17674 37258 249730 inhoud and File size optimization Transmission 78 6 KB s 8 speed time without ene minutes 42 7 seconds IP engines seconds Monitoring 33188 78040 1582576 23434 optimization and redundancy elimination Transmission speed time 65 5 KB s 61 ei j u j IP engines in seconds ocni En g 7 failsafe mode Ke Table 3 Pricing Riverbed and Ipanema for a three year contract in EUR Transmission 190 0 KB s 3 speed time 120 1 KB s IP engines first time 33 seconds dar 36 D Features file transfer a Transmission speed time 3 3 MB s 1 6 1 MB s 6 3 2 MB s 1 3 3 MB s 12 Riverbed Technolog Ipanema Technologies Table 2 Ipanema Technologies results FTP server s streamlining speed TCP Riverbed Technology uses Steelhead appliances that are Connection pooling 7 O O oOo ooo y placed on both sides of the WAN There is also a possibility to install the Steelhead Mobile client software on laptops of the mobile users When the Steelhead Mobile client software is n transformation CIFS acceleration IP engines second seconds seconds second seconds Data deduplication gt disk Multi level redundancy elimination gt time file transfer Data streamlining based architecture RAM and disk based architecture be done through the management console of the appliance or through the Central Management Co
57. 30 300000 200000 20 number of generated reports 100000 40 0 4 r A 0 0 200 400 600 Fig 7 Several simultaneous users until the server crashes In figure 8 we tested if there was a difference in response time between the load test on 1 pc and on 2 pc s with each its own Internet Several simultaneous users 600000 100 90 response time ms 500000 x 80 70 response time ms 2 60 locations 300000 50 400000 number of generated 40 reports 200000 a number of generated 20 reports on 2 locations L 10 o N 0 100000 Fig 8 Several simultaneous users divided over 2 locations It s clear that the bandwidth has no influence on the end to end response times of the SaaS application IOS Mapper C Real life approach In reality several users will never carry out the same request simultaneously and the consecution of requests will never immediately follow each other without time delay Each user will use the SaaS application at a different moment And each user has a thinking time for completing an action These thinking times will be different for each action and will differ for every user These things have to be taken into account to create a real life multi users profile If we only take the abovementioned values into account then the SaaS infrastructure will be too powerful for the number of users that can use the SaaS application This
58. 76 00 50 579 00 51 125 00 50 821 50819ms o 566ms s s 00 41 922 00 41 579 00 41 344 00 42 407 00 41 016 00 40 453 00 41 903 00 41 343 00 42 104 00 41 187 41526ms o 548ms Telnet SET no wait Ll 00 12 53 00 13 109 00 12 468 00 12 719 00 12 625 00 14 281 00 12 563 00 12 594 00 12 610 00 12 547 Is 00 10 703 00 10 890 00 10 687 00 10 484 00 10 515 00 10 890 00 10 781 00 10 484 00 10 734 00 10 672 10684ms o 143ms s l 00 12 17 00 12 109 00 12 672 00 12 640 00 12 484 00 13 172 00 12 594 00 12 422 00 12 891 00 12 703 12586ms o 299ms s s 00 09 328 00 09 156 00 09 157 00 09 828 00 09 016 00 09 063 00 09 266 00 09 250 00 09 047 00 09 094 9221ms o 225ms SNMP SET no wait Ll 00 42 906 00 42 031 00 41 875 00 41 968 00 41 984 00 41 809 00 41 582 00 42 082 00 42 734 00 41 766 42074ms o 399ms l s 00 43 483 00 42 701 00 42 014 00 44 532 00 44 751 00 43 543 00 43 832 00 42 609 00 42 986 00 43 014 43347ms o 815ms s l 00 43 811 00 43 687 00 44 312 00 43 687 00 43 544 00 41 844 00 43 206 00 41 578 00 44 057 00 41 969 43169ms o 930ms s s 00 42 657 00 41 642 00 42 157 00 41 642 00 41 860 00 41 860 00 42 578 00 41 795 00 41 781 00 41 624 41960ms o 361ms TABLE II RESULTS CFR PDF 12805ms o 520ms
59. Code 0 SSR Refl 0 Res 5 el A Code xD S Address 0 Res6 0 Altitude Ft D IMB Byte1 256 Res7 0 3D Heigth D IMB Byte2 0 Res 8 39 833752 Range 0 IMB Byte3 0 Res 9 0 000000 Range 0 IMB Byte4 0 Res 10 171 000000 Azimuth 0 MB Byte 5 0 000000 E Azimuth D MB Byte 6 0 Plot Quality 0 MB Byte 7 162 65 Power 0 Com Cap x60400 Status 0 ACAS Fig 2 RASS S6 data field The most important target properties in a RASS S6 data field for us are e Scan Number e Range e Altitude Ft e Azimuth e xX Nm e Y Nm These properties are important because they allow us to track a target in the primary radar raw video This makes it easy to calculate target radar parameters which can then be used to analyze radar performance We can display each target represented by a RASS S6 data field in an XY graph where each plot represents one target return during one antenna revolution An example of such an XY graph is shown in Figure 3 Fig 3 XY graph of secondary radar digital data in LabVIEW7 III PARAMETER CALCULATIONS Now that we have understanding of the data representation we can move on to radar parameter calculations We will discuss RCS parabolic fit error number and SNR calculation All of these parameters are calculated using LabVIEW We will give an overview of what these parameters are why they are important and how they are calculated Note that when testing a radar system we generate perfect t
60. Content INTRODUCTION isonsssennimensventmnninsernanrmnssanianinseranivenssendindisetsanwansscsvnninsersanikansentnensssaanmantseoikaninsssvanikanksentnanne 2 OPEN SHOP SCHEDULING OF A LINUX CLUSTER USING MAUI TORQUE PAPER BY MAARTEN DE RIDDER 3 PRIMARY RADAR PERFORMANCE ANALYSIS AND DATA COMPRESSION PAPER BY STIJN DELARBRE ec00 9 MIGRATION OF A TIME TRACKING SOFTWARE APPLICATION ACTITIME PAPER BY MAARTEN DEVOS 14 WAN OPTIMIZATION CONTROLLERS RIVERBED TECHNOLOGY VS IPANEMA TECHNOLOGIES PAPER BY NICK GOV VAR TS oiee seandndeebnnhanasedhnnatsanakaae enke nennkitkkennnndesdbankteseitamike bera aseamnkieias 19 LINE OF SIGHT CALCULATION FOR PRIMITIVE POLYGON MESH VOLUMES USING RAY CASTING FOR RADIATION CALCULATION PAPER BY KAREL HENRARD inicssssanisosssasnsnsnascntesinssscannavassseanasnnesssennanntessaneainesseennseaiesseraaneesasesmananss 24 INTERFACING A SOLAR IRRADIATION SENSOR WITH ETHERNET BASED DATA LOGGER PAPER BY DAVID LOOHMAN Sep E sana pagepecumigaaeseperannns 29 CONSTRUCTION AND VALIDATION OF A SPEECH ACQUISITION AND SIGNAL CONDITIONING SYSTEM PAPER BY JAN MERTENS csscvcccessisanuessdanstinecesoannaanssacaneatewedsauseauessdasstinenssssumensasauaunaneassesu saunas aaneineneseammganssadeunedeaedenusuanasscansamessssantennes 33 POWER MANAGEMENT FOR ROUTER SIMULATION DEVICES PAPER BY JAN SMETS ccsccccsccsececccescssceeseeees 39 ANALYZING AND IMPLEMENTATION
61. MICROSOFT CORPORATION Monitoring performance http www cwlp com samples tour perfmon htm 2001 4 JAY BAELE 2007 Wireshark amp Ethereal network protocol analyzer toolkit Syngress Publishing Inc Rockland 523p The implementation of wireless voice through picocells or Wireless Access Points Jo Van Loock Stef Teuwen2 Tom Croonenborghs 3 Department of biosciences and technology Department KH Kempen University College Geel Abstract Poor coverage in buildings and ensuring a good quality became the biggest problems of voice communication and are the major cause that business customers change their provider To have a maximum coverage and quality for wireless voice communication one can use Picocells or Wireless Access Points WAP s Picocells will enable voice communication through the normal Public Switched Telephone Network PSTN while WAP s will use the advancing Voice over Internet Protocol VoIP technology The choice many network designers have to make is to use picocells or VoIP technology to ensure an optimal coverage and quality in voice traffic This choice is mostly made based on a site survey Nevertheless the advantages and disadvantages of both solutions need to be known and considered Sometimes network designers can consider skipping the site survey and make the choice only based on experience in the field I INTRODUCTION Ever since 1876 people have been using voice communication technolog
62. MKL SGESV Core2Duo E6300 1 86 GHz 100 5 90 80 70 6o 50 40 4 30 20 10 5 tijd s 0 2000 4000 6000 8000 10000 12000 rijen MATLAB GotoBlas2 Ref LAPACK MKL DGESV Xeon E5506 2 13 GHz 100 5 tijd s S 0 2000 4000 6000 8000 10000 12000 rijen MATLAB GotoBlas2 Ref LAPACK MKL tijd s SGESV Xeon E5506 2 13 GHz 100 5 90 80 5 707 MATLAB 60 50 GotoBlas2 40 Ref LAPACK 30 4 MKL 20 5 10 4 A u 0 T T T T 0 2000 4000 6000 8000 10000 12000 rijen Efficiency efficientie DGESV Pentium D 940 3 20 GHz 100 5 90 80 5 70 4 MATLAB 60 50 GotoBlas2 40 Ref LAPACK 30 4 MKL 20 4 10 4 0 T T T T T 1 2000 4000 6000 8000 10000 12000 rijen efficientie SGESV Pentium D 940 3 20 GHz 100 5 90 80 o 70 4 60 50 4 40 30 20 10 4 0 2000 4000 6000 8000 10000 12000 MATLAB GotoBlas2 Ref LAPACK MKL rijen efficientie DGESV Core2Duo E6300 1 86 GHz 100 5 90 80 70 MATLAB 60 4 GotoBlas2 u Ref LAPACK 304 MKL 20 5 10 5 0 T T T T T 1 0 2000 4000 6000 8000 10000 12000
63. OLOGIES A Autonomic Networking System Ipanema s autonomic networking system or Ipanema system is an integrated application management system that consists of three feature sets e Intelligent Visibility e Intelligent Optimization e Intelligent Acceleration It is designed to manage up to very large enterprise WANs Belgacom offers application performance management APM services to their customers through the Explore platform So the Ipanema system is a managed service B Intelligent Visibility Intelligent visibility enables full control over the network and application behavior It uses IPlengines to gather real time network information The IPlengines sent this information to the central software IPlboss A synchronized global table stores volume and quality information of all connections active Application Figure 2 Synchronized global table The Ipanema system measures application flow quality metrics such as TCP RTT Round Trip Time TCP SRT Server Response Time and TCP Retransmits It also uses one way metrics to measure the performance of a protocol such as UDP User Datagram Protocol which is used by VolP Voice over IP and video Ipanema provides two application quality indicators MOS Mean Opinion Score and AQS Application Quality Score C Intelligent Optimization Intelligent optimization guarantees the performance of critical applications under all circumstances The Ipanema system uses o
64. Proceedings of the 98th AES Convention Paris France Feb 1995 3 M Lutzky G Schuller M Gayer U Kr mer and S Wabnik A guideline to audio codec delay in Proceedings of the 116th AES Convention Berlin Germany May 2004 4 S M F Smyth et al An independent evaluation of the performance of the CCITT G 722 wideband coding recommendation IEEE Proc ICCASP 1988 pp 2544 2547 5 Advanced audio distribution profile A2DP specification version 1 2 http www bluetooth org Apr 2007 bluetooth Special Interest Group Audio VideoWG 6 P P Vaidyanathan Quadrature Mirror Filter Banks M Band Extensions and Perfect Reconstruction Techniques IEEE ASSP magazine July 1987 pp 4 20 7 J H Rothweiler Polyphase quadrature filters A new subband coding technique IEEE Proc ICCASP 1983 pp 1280 1283 8 ITU Recommendation G 722 7 kHz audio coding within 64 kbit s November 1988 9 P Mermelstein G 722 a new CCITT coding standard for digital transmission of wideband audio signals IEEE Communication Magazine vol 26 February 1988 pp 8 15 10 ITU R Method for the subjective assessment of intermediate quality levels of coding systems Recommendation BS 1534 1 Jan 2003 Performance and capacity testing on a Windows Server 2003 Terminal Server Robby Wielockx K H Kempen Geel robby wielockx gmail com Abstract Using a Terminal Server inste
65. S will still be spotted Of course at very long ranges none of both planes will be spotted RCS is a function of target range and received power at the antenna 6 7 Note that clutter plays a role in RCS calculations Clutter is a term used for buildings trees surfaces that give unwanted echoes When a target with a high RCS is in a low clutter area the target will be easily spotted When the same target is located in an area where there is a lot of clutter and the reflected power received back at the antenna coming from the clutter is equivalent to the power coming from the target the target will be hard to spot or won t be spotted at all 8 9 We will therefore use secondary radar digital data to locate targets in raw video so that no targets will be lost due to clutter or no clutter will be seen as targets Calculation We will first use the previously described parabolic fitting techniques to locate the target in the raw video and to calculate the amplitude of the target s reflected signal We will then convert this voltage to decibels This will give us the received power P in decibels by the antenna coming from the target Next we are able to calculate the RCS of the target Calculating the RCS of a target consists of the following steps in our implementation note that all parameters are represented in decibels 1 First the transmitted power is added to the antenna gain during transmission This value is then sub
66. Simulating user load First we determined the actions and applications that had to be simulated We used the same series of user actions as in section II B1 To simulate a normal user speed and response time we added pauses in the script The program we used for creating a script is Autolt v32 Autolt is a freeware scripting language designed for automating the Windows GUI It uses simulating keystrokes mouse movements and window and control manipulation to automate tasks When the script is completed you end with a exe file that can be launched from the command line When the script is launched it takes over the computer and simulates user activity C Monitorring and testing 1 Performance monitoring During the testing pro cess the performance has to be monitored For collect ing the data I use the Windows Performance MMC which I also used for logging the data when testing the performance see section II B2 For testing the capac ity it is important to look at how the Terminal Server uses memory Other factors to be examined are the execution speed the processor and the network usage The counters we added in the Windows Performance MMC to examine the testing results are the following e Process gt Working Set gt _ Total e Memory gt Pages Output sec e Network Interface gt Bytes Total sec e Network Interface gt Output Queue Length e Processor gt Processor Time gt _ Total e System gt Processor Queue Le
67. Smaller sized volumes naturally have smaller deviations and volumes with a more curved surface generally have greater deviations than those with less curved surfaces These deviations can t be cured by the method of calculation itself as they are caused by the difference between a real surface and a polygonal approximation Increasing the detail of a volume by increasing its resolution provides more accurate results but this is limited by the hardware specifications It is important to note that a previous version of VISIPLAN ensured an accuracy of 0 01 cm using a different line of sight calculation From the results we conclude that the studied method using ray casting is considerably less accurate for volumes with low resolutions Only boxes small sized volumes or volumes with very high resolutions can produce good results Box Cylinder Sphere V PERFORMANCE Another area of interest is the performance of the ray casting method While we didn t have access to accurate performance test results of the previous version of VISIPLAN we know that a line of sight calculation to a single point takes about 0 01 second 10 ms in scene with 30 volumes In our tests we used similar scenes of 30 boxes cylinders or spheres We also let the number of intersected volumes vary as this was expected to have a big impact on the performance due to the use of a broad and narrow phase This is done by simply moving the volumes out of the way
68. They generate more voice and data usage and supports major customers of the operator with the best quality of service They reduce churn and drive traffic from fixed lines to mobile networks They make sales of new services possible even with improving macro cell performance They prevent more costs to the infrastructure through Pinpoint Provisioning adding coverage and capacity precisely where it s needed They provide a flexible low impact and high performance solution that integrates easily with all core networks B VoIP through WAP s VolP services convert your voice into a digital signal that travels over an IP based network If you are calling a traditional phone number the signal is converted to a traditional telephone signal before it reaches its destination VolP allows you to make a call directly from a computer a VolP phone or a traditional analog phone connected to a special adapter In addition wireless hot spots that allow you to connect to the Internet might enable you to use VoIP services The advantages that drive the implementation of VoIP networks are 1 2 Cost savings Using the PSTN network will result in bandwidth that is not being used since PSTN uses TDM that dictates a 64 kbps bandwidth per voice channel VoIP shares bandwidth across multiple logical connections Hereby we get a more efficient use of the available bandwidth Combining the 64 kbps channels into high speed lin
69. V B Results For the measurements the microcontroller circuit was placed on the sensor side and then the pulses transported over the 10m long cable to the data logger inside This would log every minute the amount of pulses it registered in the past minute In Figure 3 the energy output of the PV setup is plotted together with the sensor output in an ascending order Comparison Sunnyboy 1700 blue Cellsol 200 red 1500 1000 500 energy per surface E Wh m 13 13 5 14 14 5 15 15 5 16 16 5 17 absolute time T hour Figure 3 The correlation coefficient between the 2 signals is calculated to be 99 9 resulting in that the circuit is a good indication for the energy output of the PV setup The average ratio is calculated to be 15 7 If we multiply the microcontroller output with this ratio we can see a large resemblance as shown in figure 4 Comparison Sunnyboy 1700 blue Cellsol 200 red 250 200 150 100 energy per surface E Whim 50 0 13 13 5 14 14 5 15 15 5 16 16 5 17 absolute time T hour Figure 4 IV CONCLUSION Out of the first part of our research we conclude that the Cellsol 200 sensor is linear and that there is a high correlation with the power output of the PV setup After implementing the microcontroller circuit together with the Cellsol 200 sensor we become a correlation coefficient of 99 9 between its output data and the energy output of the PV setup What indicates that this setup i
70. ad of just a traditional desktop environment has many advantages This paper illustrates the difference between using one of those regular workstations and using a virtual desktop on a Terminal Server by setting up an RDC session Performance testing indicates that the Terminal Server environment is 24 faster and handles resources better We have also done capacity testing on the Terminal Server which results in the number of users that can connect to the server at the same time and what can be done to increase this The company this research has been conducted for desired forty concurrent terminal users Unfortunately our results turned out that at this moment only seven users can be supported without extending existing hardware memory and CPU I INTRODUCTION Windows Server 2003 has a Terminal Server com ponent which allows a user to access applications and desktops on a remote computer over a network The user works on a client device which can be a Windows Macintosh or Linux workstation The software on this workstation that allows the user to connect to a server running Terminal Services is called Remote Desktop Connection RDC formerly called Terminal Services Client The RDC presents the desktop interface of the remote system as if it were accessed locally In some environments workstations are configured so that users can access some applications locally on their own computer and some remotely from the Terminal Server I
71. akes around 10 seconds and the generation of a report around 36 seconds This falls within the frustration standards explained above in II RESPONSE TIME V CONCLUSION The SaaS application IOS Mapper of IOS International is independent of the quality of a contemporary pc used on the client side and the SaaS application is also independent of the bandwidth of the used Internet in assumption that every user has a broadband Internet As the IOS Mapper application will be more heavily loaded the response time will increase directly proportional to the user loads We showed also that the actual infrastructure meets the expected response time for an application load of 10 editors and 90 viewers This is the current clientele but it will quickly expand in the future Due to the directly proportional increase between the response time and the user loads OS International can at conscription of new clientele stipulate the expected response time and intervene prematurely to improve the SaaS infrastructure without overpowering the infrastructure These load tests can also be used in the future to control new updates of IOS Mapper A sudden increase of response time for a certain request under the same user load indicates a bug in the application These bugs can then be fixed in advance without the user having to face these bugs ACKNOWLEDGMENT I want to thank Brigitte Quanten for the linguistic advice REFERENCES 1 Yunming P M
72. an intersection found in a less detailed model This method could guarantee a much higher accuracy without the need to calculate an entire model in a high resolution VII REFERENCES 1 A Watt 3D Computer Graphics Addison Wesley 2000 pp 517 519 2 A Watt 3D Computer Graphics Addison Wesley 2000 pp 356 3 Ray Tracer Specification Available at http staff science uva nl fontijne raytracer files 200208 O1_rayspec pdf February 2010 pp 5 4 CS465 Notes Simple ray triangle intersection Available at http www cs cornell edu Courses cs465 2003fa homewo rks raytri pdf February 2010 pp 2 5 5 E Haines Point in Polygon Strategies in Graphics Gems IV P Heckbert Academic Press 1994 pp 24 26 Interfacing a solar irradiation sensor with Ethernet based data logger David Looijmans Jef De Hoon Paul Leroux IBW K H Kempen Associatie KULeuven Kleinhoefstraat 4 B 2440 Geel Belgium Porta Capena NV Kleinhoefstraat 6 B 2440 Geel Belgium david looijmans portacapena com jef de hoon portacapena com paul leroux khk be Abstract In this paper will be described how we interfaced the Carlo Gavazzi CELLSOL 200 Irradiation sensor with the Grin Measurement Agent Control data logger For this we are required to test the sensor if its output is linear to its input And also to build and calibrate a microcontroller based circuit to interface the sensor with the data logger This is require
73. annels are channels that carry voice calls These channels need a decent supervision so that appropriate call connect and call disconnect signaling be passed between end devices Codecs the job of a codec is the coding and decoding translation between analog and digital devices The voice coding and compression mechanism used for converting voice streams differs for every codec C Implementation type choice With careful consideration to both implementation methods which enable mobile communication we opted in favor of placing multiple WAP s and enabling VoIP protocol on the network The implementation cost of using WAP will be considerably higher in comparison with picocells but the expenses of making telephone calls internally will considerably decrease Above a decrease of the call cost the improved security explained in the advanced features section above was also a decisive factor for making this choice D Site survey The choice about the type of implementation was made purely on experience at De Warande Therefore I opted to make a small site survey on my own Hereby I used the following steps 4 to perform my site survey 1 Obtain a facility diagram in order to identify the potential RF obstacles 2 Visually inspect the facility to look for potential barriers or the propagation of RF signals and identify metal racks 3 Identify user areas that are highly used and the ones that are not used 4 Determine pr
74. any spherical objects a line of sight calculation using the ray casting method may take a lot longer than the old method VI CONCLUSION In this paper we showed a method for creating a line of sight between two points in a rendered 3D world Bounding volumes are used as a first crude filter to reduce the workload The intersections with polygonal models are then calculated by looking at each triangle of the model After finding the intersection with the plane of a triangle it is checked whether the intersection is located within the triangle The test results show that the method itself is accurate but deviations can be significant if the model isn t detailed enough We also conclude that the performance is problematic A scene consisting of many boxes and other not too complicated volumes can provide the desired accuracy at a very high performance level More complicated scenes with many spherical objects will struggle either with the accuracy or with the performance of the calculations An idea for future work would be to investigate the use of multiple versions of each model at different resolutions where indices of polygons in a more detailed model could be traced back to indices of polygons in a less detailed model at the same location of the surface The line of sight calculation would start with the least detailed model and work its way up through the more detailed versions only calculating the polygons near the location of
75. application It even lets our virtual machines disappear from the virtual machines list We find one solution to the problem untill now it is the only solution available but this solution works The solution is to add the directory of the virtual machines that are created to an exclude list of the trend micro real time scan application You can now say that there is no virus protection on our virtual machine but there is a workaround The proper way to protect the virtual machine is to exclude the virtual hard disk directory from the scanning list in the Trend Micro real time scan application and to install the Trend Micro real time scan on the virtual OS With these modifications the virtual machine starts to run the ActiTime installation can start and thereafter the virtual machine is known in the company as the ActiTime server D Why hyper V The ActiTime application contains several components such as the Apache Tomcat web server and the Java Virtual Machine These components can disrupt other processes or components installed on a Windows server machine Therefore there must be a proper selection of the right servers we can possibly use to accomplish the role of the ActiTime application In case of Flanders DRIVE the new infrastructure exist of several servers to choose to install the ActiTime application However no server is found to install the role to There are no specific rules to determine which server is best to choose to insta
76. argets with a RTG Radar Target Generator and inject these into the radar system so that radar performance only depends on the radar system itself 4 A Parabolic Fit Since a target s echo takes the form of a parabola in slow time we can use parabolic fitting to calculate an error number that resembles the difference between the slow time video and a parabola Fig 4 This error number can then be used to assess radar performance target slow time video parabolic fit error gt parabola Fig 4 Parabolic fit error of a target s slow time video return Another use of parabolic fitting 1s locating a target Since a target s slow time video return has a parabolic form it is easy to locate a target surrounded by noise using a parabolic fit Fig 5 Calculation Using the range and azimuth or X and Y from secondary radar data we are able to locate the target in raw video Fig 1 We will filter this target out of the raw video using a window Next we will take a look at each range in slow time as 1S shown in Figure 5 A LIRIKA 2 eN aAA n Fig 5 Slow time raw video of a target Each line in this figure represents slow time video at a certain range the higher the number the higher the range If we now cross correlate each of these lines with a given parabola and calculate the maximum correlation for each line we will have the best fit for the parabola with each of these given lines Of course
77. ase The broad phase exists of a simple inexpensive test we can use once per volume instead of per triangle to eliminate the volumes that won t be intersected This is accomplished with bounding volumes 1 The narrow phase uses a more complex test to find the exact coordinates of the intersection of the ray with a polygon which is discussed in the next section A bounding volume is defined as the smallest possible volume entirely containing the studied object In addition the bounding volume must be easily tested against intersections with a ray Three types of bounding volumes are used often spheres AABBs axis aligned bounding boxes and OBBs oriented bounding boxes OBBs generally enclose objects more efficiently than the other volumes but have more expensive intersection tests A sphere has a lower enclosing efficiency but it also has the cheapest intersection test 2 In addition a sphere is easier to describe than an oriented box For these two reasons we chose spheres as our bounding volumes A bounding sphere is easily described by determining its center point and radius which can be easily calculated based upon the polygon mesh 3 Since our primitive volumes are generated from mathematic formulae however it s easier to find the center and radius analytically The vertices of a cylinder for example are generated from a height a radius and a position vector that serves as the center point of the bottom circle It is
78. ation consisting of a scene of 3D objects and at least one radioactive source is used to calculate the radiation dose at a specific point in space The radiation originating from a source may pass through several objects before it reaches its destination decreasing in intensity To calculate the attenuation caused by each object the source model is covered by a random distribution of source points each having its own ray to the studied point This is where the line of sight calculation enters the picture It is used to calculate the distances through each material by finding the intersection points on the surfaces of the objects which in turn are submitted to further nuclear physical calculations to find the dose corresponding to a single source point It should be noted that the application requires both the geometry and the material concrete iron water of each object as this information is vital in further calculations The details considering the nuclear physical models fall outside of the scope of this paper Once a method for calculating the dose in a single point is developed it can be used in a number of applications One application is the creation of a dose map A dose map is a 2D map that uses colour codes to indicate different intensities VISIPLAN allows the user to define a rectangular grid of points with adjustable dimensions and intervals along the width and length of the grid The line of sight calculation introduced earli
79. ation that provides a common interface for operating system device configuration and power man agement of both entire systems and devices The ACPI specification defines a hardware and software interface with a data structure This large data structure is pop ulated by the BIOS and can be read by the operating system to configure devices while booting It contains information about ACPI hardware registers what I O address they can be found at and what values there may be written to The objective is to power off a simulation device In ACPI terms this maps to the global system state G2 S5 named Soft Off No context is saved and a full system boot is required to return to the GO SO Fully Working system state 2 1 ACPI compliant hardware implement various registers blocks into the silicon The Power Management Event Block includes the Status PMla_STS and Enable PM1a_EN register They are both combined to a single event block PMla_EVT_BLK This event block is used for system power state controls processor power state power and sleep buttons etc If the power button is pressed a bit will raise in the Status register If the corresponding enable bit is set a Wake Event will be generated Hardware Interface Another block is the Power Management Control Block PM1a_CNT_BLK and can be used to transition to a different sleep state This block can be used to power off the device The General Purpose Event register block c
80. aused by one of these sources we construct a ray connecting the point and the source The intensity of the dose depends on the type and thickness of the materials it crosses The aim is to find the distances traveled along the ray through each volume In essence this problem is reduced to determining which volumes are intersected and finding the coordinates of these intersections A solution using ray casting a variant of ray tracing is presented i e a method using ray surface intersection tests In this case ray triangle intersections are used Because polygon mesh models are only approximations of real surfaces the intersections deviate from the real world values We test the intersection values for each volume type against real world values and conclude that the accuracy is highly dependant on the accuracy of the model itself I INTRODUCTION To understand the importance of this work it is necessary to introduce the VISIPLAN 3D ALARA planning tool a computer application used in the field of radiation protection developed at the SCK CEN Radiation protection studies the harmful effects of ionizing radiation such as gamma rays It aims to protect people and the environment from those effects An important concept in this field is ALARA an acronym for As Low As Reasonably Achievable ALARA planning means taking measures to reduce the harmful effects e g by using protective shields by reducing the time spent near radioactive
81. big source of delay is the filter bank Almost every audio codec uses a filter bank This filter bank can be a MDCT modified discrete cosine transform or a QMF quadrature mirror filter filter Both the Philips Subband Coder and the G 722 codec use a QMF filter bank The delay introduced by these filter banks results from the shape and length of the filters When calculating this delay for the Philips Subband Codec at 32kHz sampling rate and a filter length of 80 the delay becomes 2 5 ms It becomes clear that half of the total delay comes from the filter bank Since the total delay for this codec is 5ms 2 Calculating the system delay for orthogonal filter banks is done with the following formula s 3 In these formula s N 1s the delay in number of samples N filterngen 1 del u elay fs B Prediction There are two ways to use prediction in coding Block wise prediction and backwards prediction When using block wise prediction a block of data is analyzed Hence the minimum delay introduced by this operation is equal to the block length When backward prediction is used the prediction coefficients are calculated on the base of past samples Therefore there is no delay because there s no need to wait on samples Only the G 722 codec uses prediction the Philips Subband Coder doesn t But since the Philips Subband Coder also encodes the samples in blocks a delay is introduced equal to the block length C Delay
82. bjective based traffic management to define what resources the network should deliver to each end user application flow The enterprises need to define which applications matter the most for them and what the criticalities are for their business An application with a high criticality is an important application for the business An application with a lower criticality can tolerate lower quality in a time of high demand There must also be set a per user service level for each application This per user service level defines what the network should deliver in terms of network resource for each user of a given application IPlengines exchange real time information about the flows they are controlling If the cooperating Plengines detect that they are both sending to the same destination they dynamically compute the bandwidth for each user session to this destination This computation or dynamic bandwidth allocation DBA is based on their shared knowledge of the traffic mix its business criticality and the available resources at the destination The destination doesn t have to be equipped with an appliance to prevent congestions This is also called cooperative tele optimization Ipanema s smart packet forwarding forwards packets that are belonging to real time flows Jitter delay and packet losses are therefore avoided Ipanema s smart path selection dynamically selects the best network path for each session in order to maximize applicati
83. blem for classification becomes in primal weight space 1 n min Bw esa w w a e k l w b e with Y w o X b 1 e k 1 N The classifier in primal weight space takes the form y x sign v 08 8 met we R bER After using Lagrange multipliers the classifier can be computed in dual space and is given by y x ie AAN with K x X p x p X cand b are solutions of the linear systeem with 1 1 1 Og YY p X p X and a positive definite kernel K X x AX OX It would be nice if we could solve the problem in primal space but then we need an approximation of the feature map We can handle this through random active selection with Renyi entropy criterium After this Nystr m approximation we have a sparse prediction comparison y x w G x b metw E R With that featuremap approximation we can then solve a ridge regression problem in primal space with the a sparse representation of the model which is the core of the FS LSSVM algoritme II LAPACK The mathematical core of FS LSSVM is finding the solution for a system of linear equations A general available standard software library for solving linear systems is the Linear Algebra PACKage LAPACK It depends on another library the Basic Linear Algebra Subprograms BLAS to effectively exploit the caches on modern cache based architectures Many different implementations of the LAPACK and BLAS library co
84. bs For each subpopulation we calculate W for different amounts of nodes available for that population We now add W ge and W gp 48 n eguals the number of nodes used The optimal solution can be where s and p indicate the type of the job and n found by finding n where the sum of Wis and Was 1S minimal The reason why n was chosen to represent a node and not a processor is because the queues will be static and therefore one node cannot be part of two different queues This same calculations has been done for subpopulations created by dividing the main population into 3 groups containing jobs from one department After discussing these calculations with the three departments and taking into account the company policy about the cluster usage for the different departments we decided in favor of one high performance queue and using the rest of the nodes will be placed in dedicated queues A high performance queue has the benefit of having few slots approximately 4 but with each slot many nodes This results in higher throughput Even long lasting jobs can be handled within days or weeks The disadvantage of a high performance queue is the lack of interactivity In reality it may occur that a job contains a fault in the input file In case of a fault the job will loose the reserved slot and will have to wait back in the queue To prevent users from having to use a regular queue for testing purposes we will accommodate the queuing
85. burst used in all Pentium 4 processors and a MKL as LAPACK library Pentium D 920 3 20 GHz as test CPU Core lower frequency but more efficient than the Netburst chosen CPU is a Core2Duo E6300 1 86 VI IMPLEMENTATION GHz Nehalem has a focus on performance with a Xeon We will handle the implementation in C in this section of E5506 2 13 GHz to test the paper All test are performed on Windows XP SP3 operating There are 4 important requirements we must try to realize system during this new development Memory usage we have to keep the overhead over redundant data as low as possible Goal is having an algorithm that can handle larger matrices than Me LATAE RESUMIR with MATLAB We will deal with this requirement by using pointers of C Two kind of results are available de time performance Performance We hope we dealt with it by choosing results and the efficiency results the most performant LAPACK library Datatype it would be nice if the algorithm would DGESV Core2Duo E6300 1 86 GHz also work for floats in stead of doubles Then one can test the accuracy of floats compared to doubles 90 if floats would be accurate enough than FS 70 e LSSLVM can handle larger matrices This mee requirement will be fulfilled when we use C ad Me templates 10 Code maintenance It is very import to keep de 0 2000 4000 6000 Booo 10000 12000 code structure as equal as possible with the Me MATALB code Cha
86. can use the code below to calculate the resize factor of all points We have split the code into an x component and a y component It is also possible to calculate the global resize factor with a little change LOLOrigxXDist Math Sqrt Math Pow commonOriginalPoint X originalPoint Value X 2 totorigYDist Math Sgrt Math Pow commonOriginalPoint Y originalPoint Value 1 ZJ selectedObject Resize totNewxXDist totOrigXDist MTObject Width newPoints Count 2 0 totNewYDist totOrigYDISt MTObject Height newPoints Count 2 0 i S u KG 4 Figure 7 Enlarge Shrink pull apart with hands and fingers 4 AQA AS Figure 8 Enlarge Shrink pinch and splay fingers 4 Zoom in Zoom out The zoom in and zoom out function see Figure 9 is very similar to the enlarge and shrink function explained before The only difference is that the resize function is applied on the background or parent container of the object This means that the resize factor of all the objects in the parent container needs to change depending on the resize factor SON Figure 9 Zoom in Zoom out 4 Open double tap This action see Figure 10 can be detected by using rapidly two single select taps after each other Because this is no standard gesture in Silverlight 3 0 we have to create this event manually The key question of the double click event is his time out This must be carefully chosen s
87. ccess Licenses Each CAL enables one user to connect to a Terminal Server At the moment the company has two Terminal Servers available so ideally they would like each server to support forty users By testing the capacity of each Terminal Server we can determine the number of users each server can support and discover which upgrades can be done to raise this number to the desired level A second part is testing the performance of working with a Terminal Server compared to working without a Terminal Server and just a workstation for each user which is the current way of working in the company II PERFORMANCE TESTING A Intention The purpose of the performance testing is to compare the use of a traditional desktop solution with the Ter minal Server solution which provides a virtual desktop We want to examine if users experience a difference between the two solutions in the field of working speed load times and overall easiness of use To do this a user manually performs a series of predefined tasks on both the desktop and the virtual desktop For the users the most important factor is the overall speed of the task This speed will be different at both tests because the speed of opening programs and loading documents on two different machines is never the same B Collecting data 1 Series of user actions The series of actions that a user has to perform during this performance testing consists of three parts The user needs to exe
88. ced to run on only one processor on one hand and parallel jobs which may use processors different nodes on the other hand Secondly different company policies will have to be incorporated Those company policies are not unambiguous but perhaps one of the most important elements of scheduling jobs that belong to different departments within a company A third challenging problem encountered is the lack of uniformity In order to develop a fair system we need to acquire a legitimate information system to log the usage of each department user etc with the purpose to use that information for future scheduling decisions Since the Ferm2 cluster that will be using the scheduling system explained in this paper has grown over many years and the fact that many different systems i e nodes are part of this cluster it is intricate to compare node usage In this paper we are going to discuss a manner of calculating a solution which approaches a optimal solution for a system Furthermore we will explain how to overcome the lack of scheduler to distinct between different processors en use this distinction to correctly calculate de fairshare targets for different users of groups I INTRODUCTION Parallel processing already existed in the minds of cpu designers in 1940s and 1950s Parallelism that would today be called horizontal microcode appeared in Turing s 1946 design of the Pilot ACE 1 and was carefully described by Wilkes 2 Indeed in
89. ceive an error on their request grows exponentially This means that with 100 users there are not 100 users who generate a report but only 65 of them If we take this into account and in the graph we only show the users who effectively generate a report again we will get a directly proportional increase as shown in figure 6 Severalsimultaneous users r 70 500000 60 response time ms 400000 50 40 mmm number of errors 30 300000 200000 20 number of generated reports 100000 L 19 0 0 0 20 40 60 80 100 Fig 5 Several simultaneous users up to 100 users Effective number of users who generates a report 500000 60 400000 zie 40 response time ms 300000 30 200000 20 100000 10 0 0 0 20 40 60 Fig 6 Effective number of users who generate a report If we further increase the number of users we can see in figure 7 that the SaaS application IOS Mapper can generate a maximum of 67 reports simultaneously This will result in an average response time of 580 seconds or 9 5 minutes We can also see that the number of users that effectively generates a report without receiving an error decreases from this point on to 700 users From this point on no one can generate a report anymore The server won t respond to anything Severalsimultaneous users 600000 r 70 60 gr response time ms 500000 400000 40 mmm number of errors
90. conds 3 9 MB s 1 12 2 MB s 3 3 8 MB s 1 10 5 MB s 3 seconds seconds second seconds Table 1 Riverbed Technology results FTP server 65 5 KB s 61 seconds 65 5 KB s 61 seconds 589 8 KB s 6 seconds Virtuele server IP adres 192 168 1 40 24 Default Gateway 192 168 1 254 Physical equipped sites Physical and virtual equipped sites Management station 2 sites 4 sites 100 sites 2 sites 4 sites 100 sites IP reporter Etherneti NET Ethernet NET2 network network network network network network IP boss i eee Sees eee eee r r r r Port 2 o T gi ay With 15325 27 52944 46 1003083 51 15325 27 52944 46 936546 89 a jp 921581024 ar 2162075 LN Telindus Ipanema IPlengine 5ax Ipanema IPlengine 120ax maintenance r AN en Eaters es Daenen Without 14670 08 50937 49 964300 66 14670 08 50937 49 900095 74 Management Default Gateway Telindus 192 168 2 254 z gt WAN Simulator Dre maintenance azur Ju With 13171 45140 73 854724 65 13171 45140 73 798225 51 Ethernet1 NET4 Telindus IP adres 192 168 1 254 24 5 Ethernet2 NET2 maintenance ien IP adres 192 168 2 254 24 oe eae and 192 1681 254 discounts a Without 12515 81 43133 76 815941 80 12515 81 43133 76 761774 36 Figure 4 Ipanema Technologies lab Telindus maintenance d en image_rbt_cm an FileZilla_3 2 6 image_rbt_cm Filezilla_3 2 6 21 5 n23 i discounts ex File name 1_win32 c_2_
91. ction device Determines mode ltr 48 kbit s Input Auxiliary data channel output 0 8 or 16 kbit s Mode indication FIGURE III G 722 BLOCK DIAGRAM IV G 722 CODEC The G 722 codec as specified in 8 is used with a sampling frequency of 16 kHz In the hardware that is used to test the Philips Subband Coder the G 722 codec is implemented in hardware In this setup a sampling frequency of 20 48 kHz is used The operation of the codec is identical the difference is that the bitrate goes up from 64 kpbs to 81 92 kpbs Because in most cases the standard 64kpbs is used the codec is discussed here at a sampling rate of 16 kHz The G 722 codec can operate in 3 modes In mode 1 all the bits available are used for audio coding in the other two modes an auxiliary data channel is used Since this data channel isn t useful for this application only mode 1 is discussed Figure 3 shows the block diagram for the encoder and the decoder A Quadrature mirror filters QMFs In this codec two identical quadrature mirror filters are used At the encoder side this filter is used to split the 16 kHz sampled signal with a frequency band from 0 to 8kHz into two subbands These two subbands are called the lower subband O to 4 kHz and the higher subband 4 to 8 kHz these subbands are sampled at 8 kHz These subbands are represented by the signals x and xn The receiving QMF at the decoder is a linear phase non re
92. cursive digital filter Here the signals coming from the ADPCM adaptive differential pulse code modulation decoders r and ry are interpolated The signals go from 8 kHz to 16 kHz and are then combined to produce the output signal Xou which is sampled at 16 kHz B ADPCM encoders and decoders In G 722 two ADPCM coders are used one for the lower and one for the higher subband This discussion is limited to the encoders since this is the most important step in the coding process For a complete overview of the decoders the reader is referred to 8 1 Lower subband encoder The lower subband encoder will produce an output signal of 48 kpbs so most of the available bits go to the lower subband This is because G 722 is a speech codec and most information of human speech is situated in the 0 to 4 kHz frequency band The adaptive 60 level adaptive quantizer produces this signal The input for this quantizer is the lower subband input signal substracted with an estimated signal The quantizer uses 6 bits to code the difference signal The feedback loop is used to produce the estimate signal An adaptive predictor is used to produce this signal A more detailed discussion about this decoder may be found in 8 Figure 4 show the complete block diagram of the lower subband decoder u 60 level IL 48 kbit s XL gt adaptive quantizer Delete the two LSBs A Quantizer 15 level inverse adaptive guantizer dit L Adap
93. cute these actions at a normal working speed one after another To eliminate errors as a result of hazards the series of actions are performed multiple times on both desktops We than take the average of these results to draw the conclusions First the user opens the program Isah and performs some actions Next the user opens Valor Universal jviewer and loads a PCB data model Thereafter the user opens Paperless which is an Oracle database and loads some documents Finally the user closes all documents and programs after which the test ends 2 Logging data During the execution of the ac tions data has to be logged This can be done in two ways by using a third party performance monitoring tool or by using the Windows Performance MMC Microsoft Management Console snap in The first way offers more enhanced analysis capabilities but is also more expensive For this reason we use the MMC which has sufficient features in our situation In the MMC we can add performance counters that log to a file during the test After the test the file can be imported into Microsoft Excel to be examined For this performance test we need to choose counters to examine the speed of the process and the network usage These are the most important factors Therefore the counters we add are e Process gt Working Set gt _ Total e Memory gt Pages Output sec e Network Interface gt Bytes Total sec e Network Interface gt Output Queue Lengt
94. d have a thorough understanding of certain required elements and protocols in an IP network VoIP includes these functions Signaling To establish monitor and release connections between two endpoints generating and exchanging control information is necessary This is done by signaling it To do voice signaling we need the capability to provide supervisory address and alerting functionality between nodes VoIP presents several options for signaling like H 323 Media Gateway Control Protocol MGCP Session Initiation Protocol SIP 3 We can do signaling through a peer to peer signaling protocol like H 323 and SIP or through a client server protocol like MGCP Peer to peer signaling protocols Peer to peer signaling protocols have endpoints that have onboard intelligence that enables them to interpret call control messages and initiate and terminate calls Client server protocols on the other hand lack the control intelligence but communicate to a server call agent by sending and receiving event notifications For example When a MGCP gateway determines a telephone that has gone off hook it does not know to give a dial tone automatically In this case the call agent informs the gateway to provide a dial tone after the gateway has send an event notification to it Database service includes access to billing information caller name delivery toll free database services and calling card services Bearer control Bearer ch
95. d it is possible to implement many multi touch gestures such as Single select 99 66 tap Select group hold and tap Move drag Pan drag hand Enlarge Shrink pull apart with hands Enlarge Shrink pull apart with fingers Enlarge Shrink pinch Enlarge Shrink splay fingers Zoom in Zoom out pull apart with hands Open double tap For accessing data it can easily make use of webservices like Windows Communication Foundation WCE for pulling data out of a database by using the secure and reliable Model View Controller MVC model REFERENCES 1 N Van den Vonder and D De Quint Multi Touch Screen Artesis Hogeschool Antwerpen 2009 pp 1 83 2 J Y Han Low Cost Multi Touch Sensing through Frustrated Total Internal Reflection Media Research Laboratory New York University New York 2005 pp 115 118 3 M Balliauw ASP NET MVC Wisdom Realdolmen Huizingen 2009 pp 1 13 4 J O Wobbrock M R Morris and A D Wilson User Defined Gestures for Surface Computing Association for Computing Machinery New York 2009 pp 1083 1092 5 K Dockx Microsoft Silverlight Roadshow Belgium Realdolmen Huizingen 2009 pp 1 21 Comparative study of programming languages and communication methods for hardware testing of Cisco and Juniper Switches Robin Wuyts Kristof Braeckman Staf Vermeulen Abstract Before installing a new switch it
96. d we combine 10 seconds of data from the close talk microphone s n and the handheld microphone for noise x n to form the desired signal d n The signal d n acts together with x n and the corresponding parameters as input for the 3 algorithms The parameters are for e LMS convergence factor u and filter length L e NLMS filter length L and constant a e RLS filter length L Afterwards we calculate the SNR gain for the different algorithms The SNR gain in dB 1s calculated by taking the difference in SNR between the converged enhanced signal and the desired signal d n The results for LMS NLMS and RLS can be found in figure 7 8 and 9 respectively We decide to use LMS as adaptive algorithm for the GSC To obtain the same SNR gain as LMS with a convergence factor of 0 0050 NLMS has to use larger filter lengths Next LMS is much faster per iteration than RLS Certainly for the greater filter lengths Finally LMS is also much easier in implementation So taking all these factor into account we choose for the implementation of LMS as algorithm for the GSC SNR gain dB 10 0050 40 0 01 1 0 05 o l mn emmen p 0 1 ee mel 5 u o M O __ Same g LS s LO U O L 8 LG 32 64 128 Fig 7 LMS influence of the factor u on the SNR gain SNR gain dB one eooo sT a t sr
97. d to reach a sample rate of 1Hz or higher to get an accurate energy integral estimate I INTRODUCTION AND RELATED WORK pe Capena is an energy awareness company that provides a web based interface Ecoscada Ecoscada supplies customers with information about their energy and natural resources usage Locally placed data loggers log sensor and meter data and send it to the Ecoscada database over Ethernet or GPRS This data can then be accessed through the web based interface With the growing amount of photovoltaic PV solar panel installations there is also an interest in the possibility of confirming if such an installation provided as much electrical energy as it should have done For this measuring of the solar irradiation is needed The system for now makes use of the Grin Measurement Agent Control MAC an Ethernet based data logger The MAC provides 4 Digital outputs 4 Digital inputs pulse counters 4 PT100 inputs 4 Analog inputs and 1 wire sensors As well as a 7 5v supply voltage and a calendar function The Sensor provided for measuring the solar irradiation is the Carlo Gavazzi Cellsol 200 it s a silicon mono crystalline cell that works on the same photovoltaic principle as solar panels 4 The sensor we are provided with is calibrated to give a 78 5mV DC signal at an irradiation of 1000W m and the sensor has a range from O to 1500W m Because there was no information provided about the linearity of this sensor the fir
98. d to be included in our new implementation Finally a comparison in terms of computational complexity and memory usage is performed on a MATLAB and C implementation of the FS LSSVM algorithm Index Terms Fixed Size Least Squares Support Vector Machines kernel methods LAPACK C I INTRODUCTION E this work an optimized implementation in C for the large scale machine learning algorithm called Fixed Size Least Squares Support Vector Machines FS LSSVM which was proposed in 1 is presented Although this algorithm was already found competitive with other state of the art algorithms no detailed discussion about an optimal implementation was studied This paper concerns the latter since an optimal program might result in handling even larger data sets on the same computer system The FS LSSVM algorithms resides in the family of algorithms which all are strongly connected to the popular Support Vector Machines SVM 2 which is the current state of the art in pattern recognition and function estimation Least Squares Support Vector Machines LS SVM 3 4 simply the original SVM formulation While SVM boils down to solving a Quadratic Programming QP problem the LS SVM solution is found by solving a linear system Using a current standard computer the LS SVM formulation can be solved for large data set problems up to 10 000 data points using of the Hestenes Stiefel conjugate gradient algorithm 5 6 In order to solve an
99. designing process Let us explain this using a small example If a network designer needs to implement a wireless network in a certain building and he knows the different advantages and disadvantages of both implementations he can choose between the placement options solely on experience This will result in a lower cost of implementation Suppose he would choose for the WAP implementation knowing that a WAP costs 200 to 300 and a complete site survey of the complex would cost 5000 to 7000 In this case it would be cheaper to just add a few WAP s here and there to ensure maximum coverage over a certain area then doing the survey The downside here is that the designer will never know the RF behavior in the complex what can lead to rather clumsy situations when a problem arises Some problems are not knowing where coverage holes are or areas of excessive packet loss The same example can be made with the use of picocells IL RESEARCHING POSSIBLE IMPLEMENTATION OPTIONS A Picocells To extend coverage to indoor areas where outdoor signals do not have a good reach it is possible to use picocells to improve the quality of voice communication These cells are designed to provide the coverage in a small area or to enhance the network capacity in areas that have a dense phone usage A picocell can be compared to the cellular telephone network It converts an analogous signal to a wireless one The key benefits of picocells are
100. dition PHP WampServer 2 01 with PHP 5 3 0 used packages Perl 4 5 6 7 Python 8 9 10 11 12 Ruby 13 14 15 16 Java 17 18 19 C 20 21 PHP 22 TABLE V REQUIREMENTS prompt before sending a new command Some modules or packages already contains this wait command Mostly they use a sleep command for a specific period which is extremely unefficient We wrote our own wait function similar for every language Is this wait function written as fast as possible Probably yes but if not it will not influence the result because each languages uses this function Another example is the ping command It is possible to add options to the ping command like the number of echo requests and the time out time Each language uses the same options especially 4 echo requests and 3000 ms time out time An ICMP ping is used instead of a TCP or UDP ping Table 12 4 shows the result of the benchmark 10 results language are measured to minimize effects caused by coincidence Not only speed but also memory usage and page faults have been taken into account The latter two are not mentioned because no significant differences could be found Java needs more memory but nowadays memory became very cheap Perl 01 45 546 01 43 796 01 48 546 01 43 889 01 44 780 01 42 093 01 40 705 01 42 515 01 46 440 01 43 906 104182 ms o 2142ms Ruby 02 21 156 02 18 593 02 20 296 02 20 530 02 15 999 02 17 943 02 26
101. e Small Business Server 2003 instead of the defined Windows XP client After setting up the network the software will be tested on CPU DISK memory and network performance This part is done by Windows Performance monitor 2 3 and WireShark 4 for the network part Because it is a small network the statistics that we become will be in a non working network this results in a lower network load then in real time Keeping this in mind we can start the simulations Later on we will put the best tool for the company in a real time networking environment HI SIMULATIONS The benchmark consists of tests that represent a server environment in real time Following fields will be tested 1 A non successful NTBackup of the test txt file which will result in an error in the application log file 2 Full configured Perfomance monitor onboard Windows testing tool with the following parameters a DISK scale 0 300 1 Disk read sec u Disk write sec ui Transfers sec b CPU scale 0 100 i CPU average c RAM 1 committed bytes Table 3 Test procedure Another specific requirement 1s the ability to start the service automatically when it goes down the T specialist does not have to intervene A Benchmarks The tools listed before are all tested for the specific 30 minutes testing procedure because of the large scoop of test results we will limit the results to the summarization of CPU DISK memory and network perfo
102. e On Lan rev 2 Available at liebsoft com 2006 12 W Richards Stevens TCP IP Illustrated Vol 1 The Protocols Addison Wesley ISBN 0201633469 2002 Catalog nr Catalog nr 316972 004 Analyzing and implementation of Monitoring tools April 2010 Philip Van den Eynde Kris De Backer Staf Vermeulen Rescotec Cipalstraat 3 2440 Geel Belgium Email Philipvde O pandora be Abstract The guest of analyzing monitoring tools that use the BY as SR 4 UL B Ute Services with Besadmin least of your network and server capacity to keep track of all system admin kind of resources services events disk space and BlackBerry Services One of the objectives that must be met is the automatic restart of a service when it goes offline The research starts from here First of all the tools must be tested in a standard Microsoft Exchange BlackBerry Attachment Service environment where the parameters are always the same It begins Information Store with eliminating the tools that do not have the required objectives the ten candidate tools are the ones that have it all and Microsoft Exchange BlackBerry Controller will be put in benchmark Management ke ABOE CON Microsoft Exchange BlackBerry Dispatcher N large server environments it is not obvious to manually Routing Engine monitor all running servers and services For some critical services it is even unacceptable that they go offline Therefore most company networks are auto
103. e Radar Cross Section RCS RCS is used to assess radar sensitivity e Signal to Noise ratio SNR The higher the SNR the better a target can be recognized e Pulse Compression Pulse compression processing gain will enhance detection and needs to be verified e Parabolic Fit Error We try to fit a parabola in the slow time video return of a target The difference between the slow time video return and the used parabola gives us an error number Note that we use a parabola because a radar beam can be approximated by a parabola eo These parameters can then be used to optimize or improve radar systems performance These calculated performance parameters could later on also be used to predict the performance of another equivalent radar system Offline radar system performance analysis was the first step taken to calculate the needed radar parameters This way it was easy to check if the written algorithms work correctly and if they could be used in a real time system These algorithms could then be integrated in a real time system together with a primary radar raw video data filter to filter useful data and analyze this data at the same time Real time primary radar raw video data compression is as mentioned above another step taken Data compression is important in the way of disk and memory usage If we only write data to disk that is important for future analysis there will be less memory taken and disks will be used less intens
104. e devices are capable to track multiple points of input instead of only one point This property is extremely useful for a team collaborating on the same project or computer It gives a more natural and intuitive way to communicate with the team members II SILVERLIGHT 3 0 Now that we have the hardware to test the multi touch capabilities we need the appropriate software to communicate with the multi touch device In the company Item Solutions where the research was made they introduced us to the programming language Microsoft Silverlight 3 0 Silverlight 3 0 is a cross over browser plugin which is compatible with multiple web browsers on multiple operating systems e g Microsoft Windows and Mac OS X Linux FreeBSD and other open source platforms can use Silverlight 3 0 by using a free software implementation named Moonlight that is developed by Novell in cooperation with Microsoft Mobile devices starting with Windows Mobile and Symbian Series 60 phones will likely be supported in 2010 The Silverlight 3 0 plugin SMB includes a subset of the NET framework 50MB The main difference between the full NET framework and the subset of Silverlight 3 0 is the code to connect with a database Silverlight 3 0 works client side and can not directly connect to a database For the connection it has to use a service oriented model that can communicate across the web like Windows Communication Foundation WCE Windows Communicat
105. e the BER This kind of test is not suitable because the script needs to be fast A second approach is to check the functionality of the inter faces A successful ping guarantees the interface is responding This test does not ensure that the specific interface is capable to transport an amount of data from or to another interface without any errors Therefor an FTP transfer will be used D Flowchart test operations Vlan s are necessary because data has to travel through the SlaveSwitch Below you find the vlan scheme and the corresponding traffic flow Get errors before test sea Se lt Ae prati NO Left shift Vian2 port Working _ SlaveSwitch MasterSwitch Se SE YES y SlaveSwitch MasterSwitch Ping FTP transfer X Kee YES x p lt Succes gt succesport U NO Get errors after test via succesport All ports NO tested Right shift Vlan1 port Fig 8 Flowchart Testoperations Fig 9 VLAN configuration V CUSTOM MADE BENCHMARK After the script is written it is useful to check which language is the most appropriate among those languages which are discussed at the beginning of this paper Looking at the result we consider whether or not to rewrite the script To accomplish this we designed a custom made benchmark We counted every operation which
106. e up filter from the Intel network device The system still has to decide what to do with the generated PME signal Recall the ACPI General Purpose Event register block with corresponding Enable and Status registers The Status register contains a field named PME STS that maps to the PME signal used on the Intel network device All what is left to do is set the corresponding enable bit in the Enable register When Status and Enable bit are set a wake event is generated and the system will transition to the GO SO Working state 4 3 Summary When the network device is kept powered on and configured to generate a wake event through a power management event upon reception of a Magic Packet the system will transition to the Fully Working state We can conclude that remote power on is possible and can be successfully implemented on simulation devices 5 CONCLUSION This works shows that it is feasible to implement power management features into the VxWorks operating sys tem that initially had no support for it Both remote power off and power on are successfully implemented We can conclude that all goals are achieved ACKNOWLEDGMENTS The author would like to express his gratitude to every one at Alcatel Lucent IP Division for assisting through out this work The author also wants to thank Alain Maes Erik Neel and Dirk Goethals for their assis tance and guidance during implementation of this work Thanks also go out to G
107. e used to select an object Touch FrameReported new TouchFrameEventHandler TP ActionReported TouchPointCollection tps e GetTouchPoints null foreach TP tp in tps SVILE Lp Action case TouchAction Down case TouchAction Move case TouchAction Up Figure 3 Single select tap 4 Select group hold and tap To select more than one object see Figure 4 we can reuse Code to select more objects at the same time So here we have to detect multiple select tap events for multiple objects Because there is no timer function in Silverlight 3 0 the code below can be used to make a hold function long timeInterval 1000000 100ms if DateTime Now Ticks LastTick lt timeInterval selectedObject Select LastTick DateTime Now Ticks AS RES AT Figure 4 Select group hold and tap 4 Move drag The move action see Figure 5 can be realized by using the move event in Silverlight 3 0 of a blob If a blob gives a down event followed by a move event the object must be moved equal to the movement of the blob In Silverlight 3 0 we can simply change the position of elements to change the Left and Top property of the element Figure 5 Move drag 4 Pan drag hand For this gesture see Figure 6 the method above can be reused but now we first have to detect which blobs are in the object From all the points in the object we have to calculate the midpoint by equation 1 Xit Xo
108. ecause of this 4 becomes 7 wjn 1 wh ufr hlk 6 and u n equals to 7 uni 6 A LP n In 6 we see three unknown factors First we have the factor Px n This is the power of x n at time n The power is calculated on a block of L samples Next there s a constant a The value of a lies between 0 and 2 Finally L represents the filter length NLMS solves the problem of LMS It considers the stability and optimizes the rate of convergence A drawback of the algorithm is the extra operation for the calculation of the convergence factor Recursive Least Squares Just like LMS RLS minimizes the error signal by adapting the filter coefficients However RLS uses past error signals for the calculation of the next error signal The extent to which the previous error signal counts depends from the forgetting factor A This factor is fixed but the power n 1 has as consequence that the older errors have less influence 8 So the minimization objective is 8 w arg minX 4 e i 7 v i 0 This leads to the following iterative formula for determining w n win win 1 eln s D n x n 8 where Sp n is the autocorrelation of the signal x n at time n In comparison with LMS RLS does not depend on the statistics of the signal Due to this advantage RLS converges often faster than LMS However RLS uses more multiplications 6 per update This results in a slower algorithm per iteration
109. ech signal In this manner it appears that the noise is recorded with the same microphone as the speech signal Next the error signal e n calculated by subtracting y n from d n is used to adapt the filter coefficients This adaption can happen on different ways 7 8 In this paper we discuss 3 algorithms e Least Mean Square LMS e Normalized Least Mean Square NLMS e Recursive Least Squares RLS Least Mean Square LMS tries to minimize the error signal According to 7 LMS minimizes the following objective w arg mine nl 3 by adapting the filter coefficients This boils down to iteratively solving 7 w n 1 u n ux n e n 4 In 4 u 1s the convergence factor This factor controls the stability of the algorithm and also has an influence on the rate of convergence The simplicity 1s the greatest advantage of LMS This can be seen from 4 where the only operations are an addition and a multiplication However LMS has several disadvantages If the convergence factor u is chosen too low The rate of convergence will be very slow Increasing u can solve this problem but this results in stability problems Due to a fixed convergence factor we must find a tradeoff between speed and stability Normalized Least Mean Sguare This algorithm differs from LMS in the value of the convergence factor u which depends on the time Thus there is an adaption of u every time we update the coefficients of the filter B
110. ed filter This infrared light is generated by the LED lights that are send through the acrylic pane If you put a finger on the screen the infrared light will be sent to the webcam The webcam captures this light and will be sent to the connected computer You can also notice on Figure that a projector is used This is not really necessary because the sensor webcam can be used standalone Without a projector the multi touch table is completely transparent and therefore it is particularly suited for use in combination with rear projection On the rear side of the waveguide a diffuser e g Rosco gray is placed which doesn t frustrate the total internal reflection because there is a tiny gap of air between the diffuser and the waveguide The diffuser doesn t affect the infrared image that is seen by the webcam because it is very close to the light sources e g fingers that are captured Total Internal Reflection Acrylic Pane LED Pr Scattered Light Baffle Diffuser pi Video Camera Figure 1 Schematic overview of a home made multi touch screen 2 Why multi touch The guestion is why we would use multi touch technology The problem lies in the classic way to communicate with a desktop computer Mostly we use indirect devices with only one point of input such as a mouse or keyboard to control the computer With the multi touch technology there will be a new way to human computer interaction because thes
111. ede eed ed ed ed ed Fig 1 Output from the Process gt Working Set gt _ Total counter Memory Pages Output sec E traditional workstation E terminal server 1 20E 17 1 00E 17 8 00E 16 Hit MHI 6 00E 16 HHN a 00e 16 HEIN HH 2 00E 16 HH o ooezoo oo N A m a m T Fig 2 Output from the Memory gt Pages Output sec counter Pages Output sec st AJF ede ed ed ed ed ed ed 2 Memory Figure 1 shows the output from the working set counter This counter shows the total of all working sets of all processes on the system not including the base memory of the system in bytes First of all the figure also shows the difference in execution speed we discussed in II Cl We can see that for the same series of actions it takes significantly less time to perform then on the Terminal Server desktop Another conclusion that this data shows is the mem ory usage When executing tasks on the regular desktop the memory usage varies between 400 MB and 600 MB whereas the memory usage in the virtual desktop environment varies only between 350 MB and 450 MB We can conclude that the virtual desktop uses slightly less memory than the regular desktop and the var
112. edia types uses B and D channels where B channels carry user data and D channels will direct the switch to send incoming calls to particular timeslots on the router 6 Normally the PRI will be used to make PBX to PBX calls or other internal calls and the BRI will be used when a connection to an outside network is made At the Warande it is a little different There are 8 BRI interfaces to connect to the outside world Since every BRI supports 2 channels the Warande can make 16 outgoing calls at the same time When for example a 17 user wants to make an outside call he will be routed around the network to Antwerp Here he will be connected to the telephone central that will give him an outside connection on their BRI interface Now that the outside calls can be made we have to make sure we can do internal calls This is done using a call system that is purely based on IP All the calls will travel over the network as voice packets that will be protected by configuring a Quality of Service QoS Configuring the BRI and internal IP network is not done the way students learn it Because we are configuring and managing a large amount of sites and an even larger amount of phone devices it would be too much trouble doing the installation with a console program Instead we use OmniVista 4760 This allows us to have an efficient control over all sites and on the other hand we can make changes with a few clicks A screenshot of the program can be fou
113. ee different types of nodes First there were nodes with a Intel Pentium 4 3 0 gHz processor These nodes were accompanied by nodes with a Intel Pentium 4 3 6gHz processor The third kind of nodes are AMD Opteron 3 0 gHz nodes In total this makes 48 nodes Each node has 4 logical processors In the case of the AMD Opteron those 4 logical processors consist of 4 physical processors The Intel Pentium 4 nodes only have 2 physical processors but use a Multi threading technique B Scheduling software Scheduling refers to the way processes are assigned priorities in a priority queue This assignment is carried out by software known as a scheduler It consists of mainly CPU Throughput Turnaround Waiting and Response 5 The scheduling software that was chosen by SCK CEN was Maui Maui is a scheduler developed by the Maui supercomputing center as an alternative to the default Loadleveler scheduler in their IBM SP environment 6 The Maui Scheduler is a policy engine which allows sites control over when where and how resources such as processors memory and disk are allocated to jobs In addition to this control it also provides mechanisms which help to intelligently optimize the use of these resources and generally manage the system 6 The main advantage of using scheduling software is the easy of setting different settings by manipulating elementary objects In Maui we can work with five different objects e Jobs A job consists of
114. efered because SNMP does not support every set command SET no wait operation While configuring a switch it is not necessary to wait until the previous command is really executed Note that you still need to wait for the prompt It is remarkable that SNMP is not the fastest anymore and this communication method is not influenced by command length and execution time After an SNMP set request is send an SNMP get response is received when the command is really executed So SNMP is slower because it checks automatically if the command is executed well Serial communication comes close to SNMP when we have to deal with short commands Telnet is the obvious victor The reason is already mentioned above At this point we are able to decide which communication method is the most efficient in a particular situation SNMP Serial Datalink ethernet 38 38 startbit Network IPv4 Transport 32 TCP 8 UDP stopbit Total 90 bytes 66 bytes 2 bit TABLE III OVERHEAD Blongexec long command ma Blongexec short command E Telnet S SNMP E shortexec long command B shortexec short command I 80000 70000 60000 50000 _ _ 40000 30000 20000 7 sil D Ey o r 80000 70000 60000 50000 40000 30000 20000 10000 short command long command Serial Telnet SNMP Fig 4 GET 100000 100000 30000 aao 90000 70000 80000 60000 70000 50000 30000 50
115. eliminary access point AP locations These locations include the power and wired network access cell coverage and overlap channel selection and mounting locations and antenna 5 Perform the actual surveying in order to verify the AP location Make sure to use the same AP model for the survey that is used in production While the survey is performed relocate AP s as needed and re test 6 Document the findings Record the locations and log of signal readings as well as data rate at outer boundaries Using the steps mentioned above I firstly made a theoretical site survey step 1 4 through the use of Aruba RF plan of every floor 5 floors in building A 6 floors in building B This program is able to pin point the optimal WAP locations on a certain floor where we need the 802 11 a b g wireless coverage without including the interference of concrete walls or thick glass and irradiation from other levels This is shown in the image below After this theoretical approach of the floor we need to do actual surveying on site to verify the WAP locations and make proper adjustments when needed During the survey we need to allocate possible problems When located we consider the possible level of interference it will cause and adjust the locations of the WAP s Another adjustment we need to consider is the irradiation from levels below when we are dealing with open areas since the closed areas won t have any irradiation through the
116. eneficial result 1 Load testing offers the possibility of measuring the performance of the SaaS application based on real user behavior This behavior is imitated by building an interaction script with the user requests A load generator like Jmeter then passes through the interaction script adapted with test parameters based on a real life environment on the SaaS application IOS Mapper With these load tests we can identify the reaction of the SaaS application IOS Mapper with different user loads and assess if the SaaS application IOS Mapper meets the expected real life user loads II RESPONSE TIME As mentioned in the introduction the quality of the SaaS application IOS Mapper can be measured in terms of response time So it will be very important to monitor these end to end response times to stipulate how long it lasts before the requests of the user are carried out and will be visible for the user Afterwards we can compare these results with frustration level times From studies Nah 2004 into acceptable answer times it becomes clear 2 e Delay of 41 seconds is suggested as the cut off for long delays like downloading reports e Delay of 30 seconds is suggested as the frustration level for long delays e Delay of 12 seconds causes satisfaction to decrease for normal actions like opening wizards III LOAD TESTING Load testing offers the possibility of measuring the performance of the SaaS application based on
117. er is applied to each point of the grid providing the necessary intensity values The resulting grid of values can be converted to a coloured map much like a computer screen with coloured pixels This dose map can be used to determine problematic areas areas with a high radioactive dose at a glance Another interesting application is the definition and calculation of trajectories When a person is working near radioactive material he follows a certain path or trajectory through the working area Using the line of sight method to calculate a multitude of points along the defined trajectory and taking the amount of time spent in each location into account a total dose can determined for the trajectory This allows the user to evaluate trajectories and try to find the safest route II BROAD PHASE Finding intersections between a ray and a triangulated model is generally an expensive operation Imagine there are 500 primitive volumes in a scene A simple cylinder at a resolution of 20 consists of 80 triangles while a hollow torus at the same resolution consists of as many as 1680 triangles The number of triangles in such a scene quickly adds up It s unlikely a single ray intersects every volume in a scene In many cases no more than a handful of volumes are intersected Performing expensive operations on each triangle in the scene isn t very efficient A common approach to this problem is the use of a broad phase and a narrow ph
118. erent methods we also have to consider some factors that influence making a design A Factors that influence Design When we use VolP we are sending voice packets via IP Hereby it is normal that certain transmission problems will popup Because the listener needs to recognize and sense the mood of the speaker we need to be able to minimize the effect of these problems The following factors 1 can affect clarity Echo result of electrical impedance mismatches in the transmission path Effecting components are the amplitude loudness and delay time between spoken voice and the echo Echo is controlled by using suppressors or cancellers Jitter variation in the arrival of coded speech packets at the far end of a VoIP network This can cause gaps in the playback and recreation of the voice signal Delay time between the spoken voice and the arrival of the electronically delivered voice at the far end Delay results from distance coding compression serialization and buffers Packet Loss Under various conditions like unstable network congestion voice packets can be dropped This means that gaps in the conversation can get perceptible to the user Background noise low volume audio that is heard from the far end connection Side tone the purposeful design of the telephone that allows the speaker to hear their spoken audio in the earpiece If side tone is not available it will give the impression that the teleph
119. ers at the byte seguence level Therefore even small changes of a file e g changing the file name can be detected Data streamlining works across all TCP based applications and across all TCP based protocols It ensures that the same data 1s never sent more than once over the WAN RIOS intercepts and analyzes TCP traffic Then it segments and indexes the data Once the data has been indexed it is compared to the data on the disk If the data exists on the disk a small reference is sent across the WAN instead of the entire data RIOS uses a hierarchical structure whereby a single reference can represent many segments and thus multiple megabytes of data This process is also called data deduplication FREE HEERE CI Data Reference i HHA Data Figure 1 Data references to reduce the amount of data sent across the WAN If the data doesn t exist on the disk the segments are compressed using a Lempel Ziv LZ compression algorithm and sent to the Steelhead appliance on the other side of the WAN which also stores the segments of data on disk Finally the original traffic is reconstructed using new data and references to existing data and passed through to the client C Transport Streamlining RiOS uses transport streamlining to overcome the chattiness of transport protocols by reducing the number of round trips It uses a combination of window scaling intelligent repacking of payloads connection management and other protocol
120. es Typically these procedures integrate mechanisms for signaling events during voice calls and for handling and reporting statistics about voice calls There are three protocols that can be used to implement gateways and make call control support available for VoIP 1 H 323 2 Media Gateway Control Protocol MGSP 3 Session Initiation Protocol SIP As mentioned earlier the Provincie Antwerpen uses a peer to peer signaling strategy This means that MGCP which is client server signaling can be removed from the available protocols That leaves us with H 323 and SIP H 323 is the gateway protocol used at the Warande or any other Provincial site The reason is subject to the different implementations of equipment For example The main site in Antwerp has three different kinds of telephone centrals a state of the art one and two older ones All these centrals need to be able to communicate with each other and if we would use SIP on one of them the others need to be able to support the same protocol Which in this case is impossible All the centrals do support H 323 which gives us the reason why this protocol has been used IV CONCLUSION The problem was to solve the poor coverage at De Warande and ensuring a good quality of voice communication This is possible by the use of picocells that enable voice communications through the normal PSTN network or by using WAPs with the VoIP protocol The choice made for De Wara
121. es support vector machine classifiers 1999 4 J A K Suyskens et al Least squares support vector machines 2002 5 G Golub C Van Loan Matrix computations 1989 6 J A K Suykens et al Least squares support vector machine classifiers a large scale algorithm 1999 7 T Wittwer Choosing the optimal BLAS and LAPACK library 2008 8 LAPACK benchmark Standard floating point operation counts for LAPACK drivers for n by n matrices http www netlib org lapack lug node7 1 html standardflopcount 9 C L Lawson et al Basic Linear Algebra Subprograms for FORTRAN usage 1979 10 E Anderson et al LAPACK users guide 1999 11 LibSVM Data Classification Regressin and Multi label http www csie ntu edu tw cjlin libsymtools datasets Appendix A LAPACK results Time DGESV Pentium D 940 3 20 GHz 100 5 90 80 70 60 50 4 40 4 30 5 20 4 10 5 tijd s 0 2000 4000 6000 8000 10000 12000 rijen MATLAB GotoBlas2 Ref LAPACK MKL SGESV Pentium D 940 3 20 GHz 100 al 80 70 60 50 4 40 30 20 5 10 5 tijd s 0 2000 4000 6000 8000 10000 12000 rijen MATLAB GotoBlas2 Ref LAPACK MKL DGESV Core2Duo E6300 1 86 GHz 100 7 tijd s S 0 2000 4000 6000 8000 10000 12000 rijen MATLAB GotoBlas2 Ref LAPACK
122. ese devices but also several groups of enthusiasts that are making their own home made multi touch table like the Natural User Interface group One of the methods they use is Frustrated Total Internal Reflection FTIR which was used for testing To use these devices efficiently it is necessary that new technologies are being introduced Many of the software technologies that are used nowadays are not able to communicate with multi touch devices or gestures that are made on these devices So a multi touch table that communicates with Silverlight 3 0 released in July 2009 will be presented This programming language supports multi touch but it doesn t recognize any gesture A complete description of the most intuitive gestures and how to integrate them into a Silverlight 3 0 application will be discussed We will also describe how to connect this application with a database to build a secure and reliable B2B B2C or media application I INTRODUCTION AND RELATED WORK For testing the multi touch capabilities of a Silverlight 3 0 application we used the multi touch table that was made in a previous work 1 by Nick Van den Vonder and Dennis De Quint This multi touch table was based on a research by Jefferson Y Han 2 The multi touch screen uses FTIR to detect fingers also called blobs that are pressed on the screen On Figure 1 we see how FTIR can be used with a webcam that only captures infrared light by using an infrar
123. estimated Good tools come with high prices especially when it comes to C DISK asl plementing the tool As mentioned before this is a very important part in the benchmark We do not want to lose any of the records written CPU processortime 12 000 2 000 00 10 000 8 000 1 500 00 6 000 1 000 00 4 000 500 00 2 000 0 00 0 000 Ipsentry Ipsentry Spiceworks MonitorMagic ActiveXperts Tembria server monitor ServerAssist ManageEngine PA Server Monitor Web WatchBot SolarWinds ActiveXperts ManageEngine MonitorMagic PA Server Monitor ServerAssist SolarWinds Spiceworks Tembria server monitor WebWatchBot Fig 4 Price results Fig 5 CPU results With the details listed in the following table With the details listed in the following table monitor CPU monitor Price Spiceworks 164 92 MonitorMagic 499 00 Ipsentry 520 99 ActiveXperts 0 806 ActiveXperts 690 00 WebWatchBot 0 908 ServerAssist 1 095 00 ManageEngine 1 120 69 Tembria server monitor 745 88 PA Server Monitor 1 123 69 Web WatchBot 1 495 50 Spiceworks 11 441 SolarWinds 224513 Table 7 CPU results detail Table 6 Price results detail F Memory E CPU This sections covers the same result as CPU modern servers This parameter is less important because of the high have enough memory so it wouldn t cause any problem performance of modern servers this
124. etwork After installation of the application and testing it the application works well and is accessible by the company members Thereafter a decision is made to install a new Windows xp machine on a company server This new Windows xp machine is a virtual computer that is set up through hyper V standard component of the Windows server 2008 product The application is installed in the same way it s described in previous sections While testing the application not everything is working as expected When the Tomcat Server is started the Tomcat application goes down immediately This is an unexpected behavior of the Tomcat server application With this negative behavior of the Tomcat application the ActiTime application is unable to start A search to a solution for this negative behavior can start HI WHY THE TOMCAT APPLICATION WENT DOWN To solve the problem first the following different steps are tried to determine the exact problem but they doesn t lead to a solution of the problem e Reinstallation of the ActiTime application e Reinstallation of the Apache Tomcat Server e Install a different version of the Apache Tomcat Server e Reinstallation of the java server machine e Install a different version of the java server machine These steps are used to possibly detect the problem however the problem still exists after each step After some search on the internet with the terms can t start tomcat on windows a possible solu
125. even larger set E g a computer with an Intel Core2Duo processor of problems with sizes up to 1 million of data points an approximate algorithm called FS LSSVM was proposed in 4 In 1 this algorithm was further refined and compared to the state of the art The authors there programmed the algorithm in MATLAB Such an implementation is known to be suboptimal with respect to memory usage and computational performance This is due to the fact that MATLAB is a prototyping language which enables fast algorithmic development but has the limitation that the resources cannot be accessed with full control In this work we aim at a new FS LSSVM implementation which provides solutions for the above limitations The paper is organized as follows In Section I we explained the need for a new implementation of FS_LSSVM But first will we in Section II give a small introduction to FS LSSVM In section III we will introduce LAPACK and select some candidates for a performance test Section IV explains some technical details about the test Section V will handle the test results Finally in Section VI we will implement the algorithm of which we will present the performance result in Section VII IL FIXED SIZE LEAST SQUARES SUPPORT VECTOR MACHINES In this section we will give a short introduction to LSSVM regarding classification The following steps are the same for regression According to Suykens en Vandewalle 3 the mentioned optimization pro
126. f the code significantly increases and it is designed to solve common design problems Silverlight 3 0 is not only capable to use these two concepts but there is also a minimal support for multi touch capabilities The only thing that Silverlight 3 0 can detect is a down move and up event for a blob touchpoint point or area that is detected II MULTI TOUCH GESTURES The paper User Defined Gestures for Surface Computing 4 by J O Wobbrock M R Morris and A D Wilson researched the behaviour how people want to interact with a multi touch screen In total they analyzed 1080 gestures from 20 participants for 27 commands performed with 1 or 2 hands The gestures we needed and implemented where Single select tap Select group hold and tap Move drag Pan drag hand Enlarge Shrink pull apart with hands Enlarge Shrink pull apart with fingers Enlarge Shrink pinch Enlarge Shrink splay fingers Zoom in Zoom out pull apart with hands Open double tap Single select tap For a single select tap of an object see Figure 3 it is necessary that we can detect where the user pressed the multi touch screen These coordinates must be linked to the corresponding object On this object we checked if there occurred a down and rapidly up event If these two events occur in a single object the object must be selected In Silverlight 3 0 the code below can b
127. ftp activexperts labs com samples asocket Visual 20C 4 Include 18 Regular expressions Available at http msdn microsoft com en us library system text regularexpressions aspx 19 PHP telnet 1 1 Available at http www geckotribe com php telnet 20 Philip M Miller TCP IP The Ultimate Protocol Guide BrownWalker Press 2009 21 Cisco Press CNAP CCNA I amp 2 Companion Guide Revised 3rd Edition Cisco systems 2004 22 Douglas R Mauro Kevin J Schmidt Essential SNMP 2nd Edition O Reilly Media 2005 23 Charles Spurgeon Ethernet The Definitive Guide O Reilly and Asso ciates 2000 TBW K H Kempen Associatie KULeuven Kleinhoefstraat 4 B 2440 Geel Belgium 2Telindus nv Geldenaaksebaan 335 B 3001 Heverlee Belgium
128. ful Otherwise users of the cluster will try to overcome this themselves which 1s pernicious for the utilization of the cluster Furthermore the system can accommodate continuously to changing usage of the cluster ACKNOWLEDGMENT would like to thank my professor ing Joan De Boeck of assisting me with this thesis Also specials thanks go to Prof dr F C R Spieksma Faculty of Business and Economics KU Leuven and Prof dr Roel Leus Faculty of Business and Economics KULeuven for pointing me in the right direction References 1 B E Carpenter and R W Doran Editor A M Turin s ACE Report of 1946 and Other Papers MIT Press Cambridge Massachussetts 1986 2 M V Wilkers The best way of designing a automatic calculating machine Proc Manchester University Computer Conference 1951 pp 16 18 3 J B Stringer Microprogramming and the designs of the control circuits in electronic and digital computing The Cambridge philosophical society Part 2 1953 pp 230 238 4 Avi Silberschatz Peter Baer Galvin Greg Gagne Operating System Concepts John Wiley amp Sons inc 2004 isbn 0 471 69466 5 5 Prashant Shenoy Lectures on Operating Systems university of Massachusetts 2008 Lecture 7 6 Maui Scheduler administration s guide ClusterResources chapter 1 7 David Jackson Quinn Snell Mark Clement Core Algorithms of the Maui Scheduler Brigham Young University lecture notes in compu
129. g labs at INHAM and have the following characteristics a room with shorter reverberation times ambient noise from a nearby fan of a laptop and test subjects with functional constraints or pathological voices In comparison with the setup of the first recording there are 2 differences e 4 hypercardioid PZMs are combined to a microphone array with a distance of 0 024 m between 2 adjacent microphones e An extra handheld microphone to record the noise Source 1 PZI 2 Speaker 3 Handheld microphone 4 Commercial microphone array 3 Recorders 6 Close talk T Lavalier 8 Notebook 9 Door Fig 4 Setup for the first set of recordings So 1 Notebook 2 Speaker 3 Recorders 8 on 6 acai for noise source 11 Noise source 6 Handheld for speaker 7 4 PZM 7 8 Screen with commands 9 4 s Close talk 10 Lavalier 11 Commercial microphone array Fig 5 Setup for the second set of recordings INHAM The recordings were decoded using a state of the art recognition system trained on normal non pathological voices recorded with a close talk microphone The results of the decoding are given in figure 6 where the Word Error Rate WER is defined as 12 S D I wepe 9 where S 1s the number of the replaced words D the number of substituted words I the number of inserted words and N the total number of words in the reference Figure 6 shows that for the first set of recordings the best results we
130. h By default the system records a sample of data every fifteen seconds Depending on hard disk space and test size this sample frequency can be increased or decreased Because the test endures only a few minutes we choose a sample frequency of just one second 3 Specifications The traditional workstation has an Intel Core2 CPU 2 13 GHz and 1 99 GB of RAM The installed operating system is Microsoft Windows XP Professional v 2002 with Service Pack 3 Its network card is a Broadcom NetXtreme Gigabit Ethernet card The Terminal Server has an Intel Xeon CPU 2 27 GHz and 3 GB of RAM The operating system is Microsoft Windows Server 2003 R2 Standard Edition with Service Pack 2 It has an Intel PRO 1000 MT network card C Discussion 1 Speed The most important factor is obviously the execution speed of the test When performing the actions on the traditional desktop it takes an average of 198 seconds to perform all predefined tasks On the Terminal Server on the other hand it only takes an average of 150 seconds This means that in this case the Terminal Server desktop environment is 48 seconds or approximately 24 faster than the regular desktop Saving almost a minute of time when performing a series of tasks that takes only about 3 5 minutes is a lot Working Set traditional workstation terminal server 700 600 500 4 i T VL 400 4 I 300 4 200 100 Memory in MB dd d
131. hange the parameters also a sensitivity analysis 1s possible to see which parameters can be altered an by how much Once the correct setup is found most of the work is done by Maui and of Torque The only problem that remained was the lack of Maui to work with a non uniform cluster This could be resolved by mak ng a script which will be run every day This would mean that Maui had to be restarted every day to let the change in the setup file go through V CONCLUSION For most purposes the combination of using Maui and Torque to manage a linux cluster works well The main issue is setting the different parameters in such a way that it resembles most the current real state of the cluster By using queuing theory to become the steady state parameters of a system we can predict more or less how a system will behave over time and adjust the gueuing system so to minimize the waiting time After setting up the Maui setup file the system works adeguate Nonetheless will the usage of some clusters change dramatically over time By adjusting the weight and factor of the different setup parameters the system can have a servo effect In order to overcome the problem of a non uniform cluster we had to use a extern script to supplement Maui so that the fairshare algorithm could cope with such a system To do this Maui has to be restarted every time an update of the fairshare target is being made This is of coarse not preferable but nonetheless need
132. he flow balance equations or conservation equations for our system Once the steady state 1s found other important performance measures can be derived One important factor is the utilization CPU utilization is one of the factors used to make an assessment of the quality of one system against an other An other parameter who is calculated is the average number of customers present in the queuing system L j L ED JT J 0 In this case we are interested in the expected number of jobs waiting in line in queue L We can find L by deriving it from the steady state Jt L gt G a L 1 J 0 From the data extracted from the output files from the lasts months it was possible to find the arrival rate A and the service rate 4 from the system According to Little s law 15 we can say that the average time a job spends in the system W with L W A Furthermore we can calculate the average time a job spends in line W by applying Little s law to L 5 in stead of L D Implementation When it comes to maximizing the CPU utilization this can best be done by setting the correct parameters for the scheduling software in this case Mau1 The main objective for the queuing system is to reduce the waiting time of the jobs Therefore W will be used We first divide the population of jobs into two subpopulations The first subpopulations contains the serial jobs the second subpopulations contains the parallel jo
133. ial way For example when ten users are all using the same application on the server the system does not need to physically load the executable of this application in the memory ten times It loads the executable just one time and the other sessions are referred to this executable Each session thinks that they have a copy of the executable in their own memory space which is obviously not true This way the operating system can save memory space and the overall memory usage 1S lower The Terminal Server has 3 GB of RAM see section II B3 We can calculate the maximum number of users the server could handle with the following eguation 600 x 1 350 lt 3000 1 lt 7 86 2 Only seven users can use the Terminal Server at one time when performing the same actions as simulated by the script This is a lot less than the desired number of forty If every user should perform in this way the memory of the server should be increased to 14 GB see the equation below 600 40 1 350 14000 3 The output from the Pages Output sec counter is shown in figure 6 This counter indicates how many times per second the system trims the working set of a process by writing some memory to the disk in order to make physical memory free for another process When the system is running low on physical memory when more users are connected to the Terminal Server the Pages Output sec counter will start to show high spikes Then the
134. iations are smaller The output from the Pages Output sec counter is shown in figure 2 and indicates how many times per second the system trims the working set of a process by writing some memory to the disk in order to make physical memory free for another process This is a waste of valuable processor time so the less the memory has to be written to the disk the better Windows doesn t pay much attention to the working set when physical memory is plentiful it doesn t trim the working set by writing unused pages to the hard disk In this case the output of the counter is very low When the physical memory utilization gets higher Windows will start to trim the working set The output from the Pages Output sec counter is much higher Bytes Total sec traditional workstation terminal server 1 20E 17 1 00E 17 8 00E 16 47 6 00E 16 47 4 00E 16 4 Bytes Total sec 2 00E 16 IK 0 00E 00 Ar DD MM i mm U MM dio mo mom oo iN NM sr nwo wm Ona dd cl st eed ed ed ed ed Time in s Fig 3 Output from the Network Interface gt Bytes Total sec counter Output Queue Length B traditional workstation W terminal server 3 ia 3 sj s 1 o tT mmm HAT MNT mmm Ann SERESREKBBSSLARRL ARTE SS Time in s Fig 4 Output from the Network Interface gt Output Queue Length counter We can see in figure 2 that there is plenty memory on the Terminal Server There is no need
135. icial result This paper tries to identify the reaction of the SaaS application of IOS International with different user loads and to assess if the SaaS application meets the expectations of it s clients Eventually we ll see that the response time is directly proportional to the user loads as long as there are no errors in the user loads We show also that the actual infrastructure meets the expected response time for an application load of 10 editors and 90 viewers Index Terms SaaS load testing IOS Mapper response time I INTRODUCTION IOS International nv a Belgium Company develops a software platform IOS to increase the productivity and the quality of risk management within an organization A new objective of IOS International is to make their software available on the Internet as Software as a Service SaaS This way the customer no longer has to buy the software but only concludes a contract for the services that he needs Software as a Service SaaS is one of the latest hypes in the mainstream world The quality of a SaaS application is assessed in terms of response time An inferior quality of a SaaS application can lead to frustrated users and will eventually create lost business opportunities On the other hand company expenditures on a SaaS infrastructure are linked with the application s expected traffic In an ideal infrastructure we want to spend just enough and not more allocating resources to get the most b
136. ife user 3 Fig 1 Load model of a virtual user Each rectangle in figure represents the requests that a user sends to the SaaS application IOS Mapper The SaaS application will respond to these requests and this will eventually lead to a visible window in the user s web browser This corresponds with the green ellipse in figure 1 IV USAGE SENSITIVITY OF IOS MAPPER A Single user In the first test it s the intention to find the minimum response times of the SaaS application One virtual user will pass through the complete load model which can be seen in figure 1 If the user wants to generate a report it takes a response time of 13 seconds This is the longest transaction as shown in figure 2 The generation of a report will be the most important reason for delays and crashes Response time ms 1 user 13321 1826 1477 Fig 2 Response time of 1 user Furthermore it s also important to know if the end to end response times are influenced by the pc or the bandwidth of the Internet of the users For this test we used a AMD Athlon 64 X2 Dual Core Processor 4200 GHZ with 2 00GB RAM as pe and a AMD Turion 64 Mobile 1 99GHZ with 1 00GB RAM as laptop The laptop is significantly slower than the pc We will also use the laptop on 2 different locations with its own Internet The first location has Internet with a bandwidth of 4Mbit and the second location has Internet with a bandwidth of 12Mbi
137. incoming faxes from the PSTN the router will act as an on ramp gateway Here the fax will be converted to a Tagged Image File Format TIFF file which will be attached to an e mail and forwarded to the end user For outgoing faxes the router will act as an off ramp gateway where an e mail with a TIFF attachment will be converted to a traditional fax format and delivered to a standard fax machine The converting is done with the ITU T T 37 standard The choice that was made for the Warande was to use Fax pass through with up speed This choice was made because the equipment was not suited for the fax store and forward option On the other side the fax relay method was not chosen because the available bandwidth was not an issue The choice of using up speed was because almost the whole network uses codec G 729 which is incompatible for using the first pass through method C Configuring Digital Ports Digital circuits are used when interconnecting the VoIP network to the PSTN or to a Private Branch Exchange PBX The advantage of using digital circuits is the economies of scale made possible by transporting multiple conversations over a single circuit Since the Provincie Antwerpen has a contract with Belgacom as their telecom operator they use the Integrated Services Digital Network ISDN network for their calling services The equipment used supports the ISDN Basic Rate Interface BRI and ISDN Primary Rate Interface PRI Both m
138. iness network Data transfer Figure 2 1 Flowchart of the followed way to migrate the ActiTime application B Analysis of currently used version of ActiTime and data backup Flanders DRIVE is using the ActiTime 1 5 version installed with an automatic setup In order to collect the data from the old version a way to extract the specific user data from the database must be found It is important to migrate this user data because otherwise all the time tracking information that was entered before would be lost The automatic setup allows the administrator to specify which database to use for the collection of the user data The ActiTime application can run with two database programs mysql and Microsoft access When ActiTime was installed for the first time Flanders DRIVE chose the mysql option So to extract the data from the old application a proper way to extract this data must be found The name of the database could be derived from the ActiTime support files The database is called ActiTime The exportation from the user data is a kind of backup that is made To back up the database the mysqldump 3 command could be used mysqldump u lt username gt p lt password gt ActiTime gt actiTime_data sql A short explanation of what to fill in e lt username gt fill out the username that is used to set up the mysql database e lt password gt fill out the password used for the user who created the databa
139. ingna X 2009 Load Testing for web applications First International Conference on Information Science and Engineering 2954 2957 2 Nah F 2004 A study on tolerable waiting time how long are Web users willing to wait Behaviour and Information Technology 2303 153 163 3 Grundy J Hosking J Li L Liu N Performance Engineering of Service Compositions PowerPoint presentation The University of Auckland Founded at http conferenze dei polimi it SOSE06 presentations Hosking pdf Fixed Size Least Squares Support Vector Machines Study and validation of a C implementation S Vandeputte P Karsmakers Abstract We propose an implementation in C of the Fixed Size Least Squares Support Vector Machines FS LSSVM for Large Data Sets algorithm originally developed in MATLAB An algorithm in MATLAB is known to be suboptimal with respect to memory management and computational performance These limitations are the main motivation for a new implementation in another programming language First the theory of Support Vector Machines is shortly reviewed in order to explain the Fixed Size Least Squares variant Next the mathematical core of the algorithm which is solving a linear system is zoomed into As a consequence we explore a set of LAPACK implementations for solving a set of linear equations and compare in terms of memory usage and computational complexity Based on these results the Intel MKL library is selecte
140. ion Foundation is a new infrastructure for communication and is an extention of the expanded set of existing mechanisms such as Web services Windows Communication Foundation is a new infrastructure for communication and is an extention of the expanded set of existing mechanisms such as Web services WCF makes it possible for developers using a simple programming model to build safe reliable and configurable applications This means that WCF provides a robust and reliable communication between client and server Not only the connection with the database can create a qualitybased application It is also necessary that a good structure for the code is used For this research the Model View Controller designpattern is used This pattern splits the design of complex applications into three main sections each with their own responsibilities Model A model manages one or more data elements and includes the domain logic When a data element in the model changes it notifies its associated views so they can refresh View A view renders the model into a form that is suitable for interaction what typically results in a user interface element Controller A controller receives input for the database through WCF and initiates a response by making calls to the model B er Controller Input Presentation Figure 2 Model View Controller model 3 The advantages of using a designpattern is that the readability and reusability o
141. is by way of a fairshare target Each target is specified as a percentage value where each value is interpreted as a percent of delivered utilization The use of delivered utilization as the target basis as opposed to using percent of configured or available resources allows the fairshare system to transparently take into account factors such scheduling inefficiencies system maintenance job failures etc The fairshare algorithm is composed of several parts These parts handle tasks including the updating of historical fairshare usage information managing fairshare windows determining effective fairshare usage and determining the impact of a job s various effective fairshare usage components Unfortunately Maui does not have the ability to distinct between different cpu s If we would relay completely on the fairshare capabilities of Maui then users who had used n Pentium 4 3 0 gHz processors would be penalized by the fairshare system as much as someone who had used n AMD Opteron processors This would have a pernicious influence on the cluster usage because user would tend to wait for the faster nodes instead of running on slower nodes To overcome this shortcoming we will be forced to change this externally C Resource manager Resources managers provide the low level functionality to start hold cancel and monitor jobs Without these capabilities a scheduler alone can not control jobs In addition to Maui the Fermi2 cluster uses To
142. ited SNR gain the SNR slowly increases with the number of microphones e Great number of microphones To obtain a good SNR we have to use a lot of microphones This leads to an inefficient array Non uniform spacing of the microphones might relax this issue 5 In the GSC the sum and delay beamformer is useful to obtain a reference signal which is necessary for the adaptive filter in the GSC B Blocking Matrix The goal of the blocking matrix is to get a reference of the noise at the output This is obtained by applying a spatial O in the steering direction In this manner the speech is suppressed and we only get the noise C Adaptive filter for SNR gain The third part of the GSC is an adaptive filter The filter is used to estimate the acoustic path of the noise So at the output of the filter we get an estimation of the noise The general scheme of an adaptive filter can be seen in figure 3 Here x n y n and s n are respectively the noise a filtered version of the noise and the speech Fig 2 Sum and delay beamformer with 3 microphones M 3 din x n Digital an filter Adaptive algorithm Fig 3 Adaptive noise cancellation 7 X n is obtained as x n passed the transfer function P z Combining x n and s n gives the desired signal d n This signal is composed out of speech and noise The transfer function presents the acoustic path from the noise source to the microphone who records the spe
143. ively Of course it is also possible to analyze this data immediately after it is filtered This way writing data to disk and analyzing data can be done at the same time Another advantage of data compression is the reduction of read times afterwards which speeds up offline analysis simply because there is less data to read 1 II DATA REPRESENTATION Before we can move on to calculation of radar parameters or data compression it is important to take a look at how data is represented We will therefore take a look at the representation of the two used data formats primary radar raw video and secondary radar digital data A Primary Radar Raw Video Primary radar raw video consists of a byte stream where each two bytes 16 bits represent one sample The used data format is represented in Table 1 TABLE 1 Primary Radar Raw Video Data Format Sample 1 Sample 2 Sample 3 Sample 4 Analog 1 12 Analog 2 12 Analog 1 12 Analog 2 12 ARP 1 I 1 0 1 I 1 ACP 1 0 1 I 1 I 1 PPS 1 Mode S 1 0 1 Mode S 1 Trigger 1 Trigger 1 Trigger 1 Trigger 1 This 16bit data is sampled at 16MHz using a RIM device Radar Interface Module Since I Q data is interleaved this is SMSPS 2 Analog 1 and Analog 2 represent 12bit I O data The other 4 bits are digital bits where trigger ACP and ARP together with I Q data are important The trigger bit is set when a new interrogation has started when a new pulse is transmitted
144. ks we need a vast amount of equipment Using packet telephony we can multiplex voice traffic alongside data traffic which results in savings on equipment and operations costs Flexibility An IP network will allow more flexibility in the pallet of products that an organization can offer their customers Customers can be segmented which helps to provide different applications and rates depending on traffic volume needs Advanced features o Advanced call routing e g Least cost routing and time of day routing can be used to select the optimal route for each call o Unified messaging This enables the user to do different tasks all in one single user interface e g read e mail listen to voice mail view fax messages o Long distance toll bypass Using a VolP network we can circumvent the higher fees that need to be paid when making a trans border call o Security Administrators can ensure that IP conversations are secure in an IP network Encryption of sensitive signaling header fields and massage bodies protect packets in case of unauthorized packet interception o Customer relationships A helpdesk can provide customer support through the use of different mediums such as telephone chat e mail Hereby the customer satisfaction will increase In the traditional PSTN telephony network it is clear to an end user which elements are required to complete a call When we want to do a migration to VoIP we need to be aware an
145. l Magic Quadrant for WAN Optimization Controllers Gartner 30 June 2009 Available at http mediaproducts gartner com reprints riverbed 165875 html Ipanema Technologies Smart Path Selection Combining Multiple Networks Into One Ipanema Technologies 8 July 2009 10 11 12 13 14 Ipanema Technologies Solution Overview Guarantee Business Application Performance Across The WAN Ipanema Technologies 25 May 2009 Riverbed Technology MAPI Transparent Pre Population Riverbed Technology Riverbed Technology RiOS Riverbed Technology 2009 Available at http www riverbed com products technology A Bednarz What makes a WAN optimization controller Network World 1 August 2008 Available at http www networkworld com newsletters accel 2008 0 107 netop 1 html Line of sight calculation for primitive polygon mesh volumes using ray casting for radiation calculation K Henrard R Nijs J De Boeck IBW K H Kempen B 2440 Geel Belgium SCK CEN B 2400 Mol Belgium karel henrardtedu khk be Abstract A line of sight in this context is a straight line or ray between two fixed points in a rendered 3D world populated with primitive volumes ranging from spheres and boxes to clipped hollow tori These volumes are used as building blocks to recreate real world infrastructure containing one or more radioactive sources To find the radioactive dose in a fixed point c
146. lems Finally it s possible to insist on a Proof of Concept POC to see how the WOC performs in the company network before committing to any purchase Riverbed Technology delivers WOC capabilities through their Steelhead appliances and the Steelhead Mobile client software It has a leading vision a great product reputation and some features that Ipanema doesn t have Ipanema Technologies delivers WOC capabilities through their Plengine appliances It delivers WAN optimization as a managed service These WOC solutions are described and compared in the following chapters of this paper II RIVERBED TECHNOLOGY A Riverbed Optimization System The Riverbed Optimization System or RiOS is the software that runs on the Steelhead appliances and the Steelhead Mobile client software RiOS helps organizations to dramatically simplify accelerate and consolidate their IT infrastructure RiOS provides the following benefits to enterprises e More user productivity e Consolidated IT infrastructure e Reduced bandwidth utilization e Enhanced backup recovery and replication e Improved data security e Secure application acceleration RiOS consists of four major groups e Data Streamlining e Transport Streamlining e Application Streamlining e Management Streamlining B Data Streamlining Data streamlining or Scalable Data Referencing SDR can reduce the WAN bandwidth utilization by 60 to 95 and it can eliminate redundant data transf
147. lication hyper V is used to set up the new environment and a little problem with the antivirus real time scan came up Step by step different problems were solved with a successful migration of ActiTime as a result I INTRODUCTION landers DRIVE is the Flemish competence pool for the vehicle industry The company was founded in 1996 When Flanders DRIVE moved to Lommel in 2004 they decided to buy an IT infrastructure that met the requirements at the new office in Lommel At the end of the year 2008 Flanders DRIVE decided to renew their IT infrastructure To renew an IT infrastructure it is important to correctly transfer all the components of the old infrastructure to the new one The whole transfer of the IT infrastructure and the implementation of new components can be found in my master thesis Analyse van een nieuwe IT infrastructuur When an infrastructure has to be migrated software with specific user data had to be transferred too This papers handles the migration process of one of these software applications This software application is a time tracking software tool called ActiTime ActiTime is an important tool to track time of employees Flanders DRIVE is using this software tool to create a view on how much time is spent on a customer task or a customer project which involves several employees The client billing information is partially determined from this software tool Employees who are using this software t
148. ligent optimization and intelligent acceleration Both WOC solutions have similar features but Riverbed has some additional features that Ipanema doesn t have This paper describes and compares both WOC solutions I INTRODUCTION AND RELATED WORK A WOC is a customer premises equipment CPE that is typically connected to the LAN side of WAN routers These devices are deployed symmetrically on either end of a WAN link in data centers and remote locations to improve the application response times The WOC technologies use protocol optimization techniques to prevent network latency They also use compression or caching to reduce data travelling over the WAN and they prioritize traffic streams according to business needs Therefore WOCs can also help organizations to avoid costly bandwidth upgrades Telindus offers WOC solutions from Riverbed Technology to their customers and Belgacom offers WOC solutions from Ipanema Technologies to their customers Because Telindus now belongs to Belgacom it is useful to know which the appropriate solution is for a certain customer or network This vendor selection can be difficult because vendors offer different combinations of features to distinguish themselves Therefore it is important to understand the applications and services and their protocols that are running on the network before choosing a vendor It is also useful to conduct a detailed analysis of the network traffic to identify specific prob
149. listed in the graph because of the scale When we embed the price in the overall comparison the differences between network load DISK CPU and memory will not be visible The price is already mentioned in the benchmark section Summarization 100 000 90 000 80 000 70 000 60 000 50 000 40 000 30 000 20 000 10 000 MonitorMagic Spiceworks tembria server monitor ManageEngine SolarWinds ActiveXperts PA server monitor WebWatchBot lpsentry ServerAssist m total Mb Etransfers sec m memory Fig 7 Summarization results When we bring all this together as well as taking a look at the ease of use MonitorMagic 1s the most suitable tool for Rescotec This brings us to testing it in a working network which gives approximately the same results as mentioned before We can conclude that we found the solution for the downtime of servers in the company without freguently checking the parameters ACKNOWLEDGMENT First of all I would like to thank Rescotec for giving all the necessary materlals for testing and doing the research Also a special thanks to Joan De Boeck for helping me with benchmark problems and correcting this paper REFERENCES 1 Alwin Brokmann Monitoring Systems and Services Computing in High Energy and Nuclear Physics La Jolla California March 2003 2 MICROSOFT CORPORATION Windows 2000 professional resource kit http microsoft com windows2000 library resources reskit 2000 3
150. ll an application like ActiTime but several components can be studied We have an Microsoft Exchange server it isn t recommended to install the application on this server Because this server already has a high load and there is a web server installed to access employee mailboxes through a web interface We use the Apache Tomcat Server and this server can disrupt the Microsoft IIS server installed on this machine Another possibility is a new server where Active Directory DNS DHCP the citrix licensing the backup exec etc is running Because we prefer to keep the Active Directory server separated from roles who need to set up a web server this server isn t the best option There is also a server where the Citrix remote access application is installed We don t choose this machine because citrix is also using the IIS web server to connect the citrix application with the internet It s not a good solution to install two web servers of two different vendors on the same machine Then there is another option to install ActiTime on the Microsoft SharePoint server Since this server is also using the Microsoft IIS server SharePoint is a web based environment we can t install ActiTime on this server either Our last option is to virtualize a computer where we can install our application to We find out that the Active Directory server is the server who is carrying the less load So a virtualization program can be installed on this server The
151. ll be silence In a VoIP network whether it is a conversation or silence it is packetized VAD can suppress packets containing silence With interleaving data traffic with VoIP conversations the VoIP gateways will use network bandwidth more efficiently A silence in a call can be mistaken for being disconnected This is also solved with VAD since it provides CNG CNG will make the call appear normally connected to both parties by generating white noise locally Voice sample size is a variable that can affect the total bandwidth used To reduce the total bandwidth needed we must encapsulate more samples per Protocol Data Unit PDU is the control information that is added at each layer of the OSI model when encapsulation occurs But larger PDU s will risk causing variable delay and several gaps in communication That is why we use the following formula to determine the number of encapsulated bytes in a PDU based on the codec bandwidth and the sample size 2 Bytes_per_sample Sample_Size codec_Bandwidth 8 Meaning if we would use the G 729 codec and knowing that the standard for sample size is 20 bytes and the bandwidth for G 729 is 8kHz this would result in Bytes_per_sample 0 020 8000 8 20 Another characteristic that influences the bandwidth is the layer 2 protocol used to transport VoIP Depending on the choice of the protocol it is possible that the overhead will grow substantially When the overhead is higher
152. ls of these languages Testing some network operations would be more effective Later on you will find a custom made benchmark III COMMUNICATION METHODS A General info Network programming requires interaction between hosts and network devices such as routers switches and firewalls So let s have a look at several communication methods Serial communication is mostly used to make a connection through the console port The greatest advantage is the fact that you are able to establish interaction without the need of any switch configuration This technique becomes indispensable when neither the ip address vty ports console or aux ports are configured The telnet protocol is built upon three main ideas First the concept of a Network Virtual Terminal second the principle of negotiated options and third a symmetric view of terminals and processes 5 If multiple network devices are connected to eachother a client is able to gain remote access to each device which is telnet ready All information send by telnet is send in plain text In this situation security is not an important issue SNMP is a very interesting protcol to get specific info of a device With one single command it is possible to retreive the status of an interface the amount of retreived TCP segments etc Three different versions of SNMP are possible SNMPv1 SNMPv2 SNMP V1 and V2 are very close They both use community strings to authenticate the
153. m IT responsible Flanders DRIVE B 3920 Lommel Belgium Email devosmaarten hotmail com ward vleegen O flandersdrive be tom croonenborghs edu khk be Abstract When the concept of time tracking was introduced for the first time it was used to simply determine the payroll of an employee The amount of time that was spent on a task could be converted to a reasonable payment for an employee More useful time spent on a company task translated itself into a higher payment These days time tracking has evolved to a great and handful tool to derivate several important things like how much time is spent on a project how an employee divides its time onto several tasks etc Time tracking can determine customer billing information by calculating how much time was spent on a customer project Flanders DRIVE uses a free software tool to track time of several employees 1 This software tool is called ActiTime The ActiTime application is a free application to register time dedicated to specific tasks Flanders DRIVE decided to introduce a new IT infrastructure to meet its business requirements With the migration from the old infrastructure to the new one ActiTime also needed to be migrated A few problems came up in the migration process such as e g how to convert the current database which web server application would be best to use which server is best suited to install the application etc In the migration process of the ActiTime app
154. matically monitored by dedicated agents checking the availability of all running services On the other hand when networks become large the additional network overhead caused by these tools cannot be ignored The research in this paper aims to optimize the downtime of services without using too much of the network bandwidth Microsoft Exchange BlackBerry MDS Connection Service System Attendant Table 1 Testing parameters Some examples of tools that didn t make the benchmark are Internet server monitor Intellipool IsItUp IPhost Serversalive Deksi network monitor Javvin Easy Network A Parameters that are necessary in the tool Service Monitor SCOM this because of the limitations or the overall cost II DESIGN REQUIREMENTS The following parameters must be met for a tool before it is put in benchmark All the listed items are services or resources The tools that fulfill all needs are listed in random order and that a system admin must check frequently to prevent failures will be put in benchmark for comparison and unwanted downtime is Aeon Some extra information for people who have no experience aaa with BlackBerry The besadmin is the admin to control ee NET BlackBerry services A list of tools has been checked for the 4 MonitorMagic proper specification for example Nagios 1 did not have the 5 PA Server Monitor ability to scan with another admin 6 ServerAssist 7 SolarWinds 8 Spice
155. mbination are available In order to be able to solve the linear system as fast as possible it is worth the investigation to find out the best performing implementation Four known LAPACK and BLAS implementations were tested Mathworks MATLAB R2008b MATLAB makes use of a LAPACK implementation for Intel CPUs the Intel Math Kernel Library v7 0 The test may reveal the influence of MATLAB as LAPCK wrapper Reference LAPACK v3 2 1 libraries which are reference implementations of the BLAS 9 and LAPACK 10 standard These are not optimized and not multi threaded so a bad performance is to be expected Intel Math Kernel Library MKL implementation of Intel which of course exploits the most out of Intel processors Version 10 2 4 is used GotoBlas2 a BLAS library completely tuned at compile time for best performance on the CPU it is compiled on Of course there are more LAPACK implementations available than the ones we selected for testing For some reason they were left out like e g ACML is the AMD implementation while only test on Intel processors IV TEST We developed a test application to solve the equation Ax B in C using LAPACK functions dgesv for double precision and sgesv for single precision input data in MATLAB using the operator or mldivide function During the lifetime of a software application dynamic memory which is used to store the matrices A and B can get fragmented To make sure fragme
156. me per workstation When having multiple sessions on a single workstation only the active session the session at the front of the screen would run the script correctly The session of which the window is minimized or behind another RDC session window would not execute the script correctly Therefore because we had four machines at our disposal we could only run four RDC sessions which could run the script correctly at the same time D Discussion 1 Memory Figure 5 shows the output from the Working Set counter which is the total of all working sets of all processes on the system in bytes This number does not include the base memory of the system The first thing we can conclude is that the execution time does not increase significantly when adding more users to the server around 2 seconds per extra user Next we can look at the memory usage One user running the simulation script uses a maximum of around 600 MB We see that for each extra user who runs the script the memory usage raises with approximately 350 MB For example when three users are running the script the Working Set counter has a maximum of 1300 MB 600 MB for one user and 2 times 350 MB for the extra two users Normally we would expect the memory used when three users are running the script to be 1800 MB 600 MB times 3 when in fact it turns out to be only 1300 MB The reason for this is that a Windows Server 2003 Terminal Server uses the memory in a spec
157. merdries Oost 27 B 2250 Olen Belgium stijn_delarbre hotmail com niels vanhoef intersoft electronics com guy geeraerts khk be Abstract Research on radar performance is becoming more and more essential It is important to assess radar performance based on calculated parameters and use these parameters to optimize or improve radar performance in certain situations We will discuss real time and offline radar parameter calculations in LabVIEW7 for future performance analysis based on primary radar raw video and secondary radar digital data Secondly real time compression of raw video data coming from primary radar using digital data from secondary radar in C and LabVIEW7 will be discussed Raw video data compression may have its benefits The smaller the data the longer the recordings can be to fit on the same disk It will turn out that data compression will speed up offline analysis and that disks will be used less intensively and less memory is needed It will also become clear how certain parameters are implemented that lie on the basis of future performance analysis I INTRODUCTION At present radar systems are meant to run 24 7 and faults aren t always immediately detected Most radar systems undergo maintenance on a monthly or tri monthly basis and have to function at a reasonable performance all the time Therefore it is important to calculate radar parameters to assess radar performance These parameters include
158. ming from secondary radar This will not only improve disk usage but it also speeds up the offline analyzing process Having shown the importance of data compression we will give an overview of certain decisions taken during the process of writing the filtering program These decisions have an influence on program complexity disk memory usage and determine the complexity of programs to read data afterwards Buffering Buffering is the first important decision Since secondary radar target information will not always reach the computer system at the same time the primary radar raw video of the same target does it is important to buffer raw video for a certain time Note that both primary and secondary radar are connected to the same PC laptop The used buffer has to be large enough so that no data will be lost but the buffer has to be small enough so that not all physical memory will be used for buffering We have chosen the size of the buffer to fit 1 full scan of 360 We have chosen this size because it is easy to work with and because simulations have shown that we won t lose any important data The used buffer uses the FIFO algorithm This means that the oldest data will be removed first if necessary when new data enters the buffer Threading We had to take a decision concerning threading If we would work with a single thread we would have to check if there is a secondary radar target waiting to be filtered every time we ru
159. n be seen from Table 1 where the SNR gain for periodic noise is significantly lower than for random noise However a GSC doesn t work well with random noise From Table 3 we see an additional gain of maximum 0 28 dB This 1s inferior compared to the results in Table 2 Here we reach additional gains of 30 dB and more for larger filter lengths Based on these results we can conclude that a GSC works well with periodic noise Furthermore the number of microphones plays also a role for the gain For the sum and delay beamformer the results are clear The SNR gain increases with the number of microphones Certainly for random noise but this effect can t be seen for the GSC Moreover there is no clear dependency between the SNR gain and the number of microphones Finally the distance between 2 microphones is observed Here we see no clear relation for the GSC but periodic and random noise has an influence on the SNR gain of the sum and delay beamformer Where the SNR gain increases for periodic noise a decrease for the SNR gain is observed for random noise V CONCLUSION In this paper we examined the influence of the position of a microphone on the speech recognition We showed that a microphone near the speaker gives the best performance but the speaker must have an alternative when there s no possibility to use a close talk microphone Due to the greater distance between speaker and microphone all the investigated microphones gave pr
160. n calculate the earliest time the necessary resources will be available to start that job The Backfill algorithm makes use of this knowledge by checking whether there is a low priority job who can be scheduled without delaying the high priority job On a typical system backfilling will increase system utilization by around 20 and improve average job turnaround time by even more 8 Backfill Windows Nodes Time BacKkfillable Nodes Figure 2 backfill example In this case job A is a low priority job job B and job C are high priority jobs Even tough job C has a higher priority job A can start running because he does not interfere with job C starting time Another important feature is fairshare There are a number of interpretations of the term fairshare as applied to batchsystems In general however they each involve a mechanism which controls the distribution of delivered resources across various job attribute based dimensions They do this by tracking a utilization metric over time and using this historical data to adjust scheduling behavior so as to maintain resource usage within configured fairshare constraints 9 Maui offers flexibility in configuring fairshare in areas including the tracked utilization metric the utilization to job correlation attributes the historical period and the method of fairshare enforcement Fairshare target usage can be specified on a per user group account QOS or class bas
161. n other environments the administrators choose to configure the client workstations to access all of their applications via a Terminal Server This has the advantage that management is centralized which makes it easier to do These environments are called Server Based Computing The Terminal Server environment used for perfor mance and capacity testing as described in this paper are Server Based Computing environments The Terminal Server is accessed via an RDC and the Terminal Server delivers a full desktop experience to the client The Windows Server 2003 environment uses a specially modified kernel which allows many users to connect to the server simultaneously Each user is running its own unique virtual desktop and is not influenced by actions from other users A single server can support tens or even hundreds of users The number of users Rudi Swennen TBP Electronics Geel rswennen tbp eu Vic Van Roie K H Kempen Geel vic van roie khk be a Terminal Server can support depends on which ap plications they use and of course it depends strongly on the server hardware and the network configuration Capacity testing determines this number of users and also possible bottlenecks in the environment By up grading or changing server or network hardware these bottlenecks can be lifted and the server is able to support more users simultaneously This research is done for a company which has eighty Terminal Server User CALs Client A
162. n through the raw video coming from the primary radar When using 2 threads reading raw video will become independent from processing targets thus execution becomes asynchronous Therefore when the execution of one of both threads lags the other thread will keep executing in the correct way For this reason we have chosen to use 2 threads Note that using 2 threads also makes our program easier to debug during implementation and easier to understand One thread maintains the buffer that contains the primary radar raw video and creates a list of what is inside the buffer A second thread checks if there are targets waiting for filtering and if there is a waiting target it filters this target out of the buffered raw video Of course both threads require some kind of synchronization so that no faulty data is filtered 11 In other words the second thread has to run fast enough so that no data is lost or wrong data is filtered Simulations have confirmed that without any synchronization mechanism data is filtered in the right way Writing targets to disk How we are going to write a target to disk is the last very Important decision It determines the complexity of the program it has an influence on memory usage and it determines how we are going to read data afterwards We could create one index file in which every target s header is located and one data file or we could create a header for each target and attach the target s data to hi
163. nables the generation of a Power Management Event upon reception of a Magic Packet 4 1 2 3 WUS Wake Up Status register This register is used to record statistics about all wakeup packets received Useful for testing 4 2 PCI Power Management The PCI Power Management specification 10 provides different power states for PCI busses and PCI functions devices Before transitioning to the G2 55 Soft Off state the operating system can request auxiliary power for devices that require it This is done by placing the device itself into a low power state D3 is the lowest power state with maximal savings but enough to pro vide auxiliary power for the network device Every PCI device has a Power Management Register block that contains a Power Management Capabilities PMC register and Power Management Control Status Register PMCSR The most important register is the PMCSR It contains two important fields 4 2 0 4 PowerState This field is used to change power state D3 state provides maximal savings with auxiliary power to provide Remote Wake Up capabilities 4 2 0 5 PME En Enables wake up using Power Management Events This is the same bit used in the WUC register from the Intel network device 4 2 1 Wake Event Generation Wake events can be generated using Power Management Events The PME signal is connected to pin 19 of a standard PCI connector Software can assert this signal to generate a PME That software could be the wak
164. nalyze the output data after the calculation This means that the cluster usage of the users depend highly on the state of their project which means that the overall cluster usage is very dynamic This makes it necessary to adjust the scheduling rules very The solution to this problem should resemble a servo system where the input is continuously changing according to the output The output in this case would be the data received from the scheduling software and resource manager We started our work by dividing the task into three subsystems First there is the queuing system All jobs will be put in a particular queue By setting the parameters for each queue there can be made a first sifting Then each of the queuing systems will be associated with a particular set of resources The management of those sets of resources together with the decision making when it comes to which job has priority over the other job will be the second subsystem The last subsystem will be responsible for adjusting the setup of the first two system so that the overall system will stay as close to the optimal setup as possible This will be done by changing the targets set by the scheduler IL PRELIMINARY STUDY A Fermi cluster First of all it s important to have some knowledge of the cluster and the system which is managing the cluster The Fermi2 cluster has grown over the years by adding new nodes This results in a cluster who consists momentarily out of thr
165. nd below Here we can see a couple of sites that are managed by the program teway Lar 15 Media Gateway Sn ay Large De Nekker No Pari Ze verdiep No Par ge Lozanna 2 vall Vrieselhof No Pari E Configuration OXE Provincietoren 27 0 oveliers verdiep1 No Pari Q I y La o 8 Sr 0 o 419 al 4 YES Shelf Board Address 27 0 Ci s tak No N64 4 YES Interface Type GD o 21 ay La P No Par ate GREEN gt 22 bis No Par Board Ethemet Address gt Cj 23 y Sr ari Board IP Address STARTED o c 24 Media Gateway Large Lozana GV h323 No Par NetMask IP de ee o gt 25 Media Gateway Small Archief Lange Nieuws Na Pari o 26 Media Gateway Small Hoge Mouw No Pari Default Gateway IP Address meme 4 27 Media Gateway Small Warande No Pari Interworking with Gatekeeper NO A Be d G atekeeper Id G 0 GD Idle Enabled Main Master p gt Atm Port IP Quality of service de gt Ethernet Access Numb of sig channels IP Phones Inter Act Link p 5 L A j dee t Numb of sig channels inter ACT 0 q Signaling Link kur IP Domain Number Kl 7 ae et Parameter a E164 Number List Index ae amp ds Voice Guide Ind Gateway H323 name o gt L 1 MG MIX BRA 4 US LI 4 Active Enabled Mai Cryptographic box address D VoIP gateways and gateway control protocols 3 To provide voice communication over an IP network dynamic Real time Transport Protocol RTP sessions are created and formed by one of many call control procedur
166. nde is to use a certain number of WAPs placed at strategic places These spots where calculated through experience and making a small site survey to measure and comprehend the RF behavior of the site With the choice made the next thing on the to do list was to configure the network Here we needed to watch out for some factors that have a negative influence on the design such as echo jitter delay Also a measurement of the total bandwidth that was needed for our voice traffic to travel on was calculated When the preparations were made there were two different things we had to do Firstly there was the configuration of analog ports These ports were used to connect fax machines into the network We discussed the three possibilities that could be used for enabling the faxing mechanism The fax pass through method was the one selected Secondly the configuration of digital ports was completed These port interfaces are mostly used for making connections to the PSTN network or to a PBX The configuration of the digital ports was done using an ISDN PRI and ISDN BRI interface The PRI was used for internal purposes and BRI for connecting to the outside world Finally we searched for a suitable gateway protocol These protocols will dynamically create and facilitate RTP sessions to provide voice communication of an IP network Here were three major protocols available H 323 MGCP and SIP We easily excluded MGCP from the list bei
167. ne RDC session running a script with the results from multiple RDC sessions running the same script simultaneously Most likely in the company environment with the current server hard ware memory is the bottleneck when it comes to server capacity The testing indicates that the Terminal Server could support around 7 users in the most extreme conditions of our script The goal for the company is to support forty users per Terminal Server so upgrading server memory is inevitable Also the processors need to be upgraded IV CONCLUSION There are differences between using a traditional workstation and using a virtual desktop environment on a Terminal Server which can be accessed by setting up an RDC session between a client machine and the Terminal Server itself By testing the performance we can examine these differences in the field of working speed load times and overall easiness of use To com pare these two solutions we needed to collect the data First we manually performed a series of user actions on a traditional workstation and logged certain counters Afterwards we manually performed the same series of actions on a virtual desktop on the Terminal Server By comparing the results we have learned that first of all the Terminal Server environment executes the same series of actions 24 faster than de traditional workstation We also concluded that memory usage and network us age is more efficient in a Terminal Server environmen
168. nel The choice of a particular algorithm used by the kernel may favor one class of processes over another 4 This is also the case when scheduling jobs in a cluster In choosing which algorithm to use in a particular situation we must consider the properties of the various algorithms Many criteria have been suggested for comparing CPU scheduling algorithms Which characteristics are used for comparison can make a substantial difference in which algorithm is judged to be best The criteria include the following e CPU utilization to keep the cpu as busy as possible e Throughput number of processes that complete per time unit e Turnaround total time between submission of a process and its completion e Waiting time amount of a process has been waiting in the ready queue e Response time amount of time it takes from when a request has been submitted until the first response is produced e Fairness Equal cpu time for each trhead Once we have found some parameters to assess different systems we must try to adjust the system to the real life situation within the company In our case that situation is very dynamic When working on scientific research i e nuclear energy and the effects on the environment first there is a preparatory period where the user has to prepare his input file During this time there is no need for cluster time Then a model is constructed This involves a lot of processor time Afterwards they have to a
169. neous absolute power output of the converter To estimate the correlation between the sensor and the power output of the converter Figure 2 shows the plot of the signals Comparison Cellsol 200 blue Sunnyboy 1700 red 1000 900 800 700 600 500 400 power per surface P W m 300 200 100 i 0 11 12 13 14 15 16 17 18 19 20 absolute time T hour Figure 2 The correlation coefficient between these 2 signals was calculated to be 97 3 The power output of the PV setup is on average 14 1 of what the sensor indicates This is explained by the fact that the sensor indicates the power of the incoming solar irradiation and that the Sunnyboy converter the outgoing electrical power Calculating that the sensor indicates around 25 to much according to the spectrometer this would give an efficiency of 11 3 This seems acceptable knowing the max efficiency given by the manufacturer of 12 4 and knowing there is still also a loss in the converter Out of these results we can deduce that the sensor is linear and that the correlation of the sensor output and the output of the PV setup is high HI MICROCONTROLLER CIRCUIT Now to increase the sample rate and sensitivity of our measurements we introduced a microcontroller based circuit The intention of this circuit is to sample the sensor output with a much larger sample rate using the ADC of the microcontroller The microcontroller will add every input value to its buffer When
170. ng a client server protocol Afterwards SIP was also excluded through the different implementations of equipment REFERENCES 1 Staf Vermeulen Course IP telephony Master ICT 2 Kevin Wallace Authorised Self Study Guide Cisco Voice over IP CVOICE Third Edition Cisco Press First Print 2008 125 183 185 244 3 Denise Donohue David Mallory Ken Salhoff Cisco IP communications Voice Gateways and gatekeepers Cisco Press Second printing 2007 25 52 53 78 79 114 4 http www cisco com 5 Staf Vermeulen Course CCNA 4 Accessing the WAN Master ICT 6 Patrick Colleman Course Datacommunicatie Master ICT Usage sensitivity of the SaaS application of IOS International Luc Van Roey Piet Boes Joan De Boeck IBT K H Kempen Associatie KILeuven B 2440 Geel Belgium IOS International Wetenschapspark 5 B 3590 Diepenbeek Belgium luc_van_roey hotmail com piet boes iosint be joan deboeck khk be Abstract Software as a service SaaS is one of the latest hypes in the mainstream world The quality of a SaaS application is assessed in terms of response time An inferior quality of a SaaS application can lead to frustrated users and will eventually create lost business opportunities On the other hand company expenditures for a SaaS infrastructure are linked with the application s expected traffic In an ideal infrastructure we want to spend just enough and not more allocating resources to get the most benef
171. nges in the original algorithm Figure 1 Time results of LAPACK can than easily be transferred to the new code tijd s 3 In Figure 1 is there an immediate result visible the performance of ther reference Lapack is rather bad actually the curve is O N We can also see that Matlab We are going to compare the different implementations with cannot handle more dan 8300 sized matrices due to lack regards to time of memory or good memory management inside the VII IMPLEMENTATIONRESULTS We picked randomly some datasets from 11 and used them as inputdata for the two algoritms Test were performed on the Pentium D 940 MATLA FSLSSVM s testnaam inputdata B s testdata 1 85 0 55 mpg 7 57 1 83 australian 20 27 5 97 abalone 202 60 45 56 mushroom S 1575 14 344 88 Figure 3 MATLAB FS_LSSVM t o v FSLSSVM MATLAB FSLSSV M Figure 4 MATLAB FS_LSSVM t o v FSLSSVM Even we did only some random tests and the algorithm can react differently according to the inputdata the results are much better than expected We can state that the new implementation is 70 better dan the MATLAB code REFERENCES 1 K De Brabanter J De Brabanter J A K Suykens B De Moor Optimized Fixed Size Least Squares Support Vector Machines for Large Data Sets 2009 2 V Vapnik Statistical Learning Theory 1999 3 J A K Suykens J Vandewalle Least squar
172. ngineering Technology Electronics ICT Sixteen students report here the results of their research This research was done in companies research institutions and our department itself The results are presented as papers and collected in this text which aims to give the reader an idea about the guality of the student conducted research Both theoretical and application oriented articles are included Our research areas are e Electronics e ICT e Biomedical technology We hope that these papers will give the opportunity to discuss with us new ideas in current and future research and will result in new ways of collaboration The Electronics ICT team Patrick Colleman Tom Croonenborghs Joan Deboeck Guy Geeraerts Peter Karsmakers Paul Leroux Vic Van Roie Bart Vanrumste Staf Vermeulen Open Shop Scheduling of a Linux Cluster using Maui Torque Maarten De Ridder Gert Van Den Eyndep Radu Popescu Joan De boeck 1 KHK Geel ITBT Kleinhoefstraat 4 B 2440 Geel Abstract Scheduling and information services are two main clusters which play an important role in the overall performance of an application running on that cluster When thinking of scheduling one might say that it is sufficient to make sure that the minimum requirements and QoS requirements are met for the execution of a job Unfortunately the whole process is much more complicated First of all a distinction will have to be made between serial jobs which are for
173. ngth The first four counters were also added when testing the performance 2 Testing process When the script is ready and the monitoring counters are set up correctly the actual testing process can begin When testing with tens of users the easiest way to do this is by placing a shortcut to the test script in de Startup folder so that the script runs when the RDC session is launched Because the testing in our case is only with four different users we manually launch the script in each session For testing we could use four different workstations On each workstation we launched one RDC session to the Terminal Server At approximately the same moment we kicked off the simulating script http www autoitscript com autoit3 index shtml Working Set 1 user 2users sw 3users 4 users 1800 1600 1400 1200 1000 7 800 600 400 200 Memoryin MB Mo Or UM AANA UM dam nu MN dd mm O MO Dn OO NN NN Tin OO Ooo dd dd dd dd dd dd rd di IN Time in s Fig 5 Output from the Process gt Working Set gt _Total counter Pages Output sec Miuser M4users 1 2E 17 1E 17 8E 16 6E 16 Pages Output sec 4E 16 2E 16 o Fig 6 Output from the Memory gt Pages Output sec counter Having more RDC sessions on a single workstation is possible but in this case wasnt usable Because the script simulates mouse movements and keystrokes it only works at one RDC session at the ti
174. nsole CMC The CMC Je Pr acceleration 1s a device that can manage multiple Steelheads WDS Ipanema Technologies uses IPlengines that are placed on and management both sides of the WAN There are also virtual Plengines that Management visibili a 9 streamlining Central Management Managed service framework gt must be configured in the management system IPlboss These s Steelhead Mobile Controller Managed service framework gt virtual Plengines are especially efficient for very large networks VLNs Application performance Inteligent i used there must also be placed a Steelhead Mobile Controller Aen ze change SMC in the network The management of the Steelheads can a objectives per user i optimization Riverbed uses a CAPEX Capital Expenditures model Therefore customers must buy the Steelhead devices Ipanema uses an OPEX Operating Expenditures model Belgacom offers Ipanema as a managed service to their customers by which they must pay a monthly fee Table 4 Riverbed and Ipanema features E Discussion A file transfer with WOCs placed in the network is faster than a file transfer without WOCs placed in the network When the appliances are in bypass failsafe mode the transmission time of a file 1s the same as in a network without appliances In a network with appliances the second transmission of a file is faster than the first transmission because
175. ntation is as low as possible for using the biggest possible array sizes we locate and allocate the two biggest chunks of contiguous memory immediately at the start of the test These two memory blocks are used to store the matrices A and B which increase during the lifetime of the test to do a performance test of different sizes until a row size of 10000 While it is sufficient to compare different implementations based on their time spent it may be useful to compare the theoretical and achieved performance The ratio between achieved performance and theoretical peak performance Peak known as efficiency 7 A high efficiency indicates an efficient numerical implementation Performance is measured in floating point operations per second FLOPS and can be calculated as E xk peak Nepy Lore Mepy f with 1 cpy the number of CPUs in the system Noore the number of computing cores per CPU 71 the number of FPU floating point units per core and f is the clock frequency The achieved performance can be computed as the flopcount divided by the time For the xgesv function of LAPACK 1s the standard number of floating point application The libraries GotoBlas2 and MKL are close to operations 0 67 N3 8 each other Intel CPU FLOP FLOP DGESV X E5506 2 13 GH Pier OPS Pia OPS eon 7 double float jee 80 MATLAB GotoBlas2 40 4 Ref LAPACK Core2Duo E6300 14 88 29 76 Xeon E5506 34
176. nterface Sample speed with on board averaging 0 6 msec scan Data transfer speed HD 26 connector 2 Analog in 2 Analog out 3 Digital in 12 Digital IO Digital out trigger sync Default USB power 350 mA Or with SPU2 external 12VDC 350 mA 175 x 110 x 44 mm 1 channel Dimensions weight 716 grams Tablel Power supply The spectrometer is connected to a PC by USB2 0 and controlled with the AvaSoft7 4 software that was delivered with the device It is setup to log the sum of the energy in the wavelength range from 300 1100nm every 30sec The wavelength range responses to the spectral response of mono crystalline silicon The data output is the instantaneous absolute solar irradiation in uW cm at a sample rate of 0 033Hz or 1 sample every 30 seconds Because the ultimate goal is to compare the sensor output with the energy provided by a PV installation we will also correlate the sensor data with a PV installation The PV installation used throughout our research is the setup of the KHKempen It is made up with 10 Sharp ND 175EIF solar panels with a combined surface of 11 76m 3 The panels made of polycrystalline silicon that have an efficiency of up to 12 4 Other specifications of the panels can be found in Table2 Pmax 175 0 Tolerance Short Cireuit Current ls 810 A Voltage at Point of Maximum Power Vmpp 23 3 Vv Current at Point of Maximum Power impp 752 A Maximum Sys
177. o that the user has the best look and feel experience with the multi touch application According to MSDN Windows uses a time out of 500 ms 0 5 s This time out however was too long to be useful in a multi touch environment It did not feel naturally For instance if you want to move an object from the top right corner to the bottom left corner you normally use your right hand first to move it to the middle of the screen then you use your left hand to move it from the middle to the left bottom corner With a time out of 500 ms it was not comfortable to wait while this time out was expired If the user however touches the object withing the time out the code of the doubleclick action will be executed what not always will be the intention of the user From our multi touch experience we took 250 ms as time out This gives a very intuitive feeling for this action The code that can be used is already used for the hold function in section Select group hold and tap With a little modification the code will be useful in this context 2X Figure 10 Open double tap 4 IV CONCLUSION Silverlight 3 0 is a brand new technology that is very promising for a multi touch experience on desktop computers and in the future even mobile phones The multi touch support is not very extended but it is widely customisable That makes it very useful for many programmers who are familiar with C NET and the NET framework to work with As describe
178. oblems with reverberation and noise So for a good speech recognition this factors must be suppressed To do this we applied a sum and delay beamformer and a GSC A sum and delay beamformer performs better in conditions of random noise while a GSC with LMS obtains better results in conditions of periodic noise Finally increasing the number of microphones gives better results for the reduction of random noise A better suppression of periodic noise is obtained by increasing the distance between the microphones ACKNOWLEDGMENT The authors want to thank INHAM for their assistance during the recordings which were necessary for this work In addition we give thanks to ESAT for their investigation with the speech recognizer REFERENCES 1 K Eneman J Duchateau M Moonen D Van Compernolle Assessment of Dereverberation algorithms for large vocabulary speech recognition systems Heverlee KU Leuven ESAT 2 D Van Compernolle DSP techniques in speech enhancement Heverlee KU Leuven ESAT 3 D Van Compernolle W Ma F Xie and M Van Diest Speech recognition in noisy environments with the aid of microphone arrays 2 rev Heverlee KU Leuven ESAT 28 October 1996 4 D Van Compernolle Switching adaptive filter for enhancing noisy and reverberant speech from microphone array recordings Heverlee KU Leuven ESAT 5 D Van Compernolle and S Van Gerven Beamforming with microphone arrays He
179. of both solutions by performing the same series of actions on the traditional desktop and the virtual desktop The testing indicates that the Terminal Server environment is 24 faster than the regular environment It also scores better regarding memory and network usage Working with a Terminal Server environment has many advantages but definitely saving time is an important one III CAPACITY TESTING A Intention Now that we know the difference between the tradi tional desktop solution and the Terminal Server virtual desktop solution we need to know how many users the Terminal Server can support This number can vary greatly because of different environments network speed protocols Windows profiles and hardware and software revisions For this testing we use a script for simulating user load on the server Instead of asking real users to use the system while observing the perfor mance a script simulates users using the system Using a script also gives an advantage you can get consistent repeatable loads The approach behind this capacity testing is the following First we did the test with just one user con nected to the Terminal Server The script runs simulates user activity and the performance is monitored Next we added one user and repeated the test Thereafter we did the test with three and four users because we only had four machines at our disposal Afterwards the results from the four tests can be compared B
180. of more than two then performance could be increased by for example replacing the network card with a faster one In our case when testing the network performance between a regular workstation and a virtual desktop on a Terminal Server we see that both the desktop as the Terminal Server suffice But we have to keep in mind that during the testing only one user was active on the Terminal Server The purpose of the Terminal Server is to provide a workspace for multiple users so the output from the Queue Length counter will be higher 4 User experience Also important is how the user experiences both solutions The first solution which is using a regular desktop is familiar for the user The second solution which is accessing a virtual desktop on a Terminal Server by setting up an RDC connection is not so familiar to most normal users Most of them havent used RDC connections before and having to cope with a local desktop and on top of that a virtual desktop can be confusing This problem can be solved by setting up the RDC session automatically when the client computer is starting up which eliminates the local desktop and leaves only one virtual desktop which is practically the same for an unexperienced user The only difference they experience is that most virtual desktop environments are heavily locked down to prevent users from doing things on the Terminal Server theyre not supposed to D Results We have tested the performance
181. on performance security and network usage The network path is calculated using e Path resources quality and availability e Application performance SLAs Service Level Agreements e Sensitivity level of the information carried in the flow D Intelligent Acceleration Intelligent acceleration reduces the response time of applications over the WAN so that users get the appropriate Quality of Experience QoE TCP has a slow start mechanism that tries to discover what the available bandwidth is for each session This mechanism slowly increases the throughput until the link is congested It assumes then that it has found the maximum available bandwidth Ipanema s TCP acceleration immediately sets each session to its optimum bandwidth This leads to the improvement of the response time of many applications such as those based on HTTP S Ipanema can deliver this TCP acceleration without an Plengine in the branches Devices are only required at the source of the application flows This is called tele acceleration Ipanema s multi level redundancy elimination compresses and locally caches traffic patterns in a cache in the IPlengines of the branch offices This reduces the amount of data transmitted over the network Multi level redundancy elimination uses both RAM Random Access Memory and disk caches Therefore it can compress and cache the traffic patterns of very large files and keep them longtime RAM caches have a smaller comp
182. on accuracy to an acceptable level In this paper two such noise suppression techniques are explored First we have examine the sum and delay beamformer This beamformer is used to limit the reverberation coming from other angles than the steering angle Another example is the Generalized Sidelobe Canceller GSC The GSC estimates the noise with an adaptive algorithm Possible implementations of this algorithm are LMS NLMS and RLS These 3 types were theoretically as well as practically compared Speech experiments indicate that compared to the sum and delay beamformer the GSC with LMS gives the best performance for periodic noise Index Terms sum and delay beamformer Generalized Sidelobe canceller least square noise suppression I INTRODUCTION AND RELATED WORK To change a television station we can use the remote control by pushing a button This is the easiest way but the handicapped aren t able to serve the remote control In this case voice control will be a viable solution Here disabled persons will use their voice 1 e to change the television station For such systems that use voice control it s important that the command is recognized by a speech recognizer For a good recognition the speech signal has to reach the speech recognizer in a good order A good microphone placement can solve this problem This can be achieved with a close talk microphone In some situations it is not possible to place a microphone close to
183. one is not working properly Some simple solutions for these problems are Using a priority system for voice packets Using dejitter buffers Use codecs to minimize small amounts of packet loss Making a minimized congestion network design Since we need to minimize these specific factors we will use Quality of Service QoS QoS is deployed at different points in the network With implementing this we will have a certain voice section that is protected from data bursts Two other subjects that influence design are knowing the amount of bandwidth needed for voice traffic and how we can reduce overall bandwidth consumption Because WAN bandwidth is the most expensive bandwidth there is it would be useful to compress the data we have to send This will be done by a specific codec for example G 711 G 728 G 729 G 723 iLBC The codec that is used at the Warande is the G 729 codec This codec uses Conjugate Structure Algebraic Code Excited Liner Prediction CS ACELP compression to code voice into 8kbps streams G 729 has two annexes A and B G 729a requires less computation but lowering the complexity of the 4 codec is not without a trade off because the speech quality is marginally worsened Also G 729b adds support for Voice Activation Detection VAD and Comfort Noise Generation CNG to cause G 729 to be more efficient in its bandwidth usage If we take a bundle of approximately 25 calls or more 35 of the time wi
184. ontains an Enable GPE_EN register and a Status GPE_STS register These registers are used for all generic fea tures such as Power Management Events PME If the corresponding enable bit is set a Wake Event will be generated 2 2 Software Interface Each register block is set at a fixed hardware address and cannot be remapped The silicon manufacturer deter mines its address location The ACPI software interfaces provides a way for the operating system to find out what register blocks are located at what hardware address The BIOS populates the ACPI tables and stores the memory location to the Root System Description Pointer RSDP into the Extended BIOS Data Area EBDA The operating system scans this area for a string RSD PTR which is followed by 4 bytes This 32 bit address is a pointer to the RSDP At a 16 byte offset the 32 bit address of the Root System Description Table RSDT can be found Figure 1 illustrates this layout Locatedin system s memory address s pace Root System Root System Description Pointer Description Table contents contents Figure 1 RSD PTR to RSDT layout From this point on every table starts with a standard header that contains a signature to identify the table a checksum for validation and so on Thus the RSDT table itself contains a standard header after this header a list of entries can be found The number of entries can be determined using the length field from the table heade
185. ool register their time information through a web interface because the ActiTime application is a web based tool As Flanders DRIVE has the need to introduce a new IT infrastructure several software applications must be migrated to the new infrastructure ActiTime is one of them Several problems appear in the migration process A proper way of how to extract the current user data from the ActiTime database must be found ActiTime uses java servlets through a web based application Since the internet information service IIS of Windows server 2008 doesn t support java servlets a different web server must be chosen This web server need to support the use of java servlets The developers of the ActiTime application recommend the use of an Apache Tomcat Server 2 Since Tomcat is a product of Apache a few problems must be solved to get this server work with Windows Server 2008 A determination of which server is best to use to install the Apache Tomcat server and get ActiTime to work must be made There is no server that can be used and a decision is made to create a virtual machine with Microsoft Hyper V II MIGRATION OF ACTITIME A Flowchart of migration way Gathering installation files and install application Migration of a Time tracking application AyctiT inne Setting up a test anwiranment Test application when problems are found locate the problem and find solution If everything OK install in bus
186. optimization techniques RIOS uses window scaling and virtual window expansion VWE to increase the number of bytes that can be transmitted without an acknowledgement When the amount of data per round trip increases the net throughput increases also This window expansion is called virtual because RIOS repacks TCP payloads with data and data references A data reference can represent a large amount of data and therefore virtually expand a TCP frame The RIOS implementations of High Speed TCP HS TCP and Max Speed TCP MX TCP can accelerate TCP based applications even when round trip latencies are high HS TCP uses the characteristics and benefits of TCP like safe congestion control In contrast MX TCP is designed to use a predetermined amount of bandwidth regardless of congestion or packet loss Connection pooling enables RiOS to maintain a pool of open connections for short lived TCP connections which reduces the overhead by 50 or more The SSL acceleration capability of RiOS can accelerate SSL encrypted traffic while keeping all private keys within the data center and without requiring fake certificates in branch offices D Application Streamlining RiOS is application independent so it can optimize all applications There is a possibility to add additional layer 7 acceleration to protocols through transaction prediction and pre population features Transparent pre population reduces the number of waiting requests that must be
187. oring jobs with lowest cur pansion factor rent scheduling performance promotes bal anced delivery of job queuetime and expan sion factor service levels processors Allows favoring of jobs which meet vari Requested Requested Resources memory swap local disk Jous requested resource constraints ie fa processor voring large processor jobs counters back fills proclivity for smaller jobs and improves overall system utilization nodes and equivalents User group account QoS Allows favoring jobs based on historical us and Class fairshare utiliza age associated with their credentials Fairshare Direct Prior User croup account QoS Allows political priorities to be assigned to and Class administrator various groups specified priorities delta between Allows ability to specify service targets and Allows favoring of jobs bypassed by backfill to prevent backfill based job starvation Figure 1 different factors and their explanation where each Weight value is a configurable parameter and each Factor is calculated from the subcomponents user group initial priority QoS etc After Maui has prioritized the jobs in queue it starts the jobs one by one stepping through the priority list until it reaches a job that cannot start Because all jobs posses a start time and a wallclock limit Maui can determine the finish time of the jobs in the queue When it encounters a high priority job which cannot start it ca
188. packets The community string is sent in plain text The main difference between V1 and V2 is that SNMPv2 added a few more packet types like the GETBULK PDU which enable you to request a large number of GET or GETNEXT in one packet Instead of SMIv1 SNMPv2 uses SMIv2 which is a better with more data types like 64 bit counters etc But mostly the difference between V1 and V2 is internal and the end user will probably not notice any difference between the two 6 SNMPv3 SNMPv3 was designed to address the weak V1 V2 security SNMPv3 is more secure than SNMPV2 It does not use community strings but users with passwords and SNMPv3 packets can be authenticated and encrypted depending on how your users have been defined In addition the SNMPv3 framework defines user groups and MIB views which enable an agent to control the access to its MIB objects A MIB view is a subset of the MIB You can use MIB views to define what part of the MIB a user can read or write 6 B Benchmark In this section we show some figures regarding speed using different possible communication methods serial communication telnet and SNMP Thanks to these benchmarks we are able to select the most suitable communication method in every case at every specific moment First of all the benchmark is written in two languages Perl and Python to check if the results are not determined by the programming language As you can see in figure 4 3 4 4 and 4 5 the
189. pel T Velte Planning and installation Microsoft virtualization with Hyper V 2009 pp 58 59 12 E support Trend Micro available at http esupport trendmicro com 0 Known issues in W orry Free Business Security WFBS Standard Advanced 60 aspx 13 http nl wikipedia org wiki Virtualisatie WAN Optimization Controllers Riverbed Technology vs Ipanema Technologies Nick Goyvaerts Niko Vanzeebroeck Staf Vermeulen IBW K H Kempen Associatie KULeuven Kleinhoefstraat 4 B 2440 Geel Belgium Telindus nv Geldenaaksebaan 335 B 3001 Heverlee Belgium nick goyvaerts telenet be niko vanzeebroeck telindus be staf vermeulen khk be Abstract WAN Optimization Controllers WOCs become more and more important for enterprises because of the IT centralization Telindus offers WOC solutions from Riverbed to their customers and Belgacom offers WOC solutions from Ipanema to their customers Because Telindus now belongs to Belgacom it is useful to know which the appropriate solution is for a certain customer or network Riverbed uses the Riverbed Optimization System RiOS to optimize WAN traffic RiOS consists of four main parts namely data streamlining transport streamlining application streamlining and management streamlining Ipanema uses the Autonomic Networking System or Ipanema system to optimize WAN traffic The Ipanema system is a managed system that consists of three main parts namely intelligent visibility intel
190. perty The Markov property states that the probability distribution for the system at the next step and in fact at all future steps only depends on the current state of the system and not additionally on the state of the system at previous steps 1 e it is memoryless 14 C Performance measures Once the classification of our waiting queue is determined we further described the properties of the system The common characteristic of all Markovian systems is that all interesting distributions namely of the interarrival times and the distribution of the service times are exponential distributions and thus exhibit the markov property From this property we find following important conclusions e The state of the system can be summarized in a single variable namely the number of customers in the system e Markovian systems can be directly mapped to a continuous time markov chain CTMC which can then be solved e For Markovian systems the n step transition probability P n which is the probability that after n transitions a Markov chain will be in state j given that the chain began in state I will approach a limit 7 j for large t which is independent of the initial state 1 This limit is called the steady state or equilibrium probability of state J In our case steady state means the probability that at an instant in the future j jobs will be present in the system We can derive the steady state probability by using t
191. ples The maximum bitrate is 1024 kpbs for PS in this setup this value comes from 32 kHz sampling rate and 16 bit words however when the Philips Subband Coder will be implemented using two development boards the bitrate is limited to 166 kpbs This limitation comes from the capacity of the wireless channel For this reason the maximum bitrate is set to 166 kpbs in the setup with one development board In a first phase different configurations for the Joint Stereo and Dual mode will be tested When this is done the best configuration for each mode is selected Then another test is done by comparing all the selected configurations in this phase the mono mode 1s also included The listening test s are done using the MUSHRA Multiple Stimuli with Hidden Reference and Anchor test 10 VII MUSHRA LISTENING TEST The MUSHRA listening test is used for the subjective assessment of intermediate audio quality This test is relatively simple There are a few requirements for the test signals They should not be longer than 20 s to avoid fatiguing of listeners A set of signals to be evaluated by the listener consists of the signals under test at least one anchor in this test two anchors and a hidden original signal The listener can also play the original signal The anchors and the hidden reference are used to see if the results of a listener can be trusted In this way more anomalies may be detected The anchors are the original signal
192. plexity codec like the Philips Subband Coder is used In this paper an implementation of the Philips Subband Coder SBC is discussed and a comparison with the G 722 speech codec will be made L INTRODUCTION Hearing aids have improved greatly over time Today a lot of hearing aids are binaural This means that the audio received on the right hearing aid will also be transmitted to the hearing aid in the left ear and vice versa This greatly improves the hearing quality The reason for this is simply the human brain The brain needs both ears to determine where the sound is coming from the distance and most importantly it helps to sort out speech from noise In 1 benefits of binaural hearing are discussed In this paper a hearing aid that uses the G 722 speech codec to compress audio is discussed this is a problem because this greatly diminishes audio quality for music signals Therefore a better codec that also can handle music signals is searched in this paper Hearing aids are real time devices and the sound received on one side must be heard on the other side with minimum delay For this reason delay becomes a big issue When the delay becomes too large the person wearing the hearing aid would hear an echo if there is no compensation by introducing buffering Ideally it would be best to have zero delay but since there always will be some processing delay this isn t possible It is necessary to keep the delay as low as possible fo
193. power dB at a certain range The calculated SNR can then be used to predict target visibility at a certain range or in a cluttered area and to assess radar performance IV DATA COMPRESSION Data compression is important in the way of disk and memory usage If we only write necessary data to disk data will take up less memory and disks will be used less intensively The speed of continuous writing is calculated using R f N 2 where R represents write speed in MB s f represents sampling frequency in MHz and N represents the number of bytes per sample Using a sampling frequency of 16MHz and having 2 bytes sample this gives us 32MB s As shown the write speed used for data writing without filtering is 32MB s which means that a ITB disk will be full after recording about 9hours If we exaggerate and state that there is only 1 target in unfiltered data on a 1TB disk we have wasted about 99 of disk space which is of course unwanted If we then want to analyze the radar system we will have to read all data and check all data for targets to analyze which will both take up too much time It could be depending on the number of targets that we are able to use a 1TB disk for a recording of 2 or more days which is a big improvement Therefore filtering targets before writing raw video data to disk is a big step forward We will do this by filtering a window out of primary radar raw video based on target information range azimuth co
194. r The first of many RSDT entries is the Fixed ACPI Description Table FADT This table is a key ele ment because it contains entries that describe the ACPI features of the hardware Figure 2 illustrates this At different offsets in this table a pointer an s to the I O locations of various Power Management registers Static info FIRM DSDT BLKs can be found for example the PM1a CNT BLK The FADT also contains a pointer to the Software Differentiated System Tame Dm Description Table gt DSDT table which NIT contains information and ae PX BLK hs descriptions for various port space ze C oe ce Z A do Pe 2 ce 2 KEES Rete OD dn en ONA en oS NE system features KO SU eee oe ee ce a SU seh Oe ots ite oes 25 nie te ae gt 2 3 PMI_CNTBLK This is a 2 byte register and contains two impor tant fields The SLP TYP is a three bit wide field that defines the type of hardware sleep the system enters Figure 2 FACP contents into when enabled Possible values associated with their sleeping state can be found in the DSDT When the desired sleeping states is inserted into the SLP_TYP field the hardware must be told to initiate This is done by writing a one to the one bit field SLP_EN 2 4 DSDT The Differentiated System Description Table contains in formation and descriptions for variou
195. r is doing their internship at SCK eCEN this could be used to lower his priority e Fairshare configuration the window of the fairshare is set to 15 which means that the last 15 days will be taken into account Also the decay was set to 0 95 insuring that the recent past is more important but not fully decisive The calculation for the fairshare target is depicted in figure 3 QueueTimeWeight min QueueTimeCap QueueTime 7 XPactorWoight min XFCap X Factor p R esource MIN RESOURCECAP NODEW EIGHT Nodes PROCWEIGHT Processors MEMWEIGHT Memory p SWAPWEIGHT Swapy DISKWEIGHT Disky PEWEIGHT PE p MIN FSCAP FSUSERWEIGHT FSDealtaUserUsags User z FSGROUPWEIGHT FSDeltaGroupUsage Group y FSA CCOUNTWEIGHT FSDeltaAccount Ussgs Account p PS QOSWEIGHT FSDalta OSU sage QOS 7 4 p FSCLASSWEIGHT FSDeltaClacsUsage Class USERWEIGHT Priority U s r p GROUPWEIGHT Priority Group z ACCOUNTWEIGHT Priority Account y QOSWEIGHT Priority Q057 CLASS WEIGHT Priority Clase 7 2 5 Tar get MAX 0001 X F Target XFCurrent 7 MAX 0001 QT Target QT Current z 2 NOTE XF is a unitless ratio while QT is reported in minutes Bypass Count y Figure 3 formula s for calculating different parameters To make a prediction to which setup would be most optimal we modeled the system using a simplex algorithm Not only does this make an assumption about how to c
196. r role 1s to execute submitted jobs On each compute node pbs mom runs to start kill and manage submitted jobs It communicates with pbs server on the master node Depending on the needs of the systems a compute node may double as the master node or more e Resources Some systems are organized for the express purpose of managing a collection of resources beyond compute nodes Resources can include high speed networks storage systems license managers and so forth Availability of these resources is limited and needs to be managed intelligently to promote fairness and increased utilization II QUEUING SYSTEM A Introduction In queuing theory the main objective is to manage the system is such a way that both the unit that has to wait and the one that serves get the most benefit 11 The unit demanding service is identified as the customer The unit providing service is known as the server This terminology is used in a generic sense regardless of the nature of the physical context In most cases the arrivals are a product of external factors Therefore the best one can do 1s to describe the input process in terms of random variables that can represent either the number arriving during a time interval or the time interval between successive arrivals The uncertainties involved in the service mechanism are the number of servers the number of customers being served at any time and the duration and mode of service Networks of q
197. r the audio codec A second limitation in the choice of an audio codec is the hardware Hearing aids need to be as small as possible for high comfort This means there isn t much space for hardware such as memory A third limitation is battery life A hearing aid needs a battery to operate and it isn t comfortable if the battery needs to be changed to frequently These two limitations also imply that a low complexity codec is needed These limitations are the reason why the Philips Subband Coder was used in this paper In this paper a closer look is taken at what the causes for delay are in an audio codec since this is a very important factor for a hearing aid application The delays from other codecs 3 than the Philips Subband Coder will be looked at Next a closer look is taken on how the Philips Subband Coder and the G 722 codec work and a comparison is made Next the integration of the Philips Subband Coder is discussed After the implementation an evaluation of the audio quality is made On the basis of this evaluation the configuration parameters are determined that are best used for the Philips Subband Coder The results are compared with the evaluation of the G 722 codec from 4 Il DELAY INTRODUCED BY CODECS Here some important elements that cause delay are discussed Only elements that are relevant to the codecs used in this paper are discussed A Filter Bank The delay in audio codecs has many different sources One
198. re obtained with the close talk microphone which resulted in a word error rate of 3 6 Switching to lavalier the handheld microphone the PZMs or the commercial microphone array increased the error rate to 4 68 16 2 30 96 and 43 2 respectively while the speech recognizer uses state of the art environmental compensation techniques Based on this results signal conditioning techniques were required in absence of nearby directional microphone This is necessary to limit the influence of noise and reverberation The results for the second set of recordings showed higher error rates Now the error rate starts from 48 for a person with a slight speech impairment and going up to 80 and more for pathological voices when using the close talk microphone The error rate is also influenced by several factors a short rest in the pronunciation of a command dialect of the test subjects slower speaking rate noise from other persons than the test subject WER 3a Sooo A Lavalier Handheld PZM Close talk Fig 6 WER Commercial Array Based on the results from the first experiment we investigated some techniques to limit reverberation and noise For this research we compare the sum and delay beamformer and the GSC However the GSC has an adaptive algorithm So we have to examine the most suitable algorithm for this adaptive algorithm For this experiment we use the data from the second set of recordings With figure 3 kept in min
199. relationship between serial telnet and SNMP is almost the same At this moment we can conclude that the results are independent of the programming language 80000 7 70000 7 60000 7 50000 40000 m Python 30000 E Perl 20000 10000 Serial Telnet SNMP Fig 1 GET 100000 7 90000 80000 70000 60000 50000 m Python 30000 20000 10000 04 Serial Telnet SNMP Fig 2 SET wait 70000 7 60000 50000 40000 E Python 30000 mPerl 20000 0 I I Serial Telnet SNMP Fig 3 SET no wait This benchmark is split up into three different tests Get Set with a wait function and Set without a wait function The length of the command and the execution time of a command are also considered GET Get a variable from the switch 500 times SETwait Set a parameter of the switch and wait until this parameter is in the requested state 50 times SETnowait Set a parameter of the switch and it doesn t matter if it is already in the requested state 500 times Long exectution time Long command GET sh interf gigabitEthernet 1 0 1 mtu SET inter gig 1 0 1 shut Long exectution time Short command GET sh in gig 1 0 1 mtu SET in gig 1 0 1 shu Short exectution time Long command SET hostname abcdefghij Short exectution time Short command SET hostname abc TABLE I COMMUNICATION METHODS Serial GET I B s s 00 33 344 00 34 953 00 32 984 00 33 110 00 33 141 00 33
200. ression ratio than disk caches Intelligent protocol transformation can optimize protocols to minimize the response time of applications IV COMPARISON BETWEEN BOTH SOLUTIONS A Lab We have created an equivalent test lab for both solutions to see which solution performs the best in this simple network environment Etherneti NET 1 Ethernet NET2 N Y AM Z SS P Simul T 192 168 1 0 24 cut 192 168 2 0 24 LAN Riverbed SH100 WOC02 In path 192 168 2 240 24 Primary 192 168 2 241 24 Riverbed SH100 WOCO1 In path 192 168 1 240 24 Primary 192 168 1 241 24 Clie IP ae 192 168 2100 24 fa ult Gat i De 192 168 2 DBA Management WOC server2 oe lator IP adres 192 168 150 24 Managemen Default Gateway IP ae s 192 168 0 254 24 192 168 1 254 Ether net NET1 IP adres 192 168 1 254 24 File Server Ethernet NET2 IP adres 192 168 2 254 24 Figure 3 Riverbed Technology lab image_rbt_cm FileZilla_3 2 6 image_rbt_cm geenen c_2 1 5 n23 i File name 1_win32 c_2 1 5 n23 i a mg met setup_RENAM setup exe mg gewijzigde ED exe inhoud File size 3 81 MB 39 2 MB 3 81 MB 39 2 MB Transmission speed time without WOCs Transmission speed time WOCs in bypass mode Transmission speed time WOCs with cold optimization Transmission speed time WOCs with warm optimization 78 6 KB s 8 minutes 42 seconds 78 6 KB s 8 minutes 42 seconds 161 3 KB s 4 minutes 14 se
201. rimitive volume was tested by intersecting them under similar conditions The idea behind the tests was to analytically calculate the intersections and then compare them against the outcome of the ray tracer Each volume was made to intersect with a single ray at different locations of the surface and at different resolutions 20 50 100 We let the ray intersect a vertex and the middle of a triangle The position of a vertex is the exact position of a point on the surface of a volume while the middle of a triangle is where the model deviates the most from the real surface The distances in the tests on the vertices provided perfect results no errors were measured for these volumes This means that the method itself is highly accurate however the problems arise when the intersection is closer to the middle of a triangle Boxes retain their perfect results when the intersection moves to the middle of the triangle Curved surfaces however experience significant deviations At a resolution of 20 a curved volume with a radius of 200 cm can give errors greater than 2 cm Even at a resolution of 50 there were deviations of a few mm Table 2 Deviations of the ray traced intersections at 20 cm in cm Resolution 20 50 100 20 50 100 In table 2 the same results are shown for volumes with dimension that are 10 times smaller It seems that the deviations are more or less 10 times smaller as well Results vary greatly across the various volumes
202. rmance First of all our company policy requires the server to run together with other services on a Small Business Server Our customers do not have the budgets to run such tools on dedicated servers This brings us to determine which factor is the most important for the company We ve decided that a tool for monitoring purpose to prevent problems may not cause one by tearing down the network in performance The network load of such a tool should not interfere with the normal work of a server room Followed up by the server load with as most important factor the disk operations As mentioned before the tool will not run dedicated but together with other servers like SQL Database Servers Such a server requires all data to be processed and not being lost by scans of a monitoring tool This means that disk operations transfers sec to be precisely may not reach a certain limit of IO maps sec or data can get lost in the process Other parameters like memory and CPU are not so important because servers are powerful machines that most of the time run beneath their capabilities Bringing us to the last but not least parameter the price Good tools proportionally go with the price Because the most of our customers are smaller companies the price should be in the or read by the SOL Database server We can see same order MonitorMagic 1s in the top 5 tools that use the least disk erformance B Network load P As we take a look at the
203. rque as his scheduler Torque is primarily used in batch systems This means that once a job is submitted by a users Torque will handle everything from start to execution Pooling resources in a batch system typically reduces technical administration of resources while offering a uniform view to users Once configured properly batch systems abstract away many of the details involved with running and managing jobs allowing higher resource utilization For example users typically only need to specify the minimal constraints of a job and do not need to know the individual machine names of each host on which they are running With this uniform abstracted view batch systems can execute thousands and thousands of jobs simultaneously 10 Torque uses four different components e Master Node A batch system will have a master node where a job server runs Depending on the needs of the systems a master node may be dedicated to this task or it may fulfill the roles of other components as well e Submit Interactive Nodes Submit or interactive nodes provide an entry point to the system for users to manage their workload For these nodes users are able to submit and track their jobs Additionally some sites have one or more nodes reserved for interactive use such as testing and troubleshooting environment problems These nodes have client commands such as qsub and ghold e Compute Nodes Compute nodes are the workhorses of the system Thei
204. rver supports java servlets the installation of this web server is very straightforward 5 For the java application a java runtime engine 6 machine was chosen ActiTime also need a database to store all the user data There are 2 options to use you can choose between MS Access 3 0 or later and MySQL 4 1 17 5 0 x or 5 1 x In this case we choose the MySQL server 5 1 machine We choose the MySQL option because for this application it suits better than Microsoft Access However these 2 database system are completely different we still can conclude that MySQL is better in this scenario Microsoft Access can be very slow if more than 3 clients make a read write connection to the database simultaneously Microsoft Access is more a desktop application than an application to use for internet applications MySQL is more efficient and secure in environments with multiple users connection to the database simultaneously Microsoft Access has a well developed user interface to create database scheme s while MySQL has no user interface only a command prompt window to access the database scheme 6 In the situation of ActiTime we don t need a good user interface because the web application processes the data for us Since the data is already in an MySQL format it is simpler to migrate the data to a new MySQL server because no database redesign Is necessary The test environment is set up with virtual machines who can be accessed through Microsoft Vir
205. s header so we only have one file We have chosen for the second option because it is easier to program and it is easier to read data afterwards When a target is filtered its header is created and his raw video data is added We then place this data incl header into a second buffer which hands this data over to a second program that writes this data to disk V REAL TIME SIMULATION EXPERIMENT Since we didn t have the possibility to test the real time program at a radar station we have written a program in LabVIEW7 that simulates a real time system for 1 full scan We use generated primary radar data and matching secondary radar data Since synchronizing and simulating data streams in LabVIEW isn t an easy thing to do we had to add some code in the real time C program for testing purposes only Simulations have confirmed the working of the real time filter and parallel calculation of the parabolic fit error number and SNR as described previously VI ACKNOWLEDGEMENTS We would like to express our gratitude to Peter Lievens for his technical support concerning Labview7 We would also like to express our gratitude to Erik Moons and Johan Vansant for their technical support concerning C VII CONCLUSIONS AND FUTURE WORK In this paper we have discussed radar parameter calculations which will be used in future work for radar performance analysis We have also discussed real time primary radar data compression and the decisions
206. s inside a triangle by using a half plane test Each edge of the triangle cuts a plane in half with one half plane defined as inside the triangle and the other outside This test is reduced to three simple equations 4 P is the intersection point Box Cylinder BADEN 0 15 Sphere P P x P P N gt 0 16 In table 1 we show the results for three common volumes of P F 3 x P F 3 N gt 0 17 similar sizes radius width depth and height at 200 cm The If all of the above equations are satisfied the point is inside the triangle Any equation resulting in a zero means that the intersection is exactly on an edge of the triangle Such an intersection will be shared by another triangle and could be counted double if the program doesn t take this into account Other point in polygon strategies exist but the half plane test explained above is easily the fastest for triangles 5 C Point between endpoints test The final test determines whether the intersection is between the starting and ending point of the ray P P Pe Fig 4 Point between the endpoints of a line segment d P P d P P d h P 18 This equation will only be satisfied if P is between P and P In any other case the right hand side will be greater than the left hand side IV ACCURACY The accuracy of the intersections is extremely important for further calculations The accuracy of the intersections with each type of p
207. s is proportional to the scaling factor that was calculated in the previous step The bit allocation is based on the fact that the quantization noise in a subband can be kept equal over a 6dB range An increase of of the SFI for one band increases the quantization noise with 6dB if one bit is added to the representation of a sample the quantization noise drops by 6dB Thus the quantization noise can be kept constant over all subbands within 6dB The bits are then distributed using a water filling method Subband 9 2 3 4 5 6 T SFI 0 2 3 2 8 OM 4 5 Resulting Allocation 10 0 3 5 7 2 FIGURE II WATER FILLING After the adaptive bit allocation the samples in each subband are guantized using the available bits assigned to each subband For decoding the samples the guantized samples are multiplied with the scale factor After this decoding these samples are sent to the synthesis filter bank Quantized Scale Factor Transmit guadrature mirror filters Transmit audio part da Audio signal Receive output Transmit quadrature audio part mirror filters L Higher sub band 16 kbit s XH ADPCM encoder IH MUX Lower sub band f 48 kbit s ADPCM encoder Higher sub band ADPCM decoder 16 kbit s Lower sub band ADPCM decoder 3 variants Auxiliary data channel input 0 8 or 16 kbit s 64 kbit s Data insertion device Data extra
208. s system features mostly vendor specific information of the hardware For example the DSDT tables contains a S5 object that contains three bits can be written to the The SLP_TYP field 2 5 Summary At this point we know what steps need to be taken to power off a simulation device We can conclude that it is possible to power off any ACPI compliant system which is the case for all motherboards used in simulation devices at Alcatel Lucent 3 REMOTE CONTROL POWER OFF Layer 2 packets are used to send commands to the simulation devices This means that it can only be used on the same layer 2 domain e g broadcast domain The packets are captured by the operating system kernel This means that there is no application on top of the kernel processing incoming packets This approach is chosen to capture these management packets as soon as possible in kernel space so the upper layers cannot be affected in any way All simulation devices have a unique 6 byte MAC address and a target name which is has a maximum length of 32 bytes Every device uses this target name to identify itself IP addresses are not unique and may be shared between simulation devices 3 1 A layer 2 packet also known as an Ethernet II frame starts with a 14 byte MAC header followed by variable length payload the data and ends with a 4 byte checksum Packet Layout 3 1 1 MAC Header ar The MAC header consists of the destination MAC ad SK dress to
209. s usable to confirm the output of a PV setup ACKNOWLEDGMENT Special thanks go to Wim van Dieren of Imspec for lending us the AvaSpec spectrometer 1 2 3 4 5 REFERENCES L J B McArthur April 2004 Baseline surface radiation network BSRN WCRP Operations Manual World Climate Research Programme WMO ICSU Carlo Gavazzi 2008 Datasheet Irradiation Sensor Model CELLSOL 200 Avantes April 2009 AvaSpec operating manual Sharp corporation Datasheet Solar Module No ND 175E1F Texas Instruments August 2005 Datasheet MSP430x20x3 Mixed Signal Microcontroller Construction and validation of a speech acquisition and signal conditioning system J Mertens P Karsmakers ha B Vanrumste IBW K H Kempen Associatie KULeuven B 2440 Geel Belgium ESAT SCD SISTA K U Leuven B 3001 Heverlee Belgium jan mertens peter karsmakers bart vanrumste khk be Abstract In most cases a close talk microphone gives an acceptable performance for speech recognition However this type of microphone is sometimes inconvenient Other types of microphones such as a PZM a lavalier microphone a handheld microphone and a commercial microphone array might offer solutions since these need not be head mounted On the other hand due to a larger distance between the speakers mouth and the microphone the recorded speech is more sensitive to reverberation and noise Suppression techniques are required that increase the speech recogniti
210. se ActiTime Note that there Is no space between p and lt password gt e The parameter after gt is free to choose The database backup is stored in the file specified after the gt symbol e The parameter before the gt sign specifies the name of the used database This command can be executed with the Windows command prompt In the Windows command prompt window you have to navigate to the right directory where the database is stored Now simply execute the command explained above and a database backup of the ActiTime application is made and saved in the actitime_data sql file c Opdrachtprompt C Program Files MySQL MySQL Server 5 i bin gt mysqldump u o E Qer gt actitime_data sq C Program Files MySQL MySQL Server 5 1Nbin gt Figure 2 2 command prompt example to extract the user data from the ActiTime database C Setting up a test environment The installation files of the ActiTime application can be found on the website of ActiTime 4 In this situation we choose to download the custom installation package The reason that we choose to download the custom installation package is that in this package customizations in the application can be made One of the customizations is the java application With the custom package you can choose which java application to install and which web server to use For the web server the Apache Tomcat Server version 6 0 20 is best to use because this web se
211. sing the ACPI architecture but need the operating system to manage it This paper describes how to use the ACPI framework to remotely power off a simulation device Layer 2 network frames are used to send commands to either the running operating system or powered off simulation device When powered off the network interface card cannot receive these frames Therefore limited power must be restored the PCI bus and network device Also the network device internal filter must be re configured to accept network frames that can initiate a wake up This result is an ACPI compliant system that can be remotely powered off to save energy and can be powered on when required 1 INTRODUCTION Alcatel Lucent s IP Division uses more than 7000 simulation devices These devices are mostly only used during office hours and left on at night wasting electricity Some of these run heavy simulations or test suites and must be left on overnight Every 42 unit rack has a single APC circuit that can be interrupted using a web interface This will power off all devices within the rack including the ones with heavy tasks that should have been left on The objective is to research and provide the possibility to power off a single simulation device using existing infrastructure and hardware components If remote power off is possible it is also required to power on the same device remotely 2 ACPI The Advanced Configuration and Power Interface 5 is a specific
212. so the ray doesn t intersect them anymore but we ll still have 30 volumes in the scenes Table 3 Time reguired for a line of sight calculation in ms pS EA po S099 a 09 Table 3 shows the time in milliseconds required for a line of sight calculation in three different scenes one with boxes one with cylinders at a resolution of 20 and one with spheres again at a resolution of 20 As expected the time increases significantly as more volumes are intersected this is especially true for spheres This can be explained because the polycount the number of polygons used on the volume increases more rapidly for spheres when the resolution is increased We can see that the performance for most scenes is significantly higher than the older method a few ms as opposed to 10 ms However in the previous section we concluded a much higher resolution is often needed to reach an acceptable accuracy Table 4 Time required for a line of sight calculation in ms Intersected Cylinders Cylinders Spheres Spheres 9 4s 864 039 040 Table 4 shows the results for scenes with cylinders and spheres at higher resolutions The results look good for cylinders Even in a scene with cylinders at a high resolution that are all intersected the time doesn t exceed the 10 ms of the old method It s a different story for spheres At higher resolutions the performance deteriorates dramatically This means that in complicated scenes with m
213. sources and by reducing the radioactivity of the sources as much as reasonably possible The VISIPLAN 3D ALARA planning tool allows users to simulate real world situations and evaluate radioactive doses calculated in this simulation VISIPLAN provides the tools to create virtual representations of real world infrastructure objects radioactive sources etc using primitive volumes A primitive volume is a mathematically generated polygon mesh model which means it s a surface approximation rather than an exact representation This means that only objects with flat surfaces robby nijs sckcen be joan de boeck khk be such as boxes or hexagonal prisms can be modeled in an exact way Most objects however have some curved surfaces introducing approximation errors The resolution of the approximation allows controlling the amount of the error The higher the resolution the more polygons triangles are used to render the object A cylinder with a resolution of six will use six side faces reducing it to a hexagonal prism while a resolution of 20 produces a much better approximation at the cost of performance This explanation of surface approximation seems trivial but it is crucial in this work because it s this triangulated approximation that is used directly in the calculation of intersections We can t expect to find accurate coordinates of intersections on a cylindrical storage tank if it s modeled with just six side faces A simul
214. spikes will become less and less pronounced until the counter begins rising overall The point where spiking is finished and the overall rising begins is a critical point for the Terminal Server This indicates that the Terminal Server hasnt enough memory and could benefit from more memory If this counter does not have an overall rise after the spiking is finished then this indicates that the server does have enough memory As described in section I C2 the system only trims memory when physical memory utilization gets higher We can see in the figure that the counter values are low even when four users are running the script This means that inactive pages aren t trimmed and are still in the working set Therefore we can conclude that more than seven users could use the Terminal Server at one time although the exact number can t be determined from the results Processor Queue Length E 4 users Processor Oueue Length l ded at at U U el U at at et ed Time in s Fig 7 Output from the System gt Processor Queue Length Output Queue Length E 4 users Output Queue Length THE mm u MM Hm dd ed dede ed ed Time in s Fig 8 counter Output from the Network Interface gt Output Queue Length Note the actions performed in this test are extreme and probably most users never will access al
215. st thing we need to do is test if the output of the sensor is linear with the solar irradiation The sensor output is the instant value of the solar irradiation To reference the sensor output with the electrical energy output of a PV solar panel installation we are required to integrate the samples over time For irradiation monitoring a 1Hz sampling rate is recommended minimally to ensure accurate energy integral estimates 1 However the analog input of the MAC data logger has a maximum sample rate of 1 sample a minute or 0 016Hz To address this we plan to setup a microcontroller to sample the sensor output at 1Hz or faster Then calculate the integral of these values and send pulses on the output accordingly These can then be logged with the digital input of the MAC data logger Il SENSOR LINEARITY RESEARCH A Reference Devices For testing the linearity of the Cellsol 200 sensor we require a reference to compare the values The reference device used was the Avantes AvaSpec 256 USB2 Low Noise Fiber Optic Spectrometer The specifications of the device can be found in Tablel 2 And it had a calibration report stating an absolute accuracy of 5 Wavelength range 200 1100nm 0 4 64nm stray light Sensitivity counts uW per ms integration time 120 16 bit AD CMOS linear array 256 pixels Signal Noise 2000 1 16 bit 500 kHz Integration time 0 6 msec 10 minutes USB 2 0 high speed 480 Mbps RS 232 115 200 bps I
216. t Comparisation of response times E 1 user pc E 1 user laptop m 1 user 12Mbit Fig 3 Response time of 1 user Figure 3 shows that there is no difference between the usages of a slower or a faster pc If we raise the bandwidth of the Internet from 4Mbit to 12Mbit we measure a small difference of Spercent This difference isn t significant enough and is unimportant B Several simultaneous users In figure 2 we see that the response times for report generation are the longest In the following test we measure the response time for the generation of reports when more and more simultaneous users pass through the load model seen in figure 1 Up to 25 simultaneous users there s an increase in duration of the response time directly proportional to the increase of the user loads when generating a report This is shown in figure 4 We also notice that out of 25 users there are some users who will get an error in answer to their request because the server can t process the load Several simultaneous users r 25 200000 150000 ZETA time ms r 15 number of errors 10 100000 50000 T of generated reports 0 5 10 15 20 25 Fig 4 Several simultaneous users up to 25 users If we raise the number of simultaneous users up to 100 users we can see in figure 5 that there will be a logarithmic increase in the duration of the response times The reason for this is that the number of users that re
217. t It is also pointed out that out of user experience the traditional workstation is more familiar and easier to cope with than the Terminal Server environment with a local desktop and on top of that a virtual remote desktop Next it is important to know the capacity of your Terminal Server This is indicated by the number of users that can access and use the Terminal Server simultaneously This is tested by comparing a predefined series of actions executed in only one user session with the same predefined series of actions in two three and four different user sessions The user actions were simulated by using a script We learn that the Terminal Server in our environment with the current server hardware and 3 GB of RAM can only support 7 users When considering real users the conditions are less extreme and the server can probably support a lot more users Adding more memory results in more users Other bottlenecks in Terminal Server environments are processor time and network usage Processor time in our case is likely to be a bottleneck depending on the Processor Queue Also the network isnt the limiting factor and if it ever turns out to be one installing a faster NIC in the server fixes this factor in an easy way V ACKNOWLEDGEMENTS The authors would like to thank the ICT team from TBP Electronics Belgium situated in Geel for help and support Special thanks to ICT team manager Rudi Swennen REFERENCES 1 B S Madden and
218. t cases not very difficult However when something goes wrong in the migration process it is often hard to determine the exact problem and to find a solution for it In the migration process described in this paper we found a problem with the Apache Tomcat server The problem could be fixed by placing a missing dll of the Java Virtual Machine in the right directory of the Tomcat server A selection of a possible server to move the application to had to be made After the selection process we came to the conclusion to set up a virtual machine through hyper v because there was no server available to run the time tracking application After we set up a virtual machine through hyper v a rear problem occurred The created virtual machines couldn t start and began to disappear in the hyper v management console This problem occurred because there was a conflict with the trend micro real time scan application The conflict could be solved by excluding the virtual machine directory from the real time scan list In the next figure you can find a short summary of the way followed to come to a working migration of the ActiTime application Installation In the business network Installation on a tes machine Locate problem with the tomcat Server Migration problem with accomplished hyper V Figure 5 1 Summary flowchart of the ActiTime migration ACKNOWLEDGMENT I would like to express my special thanks to Flanders
219. takes about 16 minutes 8 seconds Thanks to the script a switch can be tested in 2 minutes 41 seconds To accomplish this improvement we benchmarked three different communication methodes When SNMP is prefered in one case telnet or serial communication are recommended in another Table 13 1 offers you a short summary the x represents a don t care If two options are mentioned the first one is the most desirable Keeping these results in mind the script is written in Perl Afterwards a custum made benchmark constatates that rewriting the script doens t give a remarkable additional value Perl is the best among all scripting languages This language also provides some effective external modules to handle network operations Java and C are the fastest but requires better programming skills From now on this script will be in use at Telindus headquar ters execution time command length x long x short SNMP Telnet SNMP Telnet SET wait long long x long short xX short long SNMP Telnet short short SNMP Telnet long long Telnet long short Telnet short long Telnet Telnet TABLE VII CONCLUSION ACKNOWLEDGMENT We would like to express our gratitude to Dirk Vervoort Kristof Braeckman Jonas Spapen and Toon Claes for their technical support We also want to thank Staf Vermeulen and Niko Vanzeebroeck for supervising the entire master thesis process Also thanks to Joan
220. te the periodic and or random noise source at the right side of the array This is done in MATLAB by adding the corresponding delay to the noise signals Afterwards the noise signals must be added to the reference signal to get different desired signals Now it is just as if that the simulated signals were caught by the microphone array Finally we take from each signal 12 seconds of data sampled at 8 kHz as input for the test On this data the SNR gain is calculated by taking the difference in SNR before and after applying the sum and delay or GSC algorithm Due to the presence of the adaptive algorithm in the GSC the GSC algorithm is tested for different convergence factors and filter lengths The results for this test can be found in Table 1 Table 2 and Table 3 Here Table 1 shows the SNR gain for the different microphone arrays tested on the sum and delay algorithm Because the sum and delay algorithm is also part of the GSC algorithm an additional SNR gain is showed in Table 2 and 3 This gain is calculated by subtracting the gain Table 1 SNR gain in dB with the use of the sum and delay algorithm in different circumstances array with 2 microphones and d 0 024 A array with 4 microphones and d 0 024 m B array with 6 microphones and d 0 024 m C array with 2 microphones and d 0 072 m D This table makes also distinction between two types of noise On one hand periodic noise On the other hand random noise A
221. tem Voltage 1000 V Over Current Protection s a Maximum Power power per surface P W m Temperature Coefficient of Pmax 0 485 Temperature Coefficient of Voc 0 104 Temperature Coefficient of Isc 0 053 Table2 The converter used is the SMA Sunnyboy 1700 which is equipped with an RS485 interface that allows it to be connected to a PC and allows us to log its input and output This logs the instant input and output current and voltage instantaneous absolute output power and the meter reading of the kWh meter every 30 seconds For all measurements the Spectrometer and the sensor where installed right next to the solar panel setup pointing in the same direction under the same angle so that the input for all the 3 setups was the same B CELLSOL 200 For measuring the linearity of sensor an interfacing circuit was needed to transport the sensor signal from the PV installation outside to the data logger inside over a 10m long cable To prevent the loss of signal strength over the long cable we setup a circuit at the sensor side to convert the voltage signal of the sensor to a current signal For this we use the AD694 transmitter IC that converts a 0 to 2 5V input to a 0 to 20mA output Because the sensor needs a high impedance input and to amplify the signal from the sensor to a range of 0 to 2 5V an opamp was used A second opamp circuit at the data logger side will convert the O to 20mA current signal to a 0 to
222. ter science Volume 2221 2001 pp 87 102 8 Srividya Srinivasan Rajkumar Kettimutu Vijay Subrarnani Characterization of Backfilling Strategies for Parallel Jobs 2002 International Conference on Parallel Processing Workshops ICPPW 02 2002 pp 514 9 J Kay P Lauder A fair scheduler Communications of the ACM ACM New York Volume 31 1988 pp 44 45 10 Torque administrator s guide version 2 3 ClusterResources inc http www clusterresources com 11 U Narayan Bhat An introduction to queuing theory Birkhauser Boston 2008 pp 14 18 isbn 978 0 8176 4724 7 12 Shaler Stidham Jr Optimal Design of Queueing Systems University of North Carolina Chapel Hill North Carolina Chapman amp Hall CRC Press 2009 ch 1 isbn 978 1 58488 076 9 13 Wayne L Winston Operations Research Applications and Algorithms Fourth edition Thomson Learning inc Brooks Cole 2004 pp 1061 isbn 0 534 42362 0 14 Andreas Willig A short Introduction to Queueing Theory Technical University Berlin Telecommunication Networks Group Sekr FT 5 2 einsteinufer 25 Berlin July 1999 15 Little J D C A Proof of the Queueing Formula L A W Operations Research 9 383 387 1961 Primary Radar Performance Analysis and Data Compression S Delarbre N Van Hoef G Geeraerts BW K H Kempen Associatie KULeuven Kleinhoefstraat 4 B 2440 Geel Belgium Intersoft Electronics nv Lam
223. the buffer value surpasses a predefined threshold value the buffer will be reset by subtracting the threshold value from the buffer value When this happens a digital pulse will be sent at the output This resembles integrating the input signal over time the integration of power over time is energy so every pulse resembles a measured amount of energy The pulse output is chosen so that we are able to use the same data logger as it also has a pulse counter The most common used pulse output in energy meters is the SO interface described by DIN 43864 A The setup The used microcontroller is the MSP430F2013 from Texas Instruments it s a chip based on the 16 Bit RISC Architecture that provides us with a 16 Bit Sigma Delta A D converter with internal reference and internal amplifier a 16 Bit timer and several digital outputs 5 Since there is only 1 timer available we will use this for setting up the sample rate of the ADC as well as the timing for the digital output So at every clock interrupt the input will be converted and added to the buffer value and then compared with the threshold value If it exceeds the threshold the output will be set to high In order to produce a pulse it is required that that next interrupt after the output is set to high it is always set to low There for the threshold value must be chosen to be at least 2 times the maximum input Because the resolution of our setup increases with a lower threshold value
224. the file is stored in memory When the file is renamed and retransmitted over the WAN the results are the same as the second transmission of this file When the content of a file is changed and it is retransmitted over the WAN the transmission time increases a little bit because only the changes need to be transmitted unoptimized From the lab results we can see that Riverbed optimizes the bandwidth even more than Ipanema This is especially noticeable with the transmission of larger files Both solutions are equivalent when looking at the devices Riverbed has more features then Ipanema to optimize the network traffic When looking at the prices for both solutions it is obvious that Riverbed is more valuable for physical equipped networks and that Ipanema is more valuable when the network consists of both physical and virtual appliances This is especially noticeable for networks with many sites When there are more than five users per site Riverbed uses a physical appliance rather than a virtual appliance V CONCLUSION In this paper we have described and compared two WOC solutions that are offered by Telindus and Belgacom to their customers to optimize WAN traffic Telindus offers WOC solutions from Riverbed to their customers and Belgacom offers WOC solutions from Ipanema to their customers Both solutions have similar features but Riverbed has some additional features that Ipanema doesn t have Riverbed achieves a higher optimiza
225. the mouth Thus we must look to other types of microphones These microphones will be positioned further away from the speaker However we expected problems with reverberation and noise This results in a decrease in SNR In order to increase the SNR there are several techniques e Sum and delay beamformer this beamformer can be used for both dereverberation 1 2 and noise cancellation 3 e Adaptive noise cancelling 2 this is done by LMS e Or a combination of the above e g Griffiths Jim Beamformer 2 3 In this paper besides investigating the microphone placement also noise reduction techniques such as mentioned above are examined for periodic and random noise This paper is organized as follows In Section 2 we give an overview of the different microphones in our acquisition system Section 3 describes the GSC The sum and delay beamformer and the adaptive algorithms will also be discussed in Section 3 because they are a part of GSC The results and experiments are reported in Section 4 Finally we conclude in Section 5 II ACQUISITION The goal of the acquisition system is to pick up human speech This is done with different types of microphones First a close talk microphone is used This microphone is placed close to the mouth Due to this small distance noise and reverberation hasn t much influence on the speech This is an advantage but the placement of the close talk microphone can sometimes be annoying
226. ther test to check that the ray doesn t end before intersecting the triangle which is still possible despite the similar test used for the bounding sphere A Plane intersection Each triangle in the list is defined by three points Let these points be called P P2 and P3 and have coordinates P X Y1521 P X3 Y2 Z2 P X3 Y3 Z3 The plane of the triangle is also defined by these three points by two vectors between these points or by a single point and the normal vector N Py Fig 3 Plane with three points two vectors and a normal Vi P P 9 V P Rh oY We find the normal vector by using the cross product N V xV 11 Before we look for an intersection we have to make sure the ray isn t parallel to the plane That would give us either an infinite amount of intersections or no intersections at all which are situations we aren t interested in The condition is N R 0 12 An implicit definition of our plane is now P x y z P N 0 13 Where P x y z is an arbitrary point By substituting this point by P t from 1 we can find the value of t P P N t 14 R N Using this value in the ray equation 1 returns the intersection point application are measured in centimeters and we used volumes B Point in triangle test of different sizes Table 1 Deviations of the ray traced intersections at 200 cm in cm Resolution 20 50 100 20 50 100 We can check if a point i
227. tion and telnet are faster because less data has to be sent In this example when shorter commands are used telnet becomes 3 32 times faster Serial communication speeds up too but only 2 32 times Serial communication needs 2 extra bits to send 1 byte Telnet doesn t need extra bits because one byte can be encapsulated in the same frame SNMP seems not to be influenced that much by command length because an SNMP get request consists of an object identifier witch contains almost the same size The benchmark shows that SNMP is faster than telnet A difference in waiting time will be an additional explanation SNMP doesn t need to wait for the prompt while telnet and serial communication have to cope with this waiting time SET wait operation Imagine a programmer must shut down an interface before another interface may come up It takes some time when the interface is up and running To make sure the interface is in the right state the programmer must wait until the previous operation is ready This execution time differs from command to command Shutting down an interface takes more time than setting the hostname In this situation when the execution time is high the choice of communication method is not that important The waiting time will be the bottleneck When the execution time is low the speed in descending order is SNMP telnet and serial communication The reason can be found in previous section Sometimes telnet will be pr
228. tion than Ipanema because it is the market leader of WAN optimization controllers Riverbed is more valuable for small networks with a few sites which are equipped with physical devices Ipanema is more valuable for networks with many sites because it can equip sites with virtual appliances much faster than Riverbed ACKNOWLEDGMENT We would like to express our gratitude to Vincent Istas Telindus for his technical support concerning Riverbed We would also like to express our gratitude to Rudy Fleerakkers Belgacom and Bart Gebruers Ipanema Techonologies for their technical support concerning Ipanema Technologies REFERENCES 1 B Ashmore Steelhead Configuration amp Tuning Riverbed Technology 2 Ipanema Technologies Autonomic Networking Features and Benefits Ipanema Technologies 2009 3 K Driscoll Network Deployment Options amp Sizing Riverbed Technology 4 K Driscoll Riverbed Steelhead Technology Overview Riverbed Technology 5 B Holmes The Riverbed Optimization System RiOS 5 5 A Technical Overview Riverbed Technology 2008 6 Ipanema Technologies Intelligent Acceleration Features and Benefits Ipanema Technologies 2009 7 Ipanema Technologies Ipanema System User Manual 5 2 Ipanema Technologies 2009 8 Riverbed Technology Riverbed Certified Solutions Professional RCSP Study Guide Riverbed Technology 2008 9 A Rolfe J Skorupa S Rea
229. tion to the problem is found The solution is tried and indeed the Apache Tomcat server starts to work again The problem is a combination of the Apache Tomcat server and the installation of the Java Virtual Machine The advantage of a Tomcat server over a Windows IIS server is that a Tomcat server can run java servlets To run these java servlets Apache has the need to access the Java Virtual Machine that is installed on the machine When the Apache Tomcat server starts he searches where to find the java directory and he searches for a specific file msvcr7l dll in this java directory This dll file isn t placed in the correct directory when the Java Virtual Machine is installed To solve this problem we simply copy this dll file in the bin directory of the Tomcat server 8 The Tomcat application now can find the right dll and starts successfully The ActiTime application works properly IV HYPER V A What is Hyper V Hyper V is a role of the Microsoft Windows server 2008 product 9 With this role virtual machines can be created and managed A virtual machine is a simulated computer inside an existing operating system This operating system runs on its own set of physical hardware An illustration of how an virtual computer works can be found in figure 4 1 and 4 2 Parent Partition Application g Company network Figure 4 1 Scheme of a normal computer Parent Partition Child Partition Application L TCP IP
230. tive f predictor Lt CCITT 88980 FIGURE IV LOWER SUBBAND ENCODER 2 Higher subband encoder The higher subband encoder produces a 16 kpbs signal It works similarly to the lower subband encoder The difference is that a 4 level adaptive quantizer is used instead of a 60 level quantizer Only two bits are assigned to the difference signal As can be seen in figure 5 the block diagram is almost identical to the lower subband decoder eH 4 level IH 16 kbit s adaptive guantizer 4 level inverse adaptive guantizer SH Adaptive R predictor H O FIGURE V HIGHER SUBBAND ENCODER CCITT 88990 C Multiplexer and demultiplexer The multiplexer at the encoder is used to combine the two encoded signals from the lower and higher subband If this is done the encoding process is completed and an output signal of 64 kpbs is generated At the decoder this signal is demultiplexed such that the lower and higher subband can be decoded V COMPARISON G 722 AND PHILIPS SUBBAND CODER When comparing the structures of G 722 and the Philips Subband Coders some similarities can be found Both codecs work with subbands In order to split the input signal into these subbands similar filters are used Both codecs use QMF filters Apart from this similarity the codecs differ greatly First of all the G 722 codec uses only 2 subbands while the Philips subband coder uses 4 or 8 subbands In the G 722 codec 75 of the available
231. to trim the inactive pages On the other hand when performing the actions on a regular desktop a lot of pages need to be trimmed to make more physical memory free which results in more unwanted processor utilization and thus a longer overall speed The above explanation indicates that the working set of the Terminal Server environment in figure 1 isn t a good representation compared to the working set of the traditional desktop it shows active and inactive pages whereas the traditional desktop output shows mostly active pages 3 Network Also important when considering performance is the network usage The output from the Network Interface Bytes Total sec is shown in figure 3 The figure indicates that there is slightly more network traffic when working with the regular desktop environment The reason for this is that the desktop has to communicate with the file servers of the company which are in the basement in the server room The virtual desktop on the Terminal Server also has to communicate with these file servers but the Terminal Server itself is also located in the server room which means the distance to cross is much smaller Also the speed of the network between the two servers 1 Gbps is greater than the speed of a link between a regular workstation and the servers in the server room 100 Mbps Figure 4 shows the output from the Network Interface Output Queue Length counter If this counter should have a sustained value
232. tracted from the target power P 2 Next path loss and extra influences lens effect and atmospheric attenuation will be taken into account These influences are calculated based on the elevation angle and range of the target These influences will be added to the value obtained in step L 3 Third the antenna gain during reception will be calculated and subtracted from the value calculated in step 2 4 Finally possible range frequency and swerling influences are calculated and subtracted from the value calculated in step 3 This will return a value in dBm which is the RCS of the target We can then use this value to predict at which locations the target will not be visible for the given radar system C SNR SNR or Signal to Noise Ratio is defined as the ratio of signal power to noise power 10 SNR depends on target power clutter and of course generated noise inside the radar system We can use SNR to predict in which areas it will be hard to locate a target or to assess radar performance Calculation As with RCS calculation we will first use the previously described parabolic fitting techniques to locate the target Afterwards we will use fast time video power range on the target s azimuth location to calculate the SNR as is shown in Figure 6 R 2 Noise Rag Fig 6 Target fast time video SNR is calculated using X 0PR SNR gt 1 where R represents range Fig 6 and P represents
233. tual PC The test environment consist of a Windows server 2000 machine with an active directory installed to and two Windows server 2008 machines Full details on setting up the test environment could be found in Analyse van een nieuwe IT infrastructuur 7 D ActiTime installation on a test machine To install the ActiTime application you have to place the installation files in the web application folder of the Apache Tomcat server The web application folder is the directory where the files needed for a website are stored and can be viewed by anyone who accesses a specific website stored on our Apache Tomcat server On our test machine the ActiTime application files are unzipped to the following directory Tomcat 6 0 webapps ActiTime The application however isn t ready to use yet To prepare the application to run correctly a few variables need to be set These variables specify which database to use such as the location of the database the username and password to access the database To specify these variables for the application a visual basic script is included in the web folder setup _mysgql vbs The variables are set and the migration of the old database data to the new database on the new server can start To insert data into a database a text file that contains SQL commands can be send to the database The following command can be used to send the sql file to the database mysql u lt username gt p lt password gt
234. twork file servers that normally cause the blockage not the RDC sessions as one would think The sessions itself dont reguire a lot of network bandwidth depending on which settings are configured for the RDC session think about themes desktop background color depth For our Terminal Server environment the network isnt likely to be a limiting factor Should it have been one then fixing this bottleneck is very easy You just have to put a faster NIC in the server or implement NIC teaming or full duplexing to double the interface of the server Just like the Processor Queue Length which indicates whether or not the processor is limiting the number of user sessions on the Terminal Server see section M D2 there is a Network Interface Output Queue Length which indicates whether or not the network is the bottleneck The output from the counter which indicates this queue length is shown in figure 8 If the value of the counter sustains more than two then action should be taken if we want more users on our Terminal Server In our testing environment with one user RDC session the counter reaches three times the value of two and when testing with four users the counter indicates a few times the value of three Because this value isnt sustained there is no problem with our network interface and therefore the network isnt the limiting factor E Results We have tested the capacity of the Terminal Server by comparing the results from o
235. ueues consist of more than one server arranged in series and or parallel Random variables are used to represent service times and the number of servers when appropriate If service 1s provided for customers in groups their size can also be a random variable The number of customers that can wait at a time in a queuing system 1s also a significant factor for consideration If the waiting room is large one can assume that for all practical purposes it is infinite All other factors regarding the rules of conduct of the queue can be pooled under this heading One of these 1s the rule followed by the server in accepting customers for service In this context rules such as first come first served FCFS last come first served LCFS and random selection for service RS are self explanatory Others such as round robin RRV and shortest processing time may need some elaboration In many situations customers in some classes have priority in service over others B Kendall Classification of Queuing Systems This identification of these elements provides a taxonomy for symbolically representing with a variety of system elements 12 The Kendall classification of queuing systems 1953 exists in several modifications The most comprehensive classification uses 6 symbols A B S q c p the first characteristic A specifies the nature of the arrival process Usually we assume that the interarrival times are
236. upport the use of Remote Wake Up but need auxiliary power to do it All necessary minimal power for the network device to receive packets can be provided by the local PCI bus 7 A second requirement is that the Wake Up Filter is programmed to match Magic Packets Note that Remote Wake Up is different from Wake On LAN WOL uses a special signal that runs across a special cable between the network device and motherboard Remote Wake Up technology uses PCI Power Management 10 4 1 1 Magic Packet A Magic Packet is a Layer 2 Ethernet II frame 11 It starts with a classic MAC header that contains destina tion and source MAC address followed by an EtherType to identify the used protocol EtherType 0x4208 is used for Magic Packets The payload starts with 6 bytes 0xFF followed by sixteen repetitions of the destination MAC address Sometimes a password is attached at the end of the payload but not many network devices support this 4 1 2 Wake Up Registers Wake up filter configuration is very vendor specific At Alcatel Lucent most simulation devices use an Intel net working device Wake Up Registers are internal registers that are mapped to PCI I O space 8 There are three important Wake Up Registers 4 1 2 1 WUC Wake Up Control register This reg ister contains the Power Management Event Enable bit and is discussed later on at PCI Power Management 4 1 2 2 WUFC Wake Up Filter Control register Bit 1 from this register e
237. uy Geeraerts for supervising the entire master thesis process Last but not least special thanks go out to the author s girlfriend brother relatives and friends who encouraged and supported the author during writing of this work REFERENCES 1 S Muller Upgrading and repairing pcs 15th ed Que Pearson tech group 2004 2 Intel Corporation Intel 82801EB ICH5 Datasheet 252516 001 Available at intel com 2003 3 Intel Corporation Intel ICH9 Datasheet Available at intel com 2008 4 T Shanley D Anderson PCI System Architecture Addison Wesley Developer s Press ISBN 0 201 30974 2 1999 5 Hewlett Packard Intel Microsoft Phoenix Toshiba Advanced Configuration and Power Interface Specification ed 3 0B Available at acpi info 2006 6 Intel Corporation Intel 64 and IA 32 Architectures Software Devel opers Manual vol 3B Catalog nr 253669 032US Available at intel com 2009 7 PCI Special Interest Group PCI Local Bus Specification rev 2 2 Available at pcisig com 1998 8 Intel Corporation PCle GbE Controllers Open Source Software De velopers Manual rev 1 9 Catalog nr 316080 010 Available at intel com 2008 9 Intel Corporation ACPI Component Architecture Programmer Refer ence rev 1 25 Available at acpi info 2009 10 PCI Special Interests Group PCI Bus Power Management Interface Specification rev 1 2 Available at pcisig com 2004 11 Lieberman Software Corporation White Paper Wak
238. verlee KU leuven ESAT 1995 pp 7 14 6 B Van Veen and K Buckley Beamforming A versatile approach to spatial filtering ASSP Magazine July 1988 pp 17 19 7 Kuo Sen M Real time digital signal processing implementations and applications gu ed Bob H Lee Wenshun Tian Chichester John Wiley amp Sons Ltd 2006 ch 7 8 Paulo S R Diniz Adaptive filtering algorithms and practical implementation 3 ed New York Springer 2008 ch 5 9 LA McCowan Robust Speech Recognition using Microphone Arrays Ph D Thesis Queensland University of Technology Australia 2001 pp 15 22 10 M Moonen S Doclo Speech and Audio processing Topic 2 Microphone array processing KU Leuven ESAT 11 S Doclo Multi microphone noise reduction and dereverberation techniques for speech applications Ph D thesis 2003 12 I McCowan D Moore J Dines D Flynn P Wellner H Bourlard On the Use of Information Speech Recognition Evaluation IDIAP Research Institute Switzerland pp 2 Power Management for Router Simulation Devices Jan Smets Industrial and Biosciences Katholieke Hogeschool Kempen GEEL Belgium Abstract Alcatel Lucent uses relatively cheap Intel based computers to simulate their Service Router operating system This is a VxWorks based operating system that is mainly used on embedded hardware devices It has no power management features Traditional computers have support for power management u
239. we will set it at exactly 2 times the maximum input The second parameter that we control that has an influence on the resolution is the sample rate max pulse rate which is the same for our setup Devices for the DIN43864 standard require to send pulses of minimum 30ms This comes down to a sample rate of 33 33Hz For the ADC setup we use the internal reference voltage of 1 2V as reference this gives us an input range from 0 6V to 0 6V Setting the ADC to unipolar mode and as the max output of the sensor is 117 75mV we set the internal amplifier to a gain of 4 The resulting input range is OV to 150mV The conversion formula for the ADC is Vin Vrneg SD16MEMO 65536 Vrpos Vrneg With Vrpos 150mV and Vrneg OV If we insert Vin 117 75mV in the formula above we get SDIOMEMO 51446 resulting in a threshold value of 102892 Resulting in that 1 pulse resembles 1500w m for 60ms or 0 025Wh m Because the MSP430F2013 does not provide us with a high Impedance buffer at the input we are required to implement it ourselves since this is required for the sensor For this we use an opamp circuit with its gain set to 1 At the output we use an optocoupler at the output of the circuit that we control with the output of the microcontroller This is done to limit the current drawn from the microcontroller output and to be able to use larger voltages for the pulse output since the DIN43864 standards gives a voltage range from 0 to 28
240. with a limited bandwidth of 3 5 kHz and 7 kHz This is the original signal sent through low pass filters In the first phase 11 different audio signals are encoded with three different configurations for the three modes These configurations can be found in table 2 So in this phase the listener is presented six signals to evaluate This test is done for each mode except for mono TABLE II SBC CONFIGURATION PARAMETERS __ subbands block size bitpool bitrate a he as le Joint 1 Joint 3 samen je ie ls lie 46 fie jpo stereos 8 fs 9 hi Dull j4 ie 8 HO Dul2 js fie 8 B Then the best configuration is selected for each mode and a new test is done Now the listener is presented seven signals to evaluate because now a mono configuration is also included The listener has to grade each signal between O and 5 unacceptable to perfect The grading allows steps of 0 1 so that enough scores are available Because the development boards are made for testing purposes some noise is introduced to the audio output of these boards Also the cables connecting the boards to the PC introduce noise Therefore it was decided to generate the audio signals using a software encoder and decoder on a computer This way no additional noise can occur and more accurate results are acquired The noise introduced made it too easy to differentiate the original form the coded samples VIII RESULTS A Phase 1 Table 3 gives the
241. works 9 Tembria server monitor 10 WebWatchBot B Setting up the standard environment 3 Monitor tool set up with the capability to monitor the previous listed services and events with a scan The environment consists of one small business server where the services will be running and a monitor server with the appropriate tool for the benchmark These two servers will be connected with a Cisco 1841 router for a stable network Both systems run virtually VM Ware on two different physical systems with the following specifications TESTSERVER S MONITOR SERVER monitor server tool frequency of 5 minutes During the 30 minutes test process WireShark will monitor the network load of the specific tool under test First of all the tools both run as a service on the monitor server and follow a previous defined procedure therefore we can compare them equally time service that will go down BlackBerry Dispatcher Disabled 4 Fig 1 Standard testing environment Print Spooler testserver MSExchangeSA MSExchangelS t min 8 min 15 min 18 min 22 min 4min Smin 15min BlackBerry Server Alert 18 min 22min AMD Athlon XP 2500 Intel Core 2 Duo 2 4ghz 384MB RAM 512MB RAM Table 2 Standard testing environment Remark SolarWinds is a tool that does not follow up the standard environment because it only runs on a dedicated server environment Therefore the tool will be installed on a virtual VM War
242. y to communicate with each other It was made possible with the efforts of Alexander Graham Bell and Thomas Watson In 1907 Lee De Forest made a revolutionary breakthrough by inventing the three way vacuum tube This allowed an amplification of signals both telegraphic and voice By the end of 1991 the generation of mobile phones was introduced to the world This made mobile communication over the still developing telephone network also known as Public Switched Telephone Network PSTN possible The next couple of years the problem of poor coverage and ensuring good quality of voice communications kept growing and are nowadays the major causes of business customer churn churn the process of losing customers to other companies since switching providers is done with the utmost ease Network designers need to be able to make a choice to resolve this specific problem The two major solutions are the use of picocells or WAP s with implementing the VoIP protocol Firstly most network designers make a site survey This step will ensure that the designer comprehends the specific radio frequency RF behavior discovers RF coverage areas and checks for objects that will have a certain RF interference Based on this data he can make appropriate choices for the placements of the devices Also very important is to know the advantages and disadvantages of both options so that in some cases the cost of making a site survey can be eliminated for the

Content - KU Leuven

Contents

Download Pdf Manuals

Related Search

Related Contents