Achieving 20cm Positioning Accuracy in Real Time Using GPS – the
Contents
1. shown inset embedded in a GPS antenna GEC REVIEW VOL 14 NO 1 19992. 98 in April 1998 at Palm Springs CA USA and published in the conference proceedings References 1 MILKEN RJ and ZOLLER C J Principle of operation of NAVSTAR and system characteristics Global Positioning System Journal of the Institute of Navigation 1 pp 3 14 1980 2 LANGLEY R B RTK GPS GPS World pp 70 76 September 1998 38 RTeStar User s Manual ver 001 Canadian Marconi Company publication no 1220 GEN 0101 February 1998 4 RTCM Recommended Standards for Differential NAVSTAR GPS Service ver 2 1 RTCM Special Committee 104 January 1994 5 THEROUX Y Open architecture design for GPS applications Proceedings of the ION GPS 95 The 8th International Technical Meeting of The Satellite Division of the Institute of Navigation pp 117 121 1995 6 MaASELLA E GONTHIER M and DUMAINE M The RTeStar features and performance of a low cost RTK OEM sensor Proceedings of the ION GPS 97 The 10th International Technical Meeting of The Satellite Division of the Institute of Navigation pp 53 59 1997 7 RAMACHANDRA K V MOHAN B R and GEETHA B R A three state Kalman tracker using position and rate measurements IEEE Transactions on Aerospace and Electronic Systems 29 1 pp 215 221 January 1993 GEC REVIEW VOL 14 NO 1 1999 ACHIEVING 20CM POSITIONING ACCURACY IN REAL TIME USING GPS 27 9 The RT Star smart antenna unit an RT Star
3. the antenna on the vehicle he above routine was repeated ten times Fig 7 shows the position in the vertical plane he red marks on the time axis represent the periods at which the antenna was located on the survey point It is observed that repeatable results are obtained at each run a GEC REVIEW VOL 14 NO 1 1999 26 7 Vehicle vertical difference The error in the horizontal plane of the computed position of the marker at each run is presented in fig 8 Conclusions The tests conducted on the RTeStar show that 20 cm positioning accuracy can be obtained in realtime using GPS Furthermore the results have been proven to be repeatable The RTeStar fig 9 is considered an excellent trade off between low cost code differential GPS systems delivering l metre accuracy levels and high end two frequency RTK systems offering an accuracy level of only afew centimetres Itis suited for a variety of applications namely precision agriculture aircraft positioning and machine control It also gives the user the flexibility to confi gure the receiver either as a reference station or a roving unit via a command message Acknowledgement This paper is based on a presentation entitled Precise Kinematic Positioning Experiments with a Low Cost RTK GPS Engine given by the author at E MASELLA 8 Horizontal position error on survey point the IEEE Position Location and Navigation Symposium
4. 20 Achieving 20cm Positioning Accuracy in Real Time Using GPS the Global Positioning System by E MASELLA B Eng PE Marconi Canada Marconi Canada part of the Marconi North America Group of Marconi Electronic Systems is a recognized world leader in the design manufac ture sale and support of high technology elec tronic products for the aerospace and communications market for both military and commercial applications Its headquarters and principal design and manufacturing facility is located in St Laurent Quebec in the greater Montreal area Its facilities include a branch plant in Kanata Ontario in the Ottawa area as well as sales and service offices across Canada A photo graph of the Company s headquarters is presented in fig 1 Marconi Canada is a pioneer in the design and manufacture of GPS receivers The first receiver development took place in the early 1980s with a military receiver development for the Canadian Department of National Defence Then in the mid 1980s the first commercial GPS receiver was developed The next major development took place in the early 1990s for a GPS receiver to be installed on the new Boeing 777 aircraft This marked the beginning of a series of successes for Marconi Canada in the commercial avionics market and the Company is now recognized as a world leader in this area In the mid 1990s a new group was created with the objective of porting our GPS technology on low cost
5. Dual serial communication ports 3 Receiver block diagram 1575 42 MHz to 4 3 MHz and a 2 bit analog to digital conversion The temperature controlled crystal oscillator TCXO is the receiver s reference oscillator The DSP chip includes a 12 channel GPS signal correlator and the following peripheral circuits e two programmable UARTs e real time clock e programmable interrupts e watch dog and reset circuit and e discrete I Os The processor is the system s critical component The ARM60 RISC processor was selected because it has e the processing power to handle 12 tracking channels e 30 spare capacity for customer specific tasks and e lowcost and power consumption Software Overview The embedded software was developed using object oriented design and programming techniques in order to yield reusable software components and to encapsulate the functions most subject to change The major objects encapsulate the following functionalities e Signal Processing e Satellite List Management e I O Management e Differential Data Processing and e Navigator The Navigator is the ultimate recipient of the work produced by the other objects it receives all the data measurements differential data satellite data and produces a PVT estimate GEC REVIEW VOL 14 NO 1 1999 ACHIEVING 20CM POSITIONING ACCURACY IN REAL TIME USING GPS 23 Navigator Overview A comprehensive description of the Navi
6. actor between errors observed at both sites argely depends on the distance between the two receivers A DGPS system therefore consists of at least two units a reference station and one or several roving units The reference station broadcasts its differential data and the roving units receive it through a data port directly connected to a radio receiver The roving units can then display position velocity time PVT and other information as needed for their marine land or aeronautical Differential receiver 2 Code Differential Global Positioning System DGPS example applications A DGPS system implementation is depicted in fig 2 note that the reference station is located on the ground and the roving unit is located in the aircratt Carrier Differential GPS Overview A different and more accurate differential correction technique involves tracking of the satel lite signal s carrier phase and is called carrier phase DGPS When a receiver navigates in carrier phase mode it is measuring a different GPS observable namely the GPS carrier wave To do so it must measure the phase of the carrier continuously from signal lock time tg in order to produce the following observable t Ott P t to fdt 1 to where is the received carrier phase and f is the received Doppler frequency In order to produce a range measurement between the satellite and the user the
7. ce in the carrier phase measurement made by the same receiver observ ing two satellite signals simultaneously The single difference operator is denoted as V From n satellites n 1 single differences can be formed The simplest way of forming these n 1 observ ables is by selecting a reference satellite s and applying the operator V as follows Voie G4 g 5 gi svt where represents the phase measurement by the user of satellite i Expanding eqn 5 using eqn 3 shows that the receiver clock errors are removed in the resulting single ditterence observable To form the next observable a double ditfer nce a set of n 1 single differences must be formed with measurements from the roving user Hence a set of n 1 single differences are formed with measurements received by the reference station on the differential link A double difference would read as follows AVON Vou Voie 6 The property of this observable is that satellite clock errors are removed Expansion of eqn 6 with eqns 3 and 5 yields an equation in which the unknowns are AVR the value of the geometric distance of the double ditterence across satellites and receivers AViono the value of the double advance of the carrier because of the ionosphere across satellites and receivers AVtropo the value of the double differenced delay of the carrier because of the troposphere across satellites a
8. following observable must then be formed pt No nlt 2 where Nois an exact number of cycles between the user and the satellite at time tp m thus represents the change in cycles since time to However the initial number of wavelengths No between the satellite and the receiver is unknown This is called the carrier phase ambiguity and must be estimated In order to estimate this ambi guity itis necessary for the roving GPS unit to use information that is carrier phase measurements from a reference station This technique yields accuracies in the cm range in dynamic environ ments and is called Real time Kinematic or RTK GPS A rather good albeit brief introduction to code and carrier phase DGPS is presented in reference 2 GEC REVIEW VOL 14 NO 1 1999 22 System Specifications The RTK capable engine developed by Marconi Canada called the RTeStar consists of single frequency L1 RTK Navigation software residing on a low cost hardware platform for embedding in Original Equipment Manufacturer OEM systems It measures 4 X 2 65 100 mm X67 mm and consumes 2W Details of its use performance and specifications are presented in reference 3 This product is considered to possess one of the highest accuracy over price ratios in the OEM sensor industry The RTeStar can be configured either as a reference station or a roving unit via a command message Communication between both is i
9. fore suggest that the purpose of the Off Line Filter is to estimate the ambiguity vector measurement model in the background Because the dynamics of the ambiguities are very small they tend towards constants this filtering can be executed at a rate significantly lower than the required navigation rate The On Line Filter is thus used to provide user PVT estimates at the nominal navigation rate It will use the ambiguity vector computed by the Off Line Filter Static Tests A series of tests was carried out using three surveyed antennas located on Marconi Canada s roof in order to establish the accuracy conver gence time and functional behaviour of the RTeStar in static mode E MASELLA North error 0 57 Error m o 0 57 1 1 1 9 Chil 9 2 9 3 9 4 95 9 6 05 East error Error m io i 9 9 1 9 2 9 3 9 4 9 5 9 6 Vertical error Error m 9 4 GPS time 10 s 4 Sample reset test Repeatability and convergence time were tested by forcing a system reset every 20 minutes overa period of 24 hours A sample of one reset test isshown in fig 4 The reset times are marked witha cross on the time axis Itis observed that the system repeatedly converges towards zero error Table 1 shows some statistics generated from the results The 3 D rm s error reaches 20 cm after approximately 7 minutes TABLE 1 Reset Tests Statistics Kinematic Tests Two types of ki
10. gator s algorithms is presented in reference 6 Only a brief overview is presented below One of the first steps to be performed in an RTK Navigation solution is to form two sets of observ ables the single and the double differences Both are derived using the carrier phase observ able which is defined as p R c t AE Aiono Atropo e NGA ng 3 where R is the true range between satellite and user is the speed of light At is the user s clock offset from true time AE is the satellite s clock offset including relativistic effects Aiono is the total ionospheric advance the dispersive property of the ionosphere on he carrier actually causes an advance of he signal with respect to the modulated code Atropois the total tropospheric delay e gt t is the error introduced by the orbital data No is the initial carrier phase ambiguity A is the signal s wavelength 19 cm for L1 and Ne is the error caused by the receiver thermal noise it is a Gaussian process with zero mean Note that in eqn 3 R is the true range between the satellite and the user Geometrically this can be expressed as R Je xu ys yy zs _ zy 4 where xs y5 z5 are the satellite s coordinates as provided by the satellite s navigation data and x y 4 z are the unknown user s coordinates A single difference across satellites is defined as the instantaneous differen
11. hardware platforms This initiative has been a success as well with the development of a variety of products targeted towards the consumer 1 The headquarters of Marconi Canada Erik Masella gained his bachelor s degree in Electrical Engineering at the University of Sherbrooke in 1991 In 1992 he joined the Aerospace Group of Marconi Canada and has been involved in the design and test of airborne GPS software and GPS aided landing systems In 1996 he joined the GPS OEM Group Since then he has been involved in the development of new GPS products for the consumer market He is a chartered Professional Engineer E mail emasella mtl marconi ca Glossary ASIC Application Specific Integrated Circuit DGPS Differential Global Positioning System FEPROM Flash Erasable Programmable Read Only Memory I Intermediate Frequency VO Input Output OE Original Equipment Manufacturer ENA Position Velocity and Time RF Radio Frequency RISC Reduced Instruction Set Computer RTK Real Time Kinematic UART Universal Asynchronous Receiver Transmitter market namely automatic vehicle location systems marine navigation golf yardage systems surveying systems and also some military applications requiring commercial off the shelf COTS solutions Introduction to GPS The NAVSTAR NAVigation Satellite Timing And Ranging Global Positioning System better known as GPS is a radionavigation system using a network of satellites d
12. istributed over six orbital planes GPS provides accurate 3 D position vel ocity and time information and world wide 24 hour coverage to an unlimited number of users with all weather operation GPS is a one way ranging system signals are transmitted only by the satellites Each GPS satellite transmits signals centred on two microwave frequencies 1575 42 MHz referred to as Link 1 or simply L1 and 1227 60 MHz referred to as L2 The L1 signal is modulated with i the Coarse Acquisition or C A Code a coarse ranging signal ii the Precise P Code a precise but encrypted ranging signal and iii navigation data at 50 bits per second L2 is modulated only GEC REVIEW VOL 14 NO 1 1999 ACHIEVING 20CM POSITIONING ACCURACY IN REAL TIME USING GPS 21 with the P Code and the navigation data Auth orized users only U S Department of Defense and allies have access to the decryption keys of the P Code which by its nature has much higher anti jamming properties than the C A Code Using the ranging signal and the navigation data a GPS receiver can measure the range between the satellite and the receiving antenna and compute the exact position of the transmitting satellite Hence with three satellites the 3 D position of the receiver may be computed and a fourth is required for solution of the time An excellent overview of the GPS system operation is given in reference 1 Code Differential GPS Overview Code Diffe
13. ition errors are presented in fig 6 Field Tests The purpose of these tests was to verify the accuracy and the functionality of the RTeStar in a real life environment Several kinematic baseline 025 North error 0 E 1 raana an 5 0 eaan ii EN R a Oot A I I I I 1 0 1000 1600 2000 250 3000 3000 4000 I 1 1 1 s00 1000 150 2000 250 3000 3500 4000 Vertical error 02 Ei ako 1400 adw osm wee ako aR t GPS time s 6 Simulator test position errors tests were performed using a reference station located on the Company s roof and a vehicle equipped with an antenna and an RTeStar The vehicle followed a route located in an urban zone with a surveyed geodetic marker situated on the route The test procedure is as follows e the antenna is fixed toa surveying rod seton top of a geodetic marker located in front of the Company building e system is powered on and remains static for 2 minutes e theantennais relocated on the vehicle s roof the operator triggers a signal that logs the ime at which the antenna is removed e the vehicle drives around the Company site and stops at the geodetic marker e the antenna is removed from the rooftop and seton top of the marker the operator triggers a signal that logs the time at which the antenna is set e the antenna remains static for approximately 2 minutes the operator triggers the removal time log and relocates
14. me matched double differences can be produced This extrapolation process is implem ented using Position Velocity Acceleration PVA tracking filters which are not discussed here but interested readers may find details on these filters in reference 7 One realizes that the estimated state vector X is composed of two basic sets of parameters X User Model Measurement Model The observability on these two sets is quite differ ent The user model represents the dynamics of the roving user This set must be updated at the basic rate of the Navigator However the measurement model builds its observability on the change of geometry of the satellites that compose the unknown ambiguity A fast update rate will not significantly improve the observability on the measurement model because of the high accuracy of the carrier phase measurements The Navigation function is therefore separated into two processes e a Kalman filter that acts upon reference station and roving unit measurements sampled at the same time Because of the data link latency this solution is typically old by a few seconds This filter is referred to herein as the Off Line Filter and e a Kalman filter that acts upon extrapolated reference station measurements and roving unit measurements sampled for the current time The update rate of this filter is currently at 1Hz It is referred to herein as the On Line Filter These findings there
15. mplemented via a standard radio link and the information is encoded in RTCM 104 format Communication from a host computer with the GPS engine is performed via a serial port using an RS 232 Marconi proprietary transmission proto col The reference station is also capable of self surveying its position The RTeStar has the following features 12 parallel tracking channels GPS measurements sampling aligned on GPS time one second roll over event raw measurement output rate of 10 Hz time mark signal output aligned on GPS time one second roll over event keep alive input pin RAM and or real time clock dual UART third UART optional six input output discrete control lines reprogrammable operational code FEPROM and rechargeable lithium battery optional The RTeStar is compatible with both active and passive antennas System Architecture Hardware Overview The RTeStar hardware is a highly integrated design built around three components e RF front end e digital signal processing DSP ASIC and e RISC processor Fig 3 depicts a block diagram of the receiver The RF chip preceded by a low noise amplifier LNA performs a triple IF conversion from E MASELLA Memory control RTC crystal ROM SRAM EEPROM I I B Address bus Data bus GPS antenna correlator and ARM60 support 32 bit RISC functions si GP2010 RF a GP2021 front end 12 channel a
16. nd receivers AVN the double differenced ambiguity expressed in cycles itis itself an integer since it is the algebraic sum of four integers the value of the double ditfferenced orbital errors across satellites and bit AVE ii the double differenced phase noise AVig the sum of four Gaussian processes with zero mean Because AVy may be neglected if the receiver has relatively low phase noise as well as AVe beyond the scope of this text and because AViono and AVtropo may be modelled only AVN remains to be properly estimated in order to solve for R Fortunately No is a constant hence it may be estimated over time using the carrier phase observables So as we can see a set of synchronized carrier phase measurements that is measurements that are sampled at the same time from the reference station and the roving unit is required at the latter One problem which arises is that the data link that provides the carrier measurements from the refer ence station has a given transmission latency Consequently the roving unit will receive its reference measurements with a typical latency of l to 2 seconds Because a Navigation solution is required for the current time the roving unit will have to have some means of extrapolating the GEC REVIEW VOL 14 NO 1 1999 24 reference measurements for the current time that is coincident with the unit s measurement time then ti
17. nematic tests were performed e GPS signal simulator tests to verify the absolute accuracy of the solution and e field tests to verify real life accuracy and functionality of the system GEC REVIEW VOL 14 NO 1 1999 ACHIEVING 20CM POSITIONING ACCURACY IN REAL TIME USING GPS 25 RF cable RF cable Antenna input Antenna RTCM messages input RT Star graphical user interface 5 Simulator test set up GPS Signal Simulator Tests The purpose of the simulator test was to verify the accuracy of the RTeStar in a normal dynamic environment Two Nortel model STR2760 GPS sig nal simulators were set up in differential mode to carry out this test Fig 5 depicts the test set up The test simulates an aircraft performing dynamic manceuvres The scenario is as follows e the aircraft is static for 10 minutes e itaccelerates linearly with a lg acceleration up to 100 m s e itclimbs to an altitude of 1 km e itthencompletes a square path at a constant speed of 100 m s with 10 minutes between each turn e it touches ground and decelerates to 0 m s and e itremains static for 5 minutes The position solution generated in real time by the RTeStar was logged and then compared to the truth file of the simulator which is a file containing the true vehicle position at each GPS second The results were processed with the MATLAB data analysis tool provided by the MathWorks Inc and the North East Down pos
18. rential GPS Code DGPS is the regular Global Positioning System with the addi tion of a differential signal that conveys correction data These data significantly increase the accu racy of the GPS navigation function and can be broadcast over any authorized communication channel In these systems a GPS receiver is located at a known surveyed position this receiver is usually referred to as a reference sta tion The reference station makes measurements on the satellite s signal and estimates the measure ment errors using its surveyed geodetic position The errors include the signal transmission delays caused by the ionosphere and the troposphere as well as Selective Availability S A an intentional signal degradation introduced by the U S Depart ment of Defense in an effort to restrict the accuracy capability of most civilian GPS users In layman s terms because the reference station knows precisely where it is and computes a differ ent position using the GPS signals it can estimate the errors in its signal measurements These errors or differential corrections that is the difference between the true range and the measured range are then transmitted to roving receivers by radio or other means They can then be applied to GPS measurements from the roving GPS receiver and used to remove the systematic correctable error actors because most of these errors will be similar or the roving receivers Note that the correlation
Download Pdf Manuals
Related Search