1998 - Draper Laboratory

Validating the Validating Tool:Defining and Measuring GPSSimulator SpecificationsAvram Tetewsky, with contributions from A. Soltz, D. Fuhry, G. Barton, D. Eyring, B. Goossens, M. Dodds, L. Fava,R. Greenspan, J. Youngberg The Charles Stark Draper Laboratory, Inc.Based on the paper published in the Proceedings of the 1997 10th International Technical Meetingof the Satellite Division of the Institute of NavigationAbstractGlobal Positioning System (GPS) simulators are often used todevelop, debug, test, and validate GPS-based systems, yet thereare no universally accepted characterizations of simulators orsimulator performance metrics, nor are there formal certificationtests for product acceptance. It is virtually impossible to make arigorous comparison between specification sheets from differentvendors because there are often subtle definitional differencesbetween products. Neither is the range of dynamics for whichthe specifications are valid explicitly defined. Ultimately, a peergroup committee or perhaps the industry itself may remedy theseproblems. This paper is intended to provide a preliminary inputand to propose a basis to initiate that dialog.How does one characterize a complex, nonstandard instrumentsuch as a GPS simulator? How should one quantify instrumentperformance? How are common receiver performance metrics oruser mission requirements related to simulator specifications?How does one characterize simulator performance when it issensitive to the user dynamics being tested? How does oneprovide a reference system with performance that is known to besuperior to the simulator under test?This paper presents strawman solutions to these problemsaccompanied by an example of supporting measurements. Thepaper presents an abstract vendor-independent model of a GPSsimulator and uses it to define simulator specifications as afunction of user dynamics. A sample simulator specificationsheet is presented. User accuracy is described as a function ofsimulator specifications for point positioning tests. A procedurefor independently verifying simulator specifications is presented.Finally, we propose a Summary Simulator Capability Matrixindexed by user mission needs. This summary chart does notrequire users to understand the significance of all the detailedsimulator specifications.The discussion is limited to the simulation of GPS signals inspace. The extension to simulation of integrated sensor suites isnecessary but beyond the scope of this paper.IntroductionThere are three critical reasons why GPS simulators areexpensive, difficult for users to understand, and difficult tovalidate. By understanding these reasons, we can reduce thesimulator validation problem into a series of smaller, welldefined,solvable problems.First, GPS simulators are far more complicated than GPSsatellites. They have to create an accurate representation of theconstellation of satellite signals and then reshape them so thatthey match the signals received by a moving user. Thisreshaping includes applying the effects of the ionosphere, thetroposphere, and other intervening environments. Of thesevariables, correctly accounting for user dynamics may have thegreatest effect on all aspects of the simulator specification andvalidation process.Second, GPS simulator specifications are hard to interpretauthoritatively because they are usually linked to vendorspecifichardware and software mechanization. The definitionsof terms used in specifications found in today’s advertisementscan vary subtly from vendor to vendor. For example, mostspecification sheets provide only partial data on channelaccuracy, sampling rates, and software options. From this, a usermust figure out what the definitions mean and then map thesespecifications back into items relevant to his test objectives.Navigation accuracy may be only one of those relevant items.For example, a user may also want to test his receiver’s responseto perturbations in the 50-Hz navigation data message.Finally, one of the most difficult parts of simulator validation isachieving an independent measurement of simulator accuracy.Because the RF signals are below the noise floor, the tester eithernotes their effect on a GPS receiver or measures internal vendorproprietarysignals in the simulator. These approaches are notsatisfactory for several reasons. First, using last year’s “Goldenreceiver” (which, in turn, was developed with a previous year’ssimulator) generates circular experimental data to arrive ataccuracy measurements. Second, using a vendor’s proprietaryValidating the Validating Tool: Defining and Measuring GPS Simulator Specifications2