QRN Quantum Random Number Generator
Contents
1. ap IN PHOTON DEVICES MICRO PHOTON DEVICES QRN Quantum Random Number Generator Micro Photon Devices S r l Via Stradivari 4 39100 Bolzano BZ Italy email info micro photon devices com Phone 39 0471 051212 e Fax 39 0471 501524 ORN User Manual Version 3 1 0 November 2013 Micro Photon Devices S r l Italy All Rights Reserved QRN Quantum Random Number Generator x IM PHOTON DEVICES Contents CONTEN icona ere nigi dii ee ee ee ere eee 1 AR ace a Or OPETI 0 Immpnen rears ee rns ener REE ree Cn ene ee ere eee ree eee eer nee ern 2 RYAN ia 2 IGM CUE TIO Misano iaia NPI 2 Structure of the generator Lrrrnci 4 Empe ddcd cecuri iy EST illa 6 Qualilication or ON lara 6 BIDROG DIY eee ii nino 9 ORN MTE CSS apr aa 10 ORN OPeAlO lane 11 EISGEFICONENAraCESrISHO Soia RA RI 12 VO 5 AAG NE Rei 12 RI SS ile i i 13 POWCPCONSUMP Olga aaa 13 QRN Software interface VisualQRN Windows only cccceccceccceeceenceceeceeeceeeeeeeeceeeseeeeceeeseeeeeeeeseeeeeeeeeees 14 Stallone E ES A E i a ii 14 SEMINESARTOnNt dia a na 14 SHANE nr 15 Random stream Penetron N 17 ORN DK ini 19 System reg iremMeEn alain 19 ORIN mechanical dimensionali 20 CONVFEntend d CRM 20 Micro Photon Devices S r l Italy All Rights Reserved Page 1 QRN Quantum Random Number Generator x IM PHOTON DEVICES Theory of operation Summary The Quantum Random Number Generator QRNG designed by Micro Photon Devices MPD sh2. 0 8915 0 0920 0 2119 0 0421 0 6225 0 6221 0 6560 0 9465 0 4781 diehard_operm5 0 6092 0 7151 0 9232 0 3270 0 3980 0 4632 0 9970 0 9085 0 5261 0 4593 0 6327 diehard_rank_32x32 0 7586 0 0148 0 3911 0 2398 0 0673 0 4608 0 4535 0 9969 0 3683 0 2394 0 3991 diehard_rank_6x8 0 2750 0 4677 0 3459 0 9769 0 4058 0 9870 0 5612 0 8559 0 8352 0 1099 0 5821 diehard_bitstream 0 0664 0 6368 0 1847 0 8392 0 3445 0 3888 0 3830 0 3433 0 9891 0 4336 0 4609 diehard_opso 0 4493 0 9904 0 2566 0 7402 0 8743 0 4962 0 3230 0 9607 0 9712 0 0681 0 6130 diehard_oqso 0 1313 0 6521 0 9631 0 9070 0 9346 0 5601 0 5631 0 0604 0 7495 0 8364 0 6358 diehard_dna 0 1121 0 4518 0 3289 0 1370 0 0001 0 3805 0 1516 0 7131 0 3625 0 9875 0 3625 diehard_count_1s_str 0 9560 0 7377 0 2618 0 8614 0 2909 0 4300 0 0870 0 9520 0 2478 0 9515 0 5776 diehard_count_1s_byt 0 2358 0 7591 0 7140 0 1464 0 4822 0 4741 0 1418 0 4085 0 0355 0 0617 0 3459 diehard_parking_lot 0 3727 0 1671 0 9570 0 5313 0 6291 0 9717 0 3695 0 3190 0 7734 0 2540 0 5845 diehard_2dsphere 0 7778 0 8088 0 4922 0 9094 0 8781 0 1495 0 7450 0 7547 0 9576 0 8871 0 7360 diehard_3dsphere 0 6588 0 9591 0 5813 0 9995 0 9444 0 1866 0 2974 0 4312 0 8888 0 4421 0 6389 diehard_squeeze 0 5869 0 4116 0 7330 0 7908 0 6881 0 2039 0 7038 0 2007 0 2033 0 7529 0 5275 diehard_sums 0 0097 0 0717 0 0359 0 2323 0 0091 0 0053 0 3168 0 1191 0 6860 0 0201 0 1506 marsaglia_tsang_gcd 0 5372 0 9716 0 5264 0 1508 0 1409 0 0160 0 4270 0 4694 0 9310 0 7774 0 4948 marsaglia_
3. 3 QRN Quantum Random Number Generator x IM PHOTON DEVICES Structure of the generator As already mentioned above the photon detector exploited by the QRN is a SPAD 17 18 A SPAD is a diode that is reverse biased well above its breakdown voltage Under this operating condition a single event of quantum nature namely the absorption of a single photon causes the generation of an electron hole pair which is accelerated by the high electric field across the junction The energy of the free carriers is sufficient to trigger a macroscopic avalanche current of few milliamperes through the device Proper electronics senses and signals the event The use of a SPAD as a detector has several advantages Firstly an event can be triggered both by the internal thermal generation process and the external photons A second advantage is that no analogue measurement of voltage or current is needed since the detector acts as a digital Geiger counter It follows that no read out noise is added to the measurement process and no electrical noise affects the ignition of the events hence the randomness The QRN is based on an array of 1024 independent SPAD detectors organized as a 32x32 matrix fabricated in standard CMOS technology such as the one described in 9 As shown in Figure 2 each SPAD is controlled by a front end electronics 10 which provides the correct bias voltage above breakdown senses the avalanche curre
4. OUT before the VALID status has reached the high level If no SYNC IN is provided and no error occurred the VALID signal will remain high until the device enters sleep mode i e after about 5 seconds without any rising edge of the SYNC IN Micro Photon Devices S r l Italy All Rights Reserved Page 11 QRN Quantum Random Number Generator x IM PHOTON DEVICES Electrical Characteristics Voltages and timings The logic values and the timing properties of SYNC IN and STREAM OUT are described in Table 3 Typical propagation delays of about 14 ns from the rising edge of SYNC IN and the generation of a new random bit at STREAM OUT are expected The VALID output has an asynchronous response to SYNC IN In fact it is generated by the internal control electronics Whenever an error occurs the SYNC IN input is disabled and the STREAM OUT is set to 0 Table 3 Electrical Characteristics of the QRN measured at room temperature 25 C Symbol Parameter Signal Grade Min Typ NEVE Unit Vin Input logic high voltage 5V tolerant SYNC IN All 2 4 3 3 V Vit Input logic low voltage SYNC IN All 0 0 8 V STREAM OUT VALID All 29 V STREAM OUT VALID Von Output logic high voltage 500 load Vor Output logic low voltage 500 load All V 200Mb s Minimum time between two 120Mb s ae synchronization voltage pulses SYNC IN 64 Mb s 32 Mb s 16 Mb s Minimum duration of a synchronization voltage pulse tpp Propagation delay STREAM OU
5. according to a Poisson law with mean k p k 256 This property is usually named law of rare events When the product k p becomes very large i e k p gt 100 the Poisson distribution converges in probability to a discrete Gauss distribution of mean and variance equal to A k 256 according to the central limit theorem The distribution of each random variable n is thus the following 2 k 3 nizza n 1 id 1 AR n e e V2mA 2rk 2 256 We define now a set of random variables h which represents the normalized deviation of n from the model distribution hi ni 53 3 After a variable change in eq 2 one can easily show that the h follow a standard Gauss distribution independently on k _ hp e 2 4 1 G h Nora Micro Photon Devices S r l Italy All Rights Reserved Page 16 QRN Quantum Random Number Generator x IM PHOTON DEVICES This distribution is the same for each 8 bit number We can therefore extract 256 values which are distributed according to eq 4 from the stream These values are calculated as follows 1 The n are calculated from a stream of k bytes e g how often the number 25 appears in the stream 2 Eq 3 is applied to each n to calculate the h 3 A frequency distribution of the h is then constructed to inspect visually the match between eq 4 and the data obtained from the stream Figure 5 Right In conclusion a random stream s
6. up to 17 10 events per second are available to generate the random stream with the MPD s QRN Electronic interface PO LED DRIVER USB CONTR HW OUT EO SERIALIZER QUENCH and Detected CONTROL LL Enable PSEUDO a ee CONTROL anc LFSR COUNTER gt UNIT ae Aen SELECTOR DATA BUS SPAD matrix cell FPGA Computer Figure 2 Scheme of a single cell of the 32x32 CMOS array Each independent cell contains a SPAD detector a front end electronics and a Linear Feedback Shift Register LFSR for counting the quantum events A FPGA controls the read out of the chip and the USB interface to transmit the random binary stream to the computer Micro Photon Devices S r l Italy All Rights Reserved Page 5 QRN Quantum Random Number Generator x IM PHOTON DEVICES Embedded security test The QRN performs a continuous test of the random bit stream before providing the data to the user A failure of the generator or an active attack to the system is immediately recognized and the generator enters an idle state It is sufficient to reset the device to bring it back to the normal operating state A simple frequency test is embedded in the device electronics which looks for extremely rare events as the generation of 48 consecutive 0 or 1 One of these events might occur for example once every 7 years for a device which generates data 24 per day at a rate of 64Mbit s Although possible this event is extremely
7. 724 0 4970 0 4272 0 4666 0 5232 0 6194 0 5574 0 5212 Micro Photon Devices S r l Italy All Rights Reserved Page 7 QRN Quantum Random Number Generator x IM PHOTON DEVICES The second test suite which was applied to the QRNG was Big Crush from TestU01 This suite uses a much larger number of random data compared to Dieharder in order to remove suspicious p values i e too close to 0 or 1 which are due to less probable although not impossible data streams and not to failures of the generator under test As shown in Table 2 our QRNG successfully passed all tests of TestU01 Table 2 Results from the execution of the Big Crush test suite TestU01 The suite was executed twice All the tests were passed in both cases Test Result Test Result SerialOver PASSED WeightDistrib PASSED CollisionOver PASSED SumCollector PASSED BirthdaySpacings PASSED MatrixRank PASSED ClosePairs PASSED Savir2 PASSED SimpPoker PASSED GCD PASSED CouponCollector PASSED RandomWalk1 PASSED Gap PASSED LinearComp PASSED Run sknuth PASSED LempelZiv PASSED Permutation PASSED Fourier3 PASSED CollisionPermut PASSED LongestHeadRun PASSED MaxOft PASSED PeriodsInStrings PASSED SampleProd PASSED HammingWeight2 PASSED SampleMean PASSED HammingCorr PASSED SampleCorr PASSED HammingIndep PASSED AppearanceSpacings PASSED Run sstrings PASSED Micro Photon Devices S r l Italy All Rights Reserved Page 8 QRN Quantum Random Number Generator x IM PHOTON DE
8. T SYNC IN tmin 200Mb s 128Mb s STREAM OUT 64 Mb s 32 Mb s 16 Mb s Minimum duration of a logic output t i value tor tof Output rising and falling times 10 90 STREAM OUT tir tif Input rising and falling times 10 90 SYNC IN SYNC IN BITSTREAM OUT 50 V 10 Figure 4 Timing properties of the hardware interface Micro Photon Devices S r l Italy All Rights Reserved Page 12 QRN Quantum Random Number Generator MPP QRN status Two LEDs on the front panel of the device provide the information on the actual state of the QRN Table 4 explains the possible states Table 4 Coding for the two multi colour LED on the front panel LED 1 LED 2 Message Device ON and idle If blinking the QRN is in sleep mode it will wake as soon as O data are requested either by USB if in software mode or through SYNC IN if in hardware mode O O Reading data O O Hardware error Code 1 Contact Micro Photon Devices for assistance O O USB power error change USB port or use an externally powered USB hub Hardware error Code 2 Contact Micro Photon Devices for assistance In hardware mode QNN stops working In software mode and If security mode is enabled random bits are still provided but they come exclusively from the internal PRNG To indicate this event this LED starts blinking O Hardware mode the bit stream is read too fast for the device Decrease the frequency of the voltage pulses a
9. VICES Bibliography 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 S Cova M Ghioni A Lacaita C Samori and F Zappa 1996 Appl Opt 35 1956 N Gisin G Ribordy W Tittel and H Zbinden 2002 Rev Mod Phys 74 145 Boyle P P 1977 J Fin Econ 4 23 G Winkler 2002 Image Analysis Random Fields and Markov Chain Monte Carlo Methods A Mathematical Introduction Berlin Springer I Marcikic H de Riedmatten W Tittel H Zbinden M Legr and N Gisin 2004 Phys Rev Lett 93 180502 T Saito K Ishii Tatsuno S Sukagawa and T Yanagita 2010 Randomness and Genuine Random Number Generator With Self testing Functions Joint International Conference on Supercomputing Tokyo edited by K Todani and T Takeda C R S Williams J C Salevan X Li R Roy and T E Murphy 2010 Opt Express 18 23584 B Sunar W J Martin and D R Stinson 2007 IEEE Trans Com 56 109 119 S Tisa A Tosi and F Zappa 2007 Opt Express 15 2873 S Tisa F Guerrieri and F Zappa 2008 Opt Express 16 2232 G Marsaglia and A Zaman 1993 Comp Math Appl 26 1 Brown R G 2012 DieHarder A Gnu Public License Random Number Tester Duke University Physics Department Durham NC 27708 0305 P L Ecuyer and R Simard 2007 ACM Trans Math Softw 33 22 Rukhin J Soto J Nechvat
10. al M Smid E Barker S Leigh M Levenson M Vangel D Banks A Heckert J Dray and S Vo 2010 A Statistical Test Suite for Random and Pseudorandom Number Generators for Cryptographic Applications Special Publication 800 22 Revision 1 National Institute of Standards and Technology Gaithersburg G Marsaglia 1995 The marsaglia random number cdrom including the diehard battery of tests of randomness http www phy duke edu rgb General dieharder php M Ghioni A Gulinatti I Rech F Zappa and S Cova 2007 IEEE J Select Topics Quantum Electron 13 852 S Cova M Ghioni A Lacaita C Samori and F Zappa 1996 Appl Opt 35 1956 Micro Photon Devices S r l Italy All Rights Reserved Page 9 QRN Quantum Random Number Generator x IM PHOTON DEVICES QRN interfaces Front view Back view STREAM OUT VALID SYNC IN O O O Figure 3 Quantum random number generator front and back views USB STREAM OUT SYNC IN VALID High speed USB 2 0 connector Used both for transferring data to the PC and to power the module The device requires less than 500 mA during operation It is recommended to use short USB cables lt 2 m SMA output able to drive a 50 Ohm terminated transmission line Electric pulses of 3 3V LVCMOS are generated at this output Low Voi and high Voy voltage levels correspond to the logic values 0 and 1 This output is not enabled when the QRN i
11. are mode in this way the data are transferred to the computer and the hardware input and outputs are disabled When in hardware mode the QRN s random bits are clocked out by providing a proper signal at the SYNC IN input The SYNC IN signal can be periodic or aperiodic but the proper timings described in paragraph Electrical Characteristics must be always ensured The SYNC IN has also a maximum allowed frequency which is determined by the QRN grade and is shown also in Table 3 If a frequency higher than the maximum one is provided the QRN will signal this ERROR condition by changing the colour of the front LEDs according to Table 4 and by stopping any generation of random bits In order to enable again the generation of the random bits the frequency at SYNC IN must be decreased below the maximum one and the QRN reset by unplugging the USB cable It s thus up to the user to acquire data from STREAM OUT synchronously with SYNC IN It must be also stressed that acquiring data from STREAM OUT without providing SYNC IN or not respecting proper timings could result in a not random stream see next paragraph The cable length i e the propagation delays of the three connected signals MUST ALWAYS be considered since Figure 4 timings are measured at the QRN input outputs with no cables attached During the hardware mode operation even the VALID status must always be checked As already described when VALID s status is high the random bits are va
12. ed stored in a retrieval system or translated into any language in any form or by any means except for the documentation kept by the purchaser for backup purposes without the express written permission of Micro Photon Devices S r l All trademarks mentioned herein are property of their respective companies Micro Photon Devices S r l reserves the right to modify or change the design and the specifications the products described in this document without notice Micro Photon Devices S r l Italy All Rights Reserved Page 20
13. esired installer 32 or 64 bit and follow the guided installation procedure During the installation a command window will appear for device driver installation Wait for a command line window to be opened and wait for the following sentence to appear The driver was successfully installed message appears Then press Enter to complete the installation Settings and operation There are only two major settings for the device Enable or disable the XOR between the QRN stream and a pseudo random number generator of the KISS family This is useful in case Security mode a complete hardware failure of the QRN occurs random data are Disabled Enabled still provided by the KISS PRNG enabling the user to substitute the failing device with minimum out of service time This event is signalled with a clear message in VisualQRN window and with the blinking of LED1 on the QRN Select the working mode for the device Software the data are transferred to the computer via the API functions Hardware the Working mode input and output SMA connectors are enabled The random sla cn stream is transferred to an external hardware Upon connection to the USB QRN starts in hardware mode so it possible to use it without a PC Three different applications are included Stream Analyser It features a basic check of the random bit stream Random stream generator Allows to save a random data stream on the hard disk 3 QRN server optional A bas
14. he cipher must guarantee a broad space in which the key is selected and the probability to choose a specific key must be uniformly distributed between all possible values The latter requirement is often achieved by selecting the key using randomly generated numbers Ideally a random number generator provides unbiased and unpredictable data Accordingly the output of the generator is completely independent from the sequence of previously generated numbers Three major families of Random Number Generators RNG exist Pseudo RNG PRNG Chaotic RNG CRNG and Quantum RNG QRNG Pseudo RNGs produce sequences which look like random but are in fact generated by deterministic algorithms Although PRNG are very fast and cost effective they require a seed sequence to initialize the generator state The outcomes are then completely predictable periodical and seed dependent CRNG are instead based on chaotic physical processes complex systems in which a small variation of the initial conditions produces large changes of some observables Typically the CRNGs use thermal noise 6 optical fluctuations of laser radiation 7 jitter of oscillators in integrated circuits 8 just to cite a few High quality random streams are obtained but the generators themself are not intrinsically random In addition an attacker could get access to the chaotic system and simultaneously measure the same physical observables and reproduce the random data Conversely Quan
15. hould not show any pattern on both axis and the constructed frequency distributions should match the limit one Random stream generator With this panel it is possible to save a random data stream on the hard disk By selecting first the desired amount of megabytes and then pressing the Run button the stream is saved on the hard disk in binary format Alternatively the generation rate can be tested Please note that MPD use the following unit of measure 1MB 1 000 000 bytes WARNING The generation rate can be accurately measured only for large data streams due to the automatic data buffering in the computer RAM Measured generation rate will always be a little lower than the QRN grade due to the small amount of time needed by the PC to send the commands to the QRN Save a random stream on the disk Calculate the generation rate Size of the random data stream 10MB Size of the random data stream 1 MB 0 Bit s Figure 6 Left Generate and save a random data stream on the hard disk Right Calculate the random bit generation rate for a target device Some of the most popular battery of tests for random number generators can be then applied to the obtained streams These test suites have been created to recognize deviations of random data streams from true randomness Some statistical test suites for random numbers are 1 Test U01 http www iro umontreal ca Ysimardr testu01 tu01 html 2 NIST http cs
16. ic multithreaded TCP IP server to share random numbers over the network Micro Photon Devices S r l Italy All Rights Reserved Page 14 QRN Quantum Random Number Generator x IM PHOTON DEVICES Stream Analyser With this panel shown in Figure 5 a basic check of the random bit stream is performed in real time The size of the random data stream i e the number of bytes used for the processing is equal to 1MB 1 000 000 bytes Size of the random data stream uu gt Device QRN 2 02 SDK 1 01 Distibution of 8 bit words Deviation from the ideal distribution E 2 2 2 2 hl a a Deviation a u Figure 5 Left Frequency distribution of 8 bit words calculated from a 1 MB stream Right Frequency distribution of the deviations between model and measured data The red curves are the model distributions which are expected for perfectly random data streams The blue curves o are the frequency distributions obtained from a 1MB of random data acquired by the QRN The stream is divided in bytes which represent integer numbers between O and 255 If the stream is perfectly random the occurrence frequency of each 8 bit number is the same The frequency distribution of the 8 bit words F is constructed and it should converge in law to a discrete uniform distribution emilia 1 i lim i lim pee TRS where k is the number of used bytes and the n i 0 255 count the nu
17. improbable Qualification of the QRNG In order to test the quality of the random binary stream generated by the QRNG two of the most popular test suites for random number generators Dieharder 12 and TestUO1 13 were modified to support the QRN The C libraries which are used to control the generator were directly integrated in the source code of both test suites The reason was to prove the long term stability of the QRN in generating random numbers at high rates and at frequent request of new data streams The original test suites were also applied to random data stream from the QRN previously stored on the hard drive No differences on the results of the tests were observed The first applied test suite has been Dieharder Dieharder is an extension of the Diehard Battery of Tests of Randomness 15 a very well known random number test suite developed by G Marsaglia In particular Dieharder was extended with specific tests to check for bit level randomness of sequences produced by physical RNG Monobit Runs and Serial Furthermore it incorporates tests from the Statistical Test Suite STS developed by the National Institute for Standards and Technology NIST 14 Table 1 shows the results of 10 complete and independent executions of the Dieharder test suite In most cases the obtained p values are above 0 01 and below 0 99 and only few exceptions are obtained outside these boundaries producing a warning message However the nu
18. lid but when it is low they are not As a consequence the user should not sample the STREAM OUT status whenever the VALID status is low The following two examples will show how to properly use the VALID output Example 1 immediately after the power up the user must wait for the VALID status to go high before clocking out the random bits Example 2 as previously described when a frequency higher the maximum allowed one is applied to the SYNC IN the QRN not only stops the generation of random bits and signals this condition through the front LEDs colour but also reduces the VALID status to the low state in order to exit this condition the frequency must be reduced and the QRN restarted User must also check VALID signal before acquiring data from STREAM OUT The QRN module enters a deep sleep mode whenever it is not used for a while independently of being in hardware mode or in software mode If the QRN is in software mode the exit form the deep sleep mode is straightforward and is correctly handled by the QRN API or by the Visual QRN If the QRN is in Hardware mode the VALID signal goes low as soon as the deep sleep starts In order to get the QRN out of the deep sleep mode a sync signal must be sent to the QRN The exit from the sleep state requires some time during which the generator is restored to operating conditions At the end of this period the VALID signal goes high again of course no random bit should be sampled at the STREAM
19. mber of occurrences of each number i in the stream The n are random variables because they assume different values depending on the analysed random stream The frequency F data and the theoretical distributions F model are then plotted on the left axis of the windows form Figure 5 Left Micro Photon Devices S r l Italy All Rights Reserved Page 15 QRN Quantum Random Number Generator x IM PHOTON DEVICES These two distributions do not match perfectly In fact the frequency and uniform distributions are equivalent only in the limit case k gt The frequency distribution will be scattered around the uniform distribution and the magnitude of this deviation depends on k A second frequency distribution of the deviation between the two distributions is then constructed We can make two assumptions which are valid for an ideal random stream 1 Then are independent and identically distributed random variables 2 Then are distributed according to a Poisson distribution of mean k 256 In fact each 8 bit word has the same occurrence probability for an ideal random stream of bytes The second assumption is derived by the properties of a Bernoulli process The occurrence of a number in a stream of k bytes can be seen as k repetitions of independent Bernoulli trials of mean p 1 256 Considering that p is small and k is large i e p lt 0 01 and k gt 100 the occurrence of a number i in the stream is distributed
20. mber of suspect p values is within the expected statistical failure rate assuming a 1 significance level In addition the p values are well uniformly distributed within O and 1 The mean values calculated over the single test types and executions are sufficiently close to 0 5 The only exception to this is the diehard_sums test that seems to have some unsolved bugs and it is thus considered not reliable by the developer himself see 16 As a comparison cryptographically secure random number generators as the Advanced Encryption Scheme AES and the XOR between the QRN and KISS PRNG pass the test with similar p value distributions Micro Photon Devices S r l Italy All Rights Reserved Page 6 QRN Quantum Random Number Generator MPD MICRO PHOTON DEVICES Table 1 Summary of the results of ten executions of the statistical test suite Dieharder The p values are listed for each test family Several tests are repeated more than once in the test suite with different parameters Thus only the average p value is reported diehard_sums generates p values which are not uniformly distributed Similar p value distributions were obtained by executing the test suite with AES and making the XOR between the QRNG output and the KISS PRNG data not shown The results suggest that diehard_sums is not a reliable test as already stated in the Dieharder documentation 12 Test 1 2 3 4 5 6 7 8 9 10 Mean diehard_birthdays 0 3076 0 3885
21. nt and quenches it After ignition the SPAD is then reset back to operation after a so called dead time A custom 8 bit Linear Feedback Shift Register LFSR counts the events and provides them to the off chip electronics The LFSR can process a large number of quantum triggering events e g hundreds before producing the corresponding output bits A FPGA provides the control signals to the chip It sets the read out rate of the bits and reads the counter values In addition it implements a basic Pseudo Random Number Generator PRNG KISS Keep It Simple Stupid 11 which is used to enhance the security level of the QRN This generator is initialized once during the start up phase of the device by using a quantum random stream and it runs indefinitely When the security option is activated the XOR between the QRN stream and the PRNG is performed The output stream remains random even in case of a failure of the QRN It s worth noting that the PRNG and the security option were never active during the validation tests of the MPD s QRN As shown in Figure 2 the random stream is then processed inside the FPGA by a bit selector which removes some of the bits from the stream according with their position in the LFSR counter The bit selector is adjusted depending on the average number of quantum events detected by each cell of the array in order to obtain an equal probability of obtaining zero and ones Variable bit rates from 16 Mbit s to 200 Mbi
22. own in Figure 1 and named QRN is a physical generator of random numbers based on the intrinsic randomness of quantum optical processes The device exploits the stochastic number of detected photons within a fix observation time in order to generate a random bit stream The detection and processing of the photons is performed by a CMOS chip which includes 1024 independent Single Photon Avalanche Diodes SPAD 1 and achieves random bit generation rates of 16 32 64 128 and 200 Mbit s depending on the device grade MPD has verified the absolute randomness of the generated random data with standard statistical test suites and more stringent correlation and bias tests In all cases the data generated by the QRN passed the applied tests Introduction Different applications require the generation of high quality random number sequences such as Cryptography Monte Carlo numerical simulations and image processing 2 3 4 5 just to mention a few Cryptography for example aims to keep messages safe A private message sent via physical media in form of electromagnetic waves or electrical signals must be read only by the recipient it is addressed to The degree of security of the message depends on the cipher and the key used for encoding Figure 1 MPD Quantum Random Number Generator Micro Photon Devices S r l Italy All Rights Reserved Page 2 QRN Quantum Random Number Generator x IM PHOTON DEVICES In the optimal case t
23. rc nist gov groups ST toolkit rng stats_tests html Micro Photon Devices S r l Italy All Rights Reserved Page 17 QRN Quantum Random Number Generator x IM PHOTON DEVICES 3 Diehard http stat fsu edu pub diehard 4 Dieharder http www phy duke edu rgb General dieharder php The Micro Photon Devices QRN passed all the above listed battery of tests QRN server Optional With this panel shown in Figure 7 it is possible to start a basic multithreaded TCP IP server to share random numbers over the network Please contact sales micro photon devices com for pricing and availability This simple server application can be used to provide random numbers to several computers connected to a network TCP IP is a standard communication protocol which is supported by most programming languages A fast way to test the QRN from different computers or operating systems is to run QRN Server on a windows machine and to download on the local computers the random data streams Port Select a free port on the host computer Start Stop Allow Stop connections to the server Select a port to start the server Server listening on port 6666 ID 12 Started Connection closed 27778196 Figure 7 Simple server interface to share the random data stream over the network Micro Photon Devices S r l Italy All Rights Reserved Page 18 QRN Quantum Random Number Generator MPD ORN SDK QRN is provided with a Soft
24. s set to software mode This SMA input requires a 3 3V LVCMOS pulse V p 5V tolerated The input impedance is 50 Ohm DC This input accepts a periodic clock signal A new random bit is generated at STREAM OUT on the rising edge of the signal This output is not enabled when the QRN is set to software mode SMA output able to drive a 50 Ohm terminated transmission line Electric signals of 3 3V LVCMOS are generated at this output This signal is active high When bits from the STREAM OUT output are valid this signal goes high During start up and device sleep or if an error occurs this signal is low This output is not enabled when the QRN is set to software mode Micro Photon Devices S r l Italy All Rights Reserved Page 10 QRN Quantum Random Number Generator x IM PHOTON DEVICES QRN operation In order to power on the QRN just connect it to a PC or to a USB wall mount power adaptor Once the QRN module is connected 5 to 10 seconds are required before the internal firmware is fully functional When the QRN is powered on it starts always in hardware mode the input and output SMA connectors are enabled and the random stream is transferred to an external hardware by means of attached cables This means that it is possible to use the QRN without connecting it to a PC By using the VisualQRN PC software see the next paragraphs or the supplied APIs it is possible to change the working mode of the QRN to softw
25. t s of random data are obtained after this processing depending on the device grade Micro Photon Devices S r l Italy All Rights Reserved Page 4 QRN Quantum Random Number Generator x IM PHOTON DEVICES Finally the stream is sent to a FIFO register for buffering the random bits and maximizing the transfer rate to the user PC By choice the FIFO data can be directed to a serial hardware output HW OUT which provides 3 3 V LVCMOS logic values for the random bits In order to provide a high quantum event generation rate the chip i e the SPAD detectors is illuminated by means of a light emitting diode driven by a current generator controlled via a digital filter The filter monitors the total number of detection events during several read out operations and adjusts the LED current to keep the mean detection rate constant In this way the QRN always operates under optimal conditions independently of variations of physical parameters like LED efficiency array sensitivity warm up at power on aging etc This controller is not critical for the random bit generation process but it is necessary to assure the performance and the stability of the device over long operating times In case of failure of the illumination source the system stops and generates an error message to the user Since the typical dead time of the SPAD front end is 60 ns a maximum number of 16 6 10 photons per second and per SPAD can be acquired It follows that
26. t the SYNC IN input O O Frequency test error Too long sequence of bits equal to 0 or 1 O O Hardware error Code 3 Contact Micro Photon Devices for assistance Power consumption QRN is fully USB powered and does not require any additional power supply unit The amount of power depends on the device grade and is shown in Table 5 If no data are requested for longer than 5 seconds the device enters the idle state Table 5 QRN power consumption as the function of the grade and state Grade Mb s Idle mW Running mW 16 850 1100 32 850 1100 64 850 1100 128 850 1200 200 850 1300 Micro Photon Devices S r l Italy All Rights Reserved Page 13 QRN Quantum Random Number Generator x TM PHOTON DEVICES QRN Software interface VisualQRN Windows only VisualQRN is a basic graphic user interface to test the quantum random number generator and to familiarize with its performance The software includes a basic check of the random stream and a direct access to the generator parameters As an option it might also provide a very useful and powerful feature a server interface that allows to send the random data over the network Due to the time needed by the internal firmware to be fully working the VisualQRN should be started always 15 seconds after having powered the QRN Failing to do so will generate and error by the software Simply close and restart the software Installation Simply double click the d
27. tsang_gcd 0 8496 0 8879 0 1440 0 8815 0 3175 0 9092 0 2900 0 0506 0 3872 0 9979 0 5715 sts_monobit 0 0232 0 9088 0 8883 0 8851 0 8387 0 3365 0 4675 0 8459 0 9966 0 5923 0 6783 sts_runs 0 6791 0 7988 0 6423 0 8294 0 7319 0 1705 0 2974 0 6352 0 9967 0 7853 0 6567 rgb_kstest_test 0 9513 0 8005 0 5268 0 2059 0 6548 0 6139 0 5529 0 1210 0 4686 0 3972 0 5293 dab_bytedistrib 0 5512 0 0672 0 3609 0 6737 0 5163 0 3615 0 3049 0 4640 0 3203 0 0973 0 3717 dab_dct 0 1387 0 3679 0 5654 0 1441 0 1231 0 0406 0 4712 0 2342 0 8333 0 9845 0 3903 dab_monobit2 0 0425 0 2548 0 1360 0 9773 0 6560 0 1601 0 7439 0 5859 0 3931 0 6717 0 4621 sts_serial 0 4584 0 4807 0 5967 0 5733 0 6404 0 5538 0 6001 0 5897 0 6433 0 5137 0 5650 rgb_bitdist 0 6155 0 3641 0 5680 0 5508 0 5550 0 5925 0 6804 0 6519 0 5825 0 4730 0 5634 rgb_minimum_distance 0 1913 0 2326 0 4021 0 4785 0 3705 0 3996 0 6690 0 5090 0 7495 0 4966 0 4499 rgb_permutations 0 5470 0 6697 0 4188 0 5457 0 3954 0 4412 0 5461 0 8310 0 6045 0 6972 0 5697 rgb_lagged_sum 0 4910 0 5497 0 6195 0 5459 0 5702 0 5379 0 4864 0 4917 0 5671 0 6088 0 5468 dab_filltree 0 2873 0 8643 0 8436 0 6482 0 7891 0 5431 0 1971 0 5901 0 7201 0 4563 0 5939 dab_filltree2 0 2854 0 8174 0 4092 0 2859 0 3400 0 2796 0 1865 0 1648 0 7603 0 7939 0 4323 diehard_runs 0 5100 0 3717 0 8628 0 4501 0 8081 0 9790 0 6927 0 4991 0 1212 0 6766 0 5971 diehard_craps 0 2298 0 8773 0 2641 0 7589 0 2988 0 5364 0 6011 0 3626 0 4948 0 3664 0 4790 Mean 0 4436 0 5787 0 5261 0 5
28. tum RNGs are based on truly quantum physical processes whose randomness is guaranteed by theory and experiments In addition an attacker who would attempt to measure the physical observables which are used to generate the data would destructively perturb the system in most cases or would not be able to clone the intercepted message e g to replicate the number of photons measured by a photodetector A special class of QRNG is the one based on optical phenomena such as the reflection or transmission of photons by a semi transparent mirror or the number of measured photons within a fix sampling time MPD s QRN falls in this category since as said exploits the stochastic number of detected photons within a fix observation time The system is based on a monolithic CMOS chip containing an array of independent cells each capable of detecting single photons and properly generating random bits The photon detection is performed by a SPAD in each cell The array based architecture of the chip allows to boost the rate of processed photons hence random bits per second With 1024 cells laid out as an array of 32 columns and rows the present chip achieves a maximum bit rate of 200 Mbit s This fast bit rate makes the chip for example suited for One Time Pad quantum cryptography systems in which a random bit is used for each information bit sent through the communication channel 8 Micro Photon Devices S r l Italy All Rights Reserved Page
29. ware Development Kit SDK composed by a shared library and few example projects The SDK allows the user to incorporate random numbers from QRN in their own application and it is available for Windows Mac OS X and Linux operating systems both in 32 and 64 bit versions Please read the related documentation for more information about installing and using the SDK System requirements e High speed USB 2 0 interface e Host computer minimum requirements o 300 MHz processor and 256 MB of RAM e Supported operating systems o VisualQRN Microsoft Windows XP Vista 7 8 32 or 64 bit versions o SDK Microsoft Windows XP Vista 7 8 32 or 64 bit versions Linux Ubuntu 12 04 LTS or compatible distributions 32 or 64 bit versions Different distributions should work but were not tested MacOSX 10 7 5 and above Micro Photon Devices S r l Italy All Rights Reserved Page 19 QRN Quantum Random Number Generator x IM PHOTON DEVICES QRN mechanical dimensions nl A O NI re T Third angle projection in Units are in millimeters mM aie ni D m I un a 90 23 9 ia l v n O O O g i SZ RY GY N d ry f O O 1 BIT STREAM OUT VALID SYNC IN Figure 8 Dimension of the case of the QRNG The values are in millimetres Copyright and disclaimer No part of this manual including the products and software described in it may be reproduced transmitted transcrib
Download Pdf Manuals
Related Search