Download - 1 Thompson

Transcript
Page 1: 1 Thompson

T. THOMPSON, L. J. LEVY, AND E. E. WESTERFIELD

S

The SATRACK System: Development and Applications

Thomas Thompson, Larry J. Levy, and Edwin E. Westerfield

ATRACK has been a significant contributor to the development and operationalsuccess of the Trident Weapon System, and it continues to provide a uniquemonitoring function that is critical to the maintenance of the U.S. sea-based strategicdeterrent. This article reviews the background and evolution of this unique GlobalPositioning System user application (the first committed user) and discusses its supportof other systems of national importance. Its connection to APL’s pioneeringcontributions to the development of the technology and methodology supportingsuccessful deployment of the sea-based strategic deterrent is reviewed. Basic conceptsand implementation specifics of the evolving system design are described, and itsextended use and performance improvements over the past 25 years are presented.(Keywords: GPS translators, Missile system test and evaluation, Satellite positioning.)

INTRODUCTIONSATRACK was developed to validate and monitor

the Trident missile guidance error model in the SystemFlight Test Program. It is the primary instrumentationand processing system responsible for accuracy evalu-ation of the Navy’s Strategic Weapon System. Instru-mentation and processing systems available when theTrident Development Program began could not meetthis need. APL conceived and led the development ofthe SATRACK system to fulfill this requirement. Pro-totype instrumentation required for the missile andground station data collection functions were devel-oped at APL to validate the concept, and we generatedspecifications controlling the development of the op-erating missile and ground station hardware. We alsodeveloped and operate the SATRACK processingfacility, which includes a unique preprocessing hard-ware and software configuration and an extensive

436 JOH

postprocessing analysis capability. Additionally, theAPL satellite tracking facility has operated as a backupSATRACK recording site for all East Coast test flightssince 1978.

SATRACK fully met all its guidance subsystemevaluation requirements and also provided weaponsystem error model insights that would not have beenpossible without its unique ability to detect and allo-cate small error contributors to miss distances observedin the flight test program. SATRACK not only vali-dated the Trident system accuracy, but the test-derivedand -validated system error models have allowed thecommand authority to confidently assign and allocatetargets to sea-based strategic resources.

The next section will discuss the background lead-ing to SATRACK’s development as a natural exten-sion from APL’s role in the development of the Fleet

NS HOPKINS APL TECHNICAL DIGEST, VOLUME 19, NUMBER 4 (1998)

Page 2: 1 Thompson

Ballistic Missile Evaluation System and the Navy Nav-igation Satellite System (Transit). Then, following adiscussion of the basic concepts, this article will tracethe system’s evolution:

• SATRACK I—A technology project to develop anddemonstrate the instrumentation and processing sys-tem using the Trident I missile (1973–1983)

• SATRACK II—The operational system designed tomeet system requirements for the Trident II missile(1983–present)

• Other applications—Uses of SATRACK for Armyand Air Force missile test applications (1983–present)

• SATRACK III—Current system upgrade and futureapplications

BACKGROUNDFrom 1967 through early 1971, APL conducted a

series of studies to support concept development of theDefense Navigation Satellite System (DNSS). Thesestudies addressed a variety of configurations capable ofmeeting DNSS requirements and eventually led to aconcept proposal called Two-In-View (TIV) Transit.The name was chosen to indicate that DNSS require-ments could be met with only two visible satellites (incontrast to the alternate concepts that required fourvisible satellites) and that it could evolve naturallyfrom the already operational Transit system. We chosethis concept because it was the lowest-cost approachto meeting DNSS requirements.

The ability to provide three-dimensional position-ing with the required accuracy using two satellites ispossible only because their motion relative to a useris significant. The alternate concepts were based onsimultaneous range measurements to four satellites.Since these concepts were not benefited by highrelative motion, they incorporated satellite constella-tions at higher altitude. The higher-altitude constel-lations were selected because they achieved therequired global coverage with a smaller number ofsatellites. This choice was partly motivated by theincorrect assessment that system costs increased as therequired number of satellites increased. However, few-er satellites were possible because the area that eachserved expanded in direct relation to the reduction inthe number of satellites, and costs for signal servicestend to have a direct relationship to service area, notto the number of satellites. To a first-order approxima-tion, positioning satellite system costs are independentof constellation altitude. Further discussion of thesetopics and the prevalent views of the time is availablefrom Refs. 1–3.

The last performance study of the TIV Transitsystem addressed its ability to provide trajectory mea-surements of SLBM test flights. The study showed thatSLBM measurement objectives could be met, and an

JOHNS HOPKINS APL TECHNICAL DIGEST, VOLUME 19, NUMBER 4 (

THE SATRACK SYSTEM

interim report published in late 1971 formulated thetracking concepts and missile and ground station ca-pabilities needed to support TIV Transit measurementsof SLBM flight tests. However, by mid-1972 it wasclear that none of the then-proposed DNSS conceptswould be developed, and the missile tracking conceptwas temporarily forgotten.

In a parallel chain of events, the Navy’s StrategicSystems Programs organization was asked to addressthe suitability of the Trident Weapon System for moreaccurate targeting requirements. Several studies wereinitiated to consider this question. The primary issuesconcerned possible system improvements to achievethe desired accuracy. An important secondary concernregarded the method by which the accuracy of the newsystem would be validated.

It was soon evident that the impact scoring tech-niques used for Polaris and Poseidon evaluations wouldnot be adequate. A new methodology that providedinsights into major error contributors within the flighttest environment would be needed so that accuracyprojections could be based on a high-confidence un-derstanding of the underlying system models. Thetechnique of comparing observed test impact statisticswith results computed from models used for develop-ment (i.e., “shoot and score” approach) was unaccept-able. Assessing performance models in the flight testenvironment requires guidance-independent measure-ments with sufficient precision to separate out theimportant contributors to system inaccuracy. Theexisting range instrumentation (missile tracking andtrajectory estimation) was largely provided by radarsystems, and it was not clear that they could providethe needed measurement accuracy or coverage in thebroad ocean test areas.

In early 1973, we initiated a study to compare cur-rent range radar with TIV Transit measurement capa-bilities in relation to needed SLBM accuracy evalua-tion objectives. The study results showed that only asatellite-based measurement system could meet futurerequirements. APL presented the SATRACK conceptto the Navy’s Strategic Systems Programs staff in May1973. The proposed system was based on a customsatellite design patterned after TIV Transit satellites,but simplified by the removal of any requirement forreal-time positioning service. A six-satellite constella-tion would support two flight test windows per day.These concepts were proposed to minimize costs.Missile hardware and ground support capabilities wereunchanged from the concepts defined in 1971.

The proposal was accepted and preliminary devel-opment was initiated. However, 1973 was also the yearthat the Global Positioning System (GPS) develop-ment began. With the emergence of GPS, we wereasked to consider its use in place of the specializedsatellite constellation. Our studies indicated that the

1998) 437

Page 3: 1 Thompson

T. THOMPSON, L. J. LEVY, AND E. E. WESTERFIELD

GPS could be applied to the SLBM accuracy evaluationsystem with some changes to the missile and groundstation designs. There was a technical concern regard-ing the available signal power and ionospheric correc-tion capabilities as well as a programmatic concernregarding the number of satellites that would be avail-able for early Trident test flights, but otherwise, theGPS capabilities were expected to be adequate. In July1974, the Improved Accuracy Program was initiated toconsider the implications of modifying Trident to sup-port an improved accuracy requirement. SATRACKdevelopment was initiated to support the program, andin September 1974, DoD Research and Engineeringdirected the Navy to use GPS for SATRACK, makingit the first committed GPS user system.

BASIC CONCEPTSThe SATRACK concept is shown in Fig. 1. Signals

from GPS satellites are relayed by the test missile toreceiving equipment at a launch-area support ship(LASS) and a downrange support site (DRSS). At the

beginning of the Tused for downrangvided by a land called a translator functions are accsignals are simply mitted). The evocussed in a comWesterfield elsewequipment at the vide a SATRACKthe DRSS does ptracking capabilitylated signals receihigh enough rate trum. Precision traaccomplished at tthrough playback

After the signaseveral systematicsatellite-to-missiledata are used in

Figure 1. SATRACK measurement concept. Signals transmitted from the Global Posi-tioning System satellites are received at the missile, translated to another frequency, andrelayed to the telemetry station, where they are recorded for later playback andpostprocessing at APL.

Recordingequipment

L-bandlow noiseamplifier

M1First IF and

GPSsignal filter

M2S-band power

amplifier

Oscillator andfrequency

synthesizer

APLpostprocessing

facility

All-in-view GPS satellites(L-band signals)

MissileTranslator

TelemetryStation

(S-band signals)

(Signal data)

438 JOHN

rident Flight Test Program a ship wase support; now that support is pro-station. The missile hardware wasto emphasize that no signal trackingomplished in the missile (i.e., the“translated” to S-band and retrans-lution of translator systems is dis-panion article by Thompson andhere in this issue. The receivingLASS and DRSS also does not pro- signal tracking function (althoughrovide a lower-accuracy real-time to support range safety). The trans-ved at both sites are sampled at ato capture the desired signal spec-cking of the GPS signals is actuallyhe APL postflight tracking facilityof the recorded translator signals.l tracking data are recovered and

corrections are applied, the derived (link) range and integrated Dopplera large Kalman filter that provides

estimates of trajectory-observablemodel parameters (this terminolo-gy is used because not all modelparameters produce observable sig-natures in any specific trajectorygeometry). The processing meth-odology developed at APL proper-ly combines a priori model data andtrajectory-observable model datafor each flight test.

Significantly, the critical mea-surements are provided by carrierphase tracking of the GPS-to-missile signals. The GPS signalphase measurements (i.e., integrat-ed Doppler) sense range changesalong each signal line of sight to asmall fraction of a wavelength (i.e.,a few millimeters). These measure-ments, which are compared withtheir values computed from guid-ance sensor data and satelliteposition and velocity estimates,provide most of the information.Range measurement noise in therecovered GPS range-code signalsis of secondary importance. Inessence, the inertial sensors pro-vide high-frequency motion infor-mation better than the signal pro-cesses, the Doppler informationsenses the systematic errors associ-ated with the inertial sensors, andthe range data provide an initial

S HOPKINS APL TECHNICAL DIGEST, VOLUME 19, NUMBER 4 (1998)

Page 4: 1 Thompson

THE SATRACK SYSTEM

condition for all the dynamic measurements. There-fore, range noise (i.e., noise in the range trackingloops) can be smoothed over the full flight interval.The range noise remaining after this process is smallerthan other bias-like uncertainties that set the limit onabsolute position accuracy (e.g., satellite position).

The SATRACK system configuration is shown inFig. 2. Signals transmitted by GPS satellites are trackedby the GPS tracking network for several days surround-ing the missile test flight. The tracking data from thisoperation are processed to derive satellite ephemeridesand clock estimates that have the highest possibleaccuracy during the missile flight interval. The eph-emerides and clock estimates are used by the postflightreceiver and missile processor to provide the SA-TRACK data products for each flight.

During the missile flight, all in-view GPS satellitesignals are received at the missile, translated to S-band,and retransmitted to the surface station. The translat-ed GPS signals are recovered with the same stationtracking antenna used for all the missile telemetrysignals. The translated GPS signal data are separatedinto range safety and accuracy operations. The rangesafety function processes the lower-accuracy GPS sig-nals to provide a real-time tracking solution for therange command. The real-time accuracy function isprovided by simply sampling and recording the GPSsignals (i.e., by sampling the signals at a rate that

JOHNS HOPKINS APL TECHNICAL DIGEST, VOLUME 19, NUMBER 4 (1

captures the full bandwidth of the selected range-codemodulation).

The telemetry data are also recorded for postflightanalysis. These data, along with GPS ephemeris andclock data, are used to provide tracking aids for thepostflight receiver and measurement estimates for themissile processor. Raw tracking data from the postflightreceiver are corrected for known systematic errors(e.g., antenna lever arm) before being passed to themissile processor. The missile processor is a largeKalman filter that operates on the raw guidance sensormeasurement data supplied by missile telemetry. Withthese data and satellite ephemeris and clock estimates,the processor computes the expected satellite measure-ment data. These expected data are compared to theactual satellite measurements, and the observed differ-ences are used to drive the filter model to a best es-timate of the underlying guidance model errors.

Assuming that the data processing has not identifieda system fault (i.e., an error component well outside itsexpected performance), the processed data from eachflight test are used to provide estimates of major con-tributors to impact miss. Figure 3 shows a hypotheticaldiagram used to allocate contributions to impact miss.The method is based on projecting each error contrib-utor and its uncertainty into the impact domain. Thefirst-level allocation is at the subsystem level (initialconditions, guidance, and deployment and reentry); a

Trackingantenna

Telemetryprocessor

Postflightreceiver

Trackercorrections

Missileprocessor

Systemmodels

Guidance data

Estimated initialconditions

Ephemerisand clockestimates

Real time Postflight

Orbit/clockdetermination

facility

GPStrackingnetwork

GPSsatellites

Testmissile

Rangesafetysystem

Preamplifier

Telemetryreceiver

Translatorreceiver

Translatorrecorder

SATRACKrecorder

Figure 2. Basic SATRACK configuration. For a number of days surrounding the missile flight, GPS signals are received, tracked, andrecorded at the GPS tracking sites. During the missile flight, GPS signals are received by the missile, translated in frequency, andtransmitted to the surface station(s). A tracking antenna at the station receives the missile signals, separates the various components,and records the data. The postflight process uses the recorded data to give satellite ephemerides and clock estimates, tracked signaldata from the postflight receiver, and missile guidance sensor data. After the signal tracking data are corrected, all the data elementsand the system models are used by the missile processor to produce the flight test data products.

998) 439

Page 5: 1 Thompson

T. THOMPSON, L. J. LEVY, AND E. E. WESTERFIELD

second level provides data for major error groups withineach subsystem (e.g., accelerometers). A third level,not shown in the figure, produces estimates of funda-mental error terms of the guidance model (e.g., anaccelerometer scale factor error). In addition, SA-TRACK provides a point estimate of the initial con-ditions and a pre-deployment estimate; however, itsprincipal purpose is to evaluate guidance subsystemerror. The results from each flight are checked for sta-tistical consistency with independent measures of im-pact, initial conditions, and several other factors.

As noted previously, limitations of test geometrywill prohibit observations of all errors in any singleflight test. However, since each flight provides obser-vations of the underlying missile guidance error mod-els, the data from many flight tests can be combined.The final cumulative analysis of flight test data pro-duces a guidance error model for the Trident WeaponSystem. It combines observations from each flight toderive a tactically representative missile guidancemodel based completely on flight test data. This modelis used with other similarly derived subsystem modelsto develop planning factors used to assign weaponsystem targets.

SATRACK has evolved over a quarter century. Theoriginal development for the Trident I missile (C4)began in earnest in 1974.4 The C4 version (SATRACKI) was a technology development program aimed at

TRACK II has A general u

aging componeSATRACK groin the final stagflight facility is replacement misthermore, effortyears on a new dent reentry bonot only modernfunctions, it wilperformance ca

SATRACK E

SATRACK IThe C4 miss

was decided to improposed designadditional missiof the existing Santenna had toheld the GPS aspaced uniforml

With the widsignal sums wo

Downrange miss

Crossrange miss

Velocity

Orientation

Deploymentand reentry

Total S

ATRACKGuidance

Initialconditions

Position

Accelerometers

Gyros

“Observed” impact

Impactuncertainties

EstimatedSATRACKimpact

TotalSATRACKuncertainty

Uncertaintyin guidance

estimate

Uncertainty ininitial conditions

estimate

Figure 3. Hypothetical example of SATRACK impact evaluation. Each error estimateprovided by the SATRACK process is projected into the impact domain, showing itsdownrange and crossrange contribution to impact miss (the center of the coordinatesystem is the aim point). Uncertainties for each estimate are also projected. The estimatederrors and their uncertainties are tested for statistical consistency with system models andother independently measured results (e.g., the initial conditions derived from launch areainstrumentation).

440 JOH

(1) gaining insight into whatwas needed for improved accuracyand (2) developing an adequateaccuracy evaluation system. SA-TRACK, as it evolved, is the ful-fillment of the second major ob-jective. SATRACK I proved thatit could provide adequate esti-mates of guidance subsystem errorsfor individual flights. It was a pi-oneering effort in that both thetracking methods and the largeKalman filter processing tech-niques were pushing the state-of-the-art.

The second phase (SATRACKII) was a major upgrade in responseto the stringent measurement re-quirements set by the AccuracyEvaluation System study.5 Thestudy established the total weaponsystem instrumentation require-ments for the Trident II (D5) mis-sile in accordance with specifiedaccuracy evaluation objectives,including SATRACK as well asother prelaunch, in-flight, andreentry area instrumentation. SA-

been operational since 1987.pgrade has been initiated to replacents of the D5 test system. The newund recording equipment is currentlyes of checkout, upgrading at the post-well along, and preliminary design ofsile test components is beginning. Fur-s have been focused for the last severalGPS translator system to support Tri-dy testing. The upgraded system willize the facility hardware and software

l also substantially extend SATRACKpabilities.

VOLUTION

ile design was well under way before itplement the SATRACK system. The

approach did minimize the requiredle electronics and it did take advantage-band antenna; however, a new GPS

be added. The C4 design constraintsntenna configuration to four elementsy around the missile circumference.e spacing between antenna elements,uld cause strong interferometer null

NS HOPKINS APL TECHNICAL DIGEST, VOLUME 19, NUMBER 4 (1998)

Page 6: 1 Thompson

patterns. To minimize the problems associated withremoving antenna phase effects in the postprocessor,opposite pairs of antennas were summed to form two-element interferometers oriented 90° from each other.The translator input was time multiplexed betweenthe two interferometers. The time multiplex rate wasset high enough to be outside the bandwidth of thesignal tracking loops so that both inputs could betracked simultaneously when the signals were strongenough. However, the tracking data used by the sys-tem’s Kalman filter were selected to include only track-ing data from the regions of each antenna where thephase effects were well understood (i.e., away from thenull regions). Although this was a reasonable compro-mise within the set constraints, it led to an antennadesign with a very poor gain over a large region (i.e.,the specified gain was more than 14 dB below an idealisotropic antenna, 0 dBi, over 15% of the coverageregion). This poor gain, coupled with the levels of GPSsatellite signals, represented a challenging conditionfor signal tracking.

Signal refraction through the ionosphere at theGPS prime frequency (L1 = 1575.42 MHz) is signifi-cant, and it must be corrected for precision positioningapplications. GPS provides a second frequency signal(L2 = 1227.60 MHz) that is used with the prime fre-quency to compute the needed correction for signalrefraction. Modulations applied to each frequencyprovide the basis for epoch measurements used todetermine the distance to each satellite (range mea-surements). Two range-code modulations are appliedto the L1 frequency, one having a 2-MHz bandwidthand a second having a 20-MHz bandwidth. The L2frequency, however, is modulated only with the 20-MHz bandwidth ranging code.

The strongest GPS signal is the narrow-bandwidthL1 signal; the L2 signal is at one-fourth the power levelof the L1 signal. The wide bandwidth and lower powercharacteristics of the L2 signal, combined with theantenna constraints for C4, precluded its use in theSATRACK I system. Therfore, the C4 translator wasdesigned to use only the narrow-bandwidth GPS sig-nal. The narrow bandwidth code signal is referred toas the clear/acquisition (C/A) code, and the wide-bandwidth code is normally referred to as the protectedor precision (P) code. Sometimes the P code is calledthe P/Y code to indicate that the P code is encrypted.

Since the C4 translator would be using only theGPS L1 C/A signal, another means was needed tocorrect for ionospheric refraction. Because of our de-velopment of and operational experience with Transit,we had an extensive background in ionospheric sci-ence. We had developed and evaluated models of theionosphere, and we understood their limitations. Wetherefore decided to provide a ground-based satellite-like transmitter (i.e., a pseudosatellite) with two

JOHNS HOPKINS APL TECHNICAL DIGEST, VOLUME 19, NUMBER 4 (

THE SATRACK SYSTEM

frequencies. Measurements of these two signals yieldeda measure of refraction along the signal path betweenthe missile and the pseudosatellite. These data gave anestimate of the electron density profile in the regionof the missile flight path. The profile was then used toadjust our best available ionosphere model, which wasthen applied to estimate the refraction correction foreach GPS satellite-to-missile L1 signal path. The pri-mary pseudosatellite signal is similar to the GPS L1signal. The second pseudosatellite signal was set atone-fourth the L1 signal frequency (the L1/4 signal),and it was modulated with a 200-KHz bandwidth rang-ing code. Another natural use of the L1/4 signal wasfor SLBM range safety. By adding extra L1/4 pseudo-satellites at selected range sites, the required (lower-accuracy) real-time trajectory measurement is deter-mined in relation to the three or more pseudosatellitelocations. This range safety system was qualified duringearly C4 flights, which were equipped with both trans-lators and C-band radar transponders. (Although therange safety system shares components with SA-TRACK, it is normally treated as a separate system andwe will not discuss it further.)

The Laboratory provided overall technical direc-tion for the SATRACK system and developed andcontinues to operate the unique postflight processingsubsystem. Missile hardware development was part ofthe Navy’s missile development contract with Lock-heed Missiles and Space Company, and the groundrecording and range safety equipment developmentwas added to the Navy’s contract with Interstate Elec-tronics Corporation.

Another major challenge in SATRACK develop-ment was implementation of the large Kalman filterprocessing technique. To get a head start on develop-ing the required C4 postflight processing software, theproposed methodology was applied to radar data col-lected for Poseidon (C3) missile flight tests. We rec-ognized from the outset that the radar data would beinadequate because of shortfalls in system geometryand velocity measurement accuracy; however, valuableanalysis insights and practical experience would begained in the process. The C3 processing would beapplied to data available from selected earlier Demon-stration and Shakedown Operation (DASO) flighttests and all subsequent C3 DASOs (after March1975) until the beginning of the C4 Missile Flight TestProgram. DASOs were selected because the radar ge-ometry was very weak in the operational test area.

An improved tracking capability was subsequentlyadded to collect Doppler data from missile telemetrysignals to supplement the radar data. This capability,called the Telemetry Doppler Metric MeasurementSystem (TDMMS), was based on the use of telemetrysignal Doppler differences as observed at multiple re-ceiving sites. The hardware to record these data for

1998) 441

Page 7: 1 Thompson

T. THOMPSON, L. J. LEVY, AND E. E. WESTERFIELD

postflight processing was in place to support two C3DASO tests and C4 tests beginning with C4X-13.(The “X” flight designation applies to developmentalflight tests conducted from a launch pad.) In all, sixC3 radar, two C3 radar/TDMMS, two C4X radar, sixC4X radar/TDMMS, six C4-PEM (Production Evalu-ation Missile) radar/TDMMS, and three C4-DASOradar/TDMMS flight tests were processed with themodified SATRACK postflight processing software.The last three C4X flights and all the C4-PEM/DASOflights processed with the radar/TDMMS system werealso processed with the GPS/SATRACK system.TDMMS processing was improved on C4 flights be-yond C4X-19 by using a translator signal, known as thepilot carrier, for the Doppler measurement rather thanthe telemetry signal.

The overlap of radar/TDMMS and GPS measure-ments was extended to overcome the limited numberof GPS satellites available through the early C4 tests;it also provided an opportunity to compare the GPSprocessing results with radar and radar/TDMMS re-sults. In these early C4 tests, the final best estimatesof missile system performance were derived from aprocess that combined the data from all availablemethods. However, once the satellite support reachedthe expected levels, its accuracy was sufficiently supe-rior to the radar/TDMMS that there was no longer anybenefit from the combining process. Radar/TDMMSprocessing yielded only a part of the total validationactivity associated with SATRACK development.

It is easy now to forget how much of this technologywas yet to be proven when SATRACK developmentstarted. GPS was in a concept demonstration phase,the missile translator and digital postflight trackingcapabilities were untested, and the Kalman filter pro-cessing techniques were being extended significantlybeyond the existing state-of-the-art. The SATRACKDevelopment Program included a substantial effort tovalidate all aspects of its accuracy (i.e., we had tovalidate the validation system).6 In addition to theradar/TDMMS work, SATRACK accuracy was vali-dated with a comprehensive simulation system and theuse of a unique SATRACK test satellite.

A hardware-in-the-loop simulation facility wasdeveloped that produced simulated data inputs for allthe major processing elements of the system fromsoftware simulations of GPS satellite motions and atest missile flight trajectory. The software simulationproduced GPS satellite tracking data files, includingsimulated errors, as if they were provided by the GPStracking facilities. These simulated data were sent tothe Naval Surface Warfare Center (NSWC) for pro-cessing (using prototype software being developedfor SATRACK operation). NSWC then produced sat-ellite position and velocity estimates (i.e., ephemeri-des) and clock files that were returned to the APL

442 JOH

postprocessing facility. The missile flight simulationproduced simulated telemetry guidance data, includingerrors, and these were also forwarded to the postpro-cessing facility.

The missile and satellite trajectories, including sim-ulated errors for satellite positions and clocks, werealso used to drive satellite signal generators to producesimulated GPS signals. These, in turn, were passedthrough digitally controlled phase shifters and a timemultiplexing switch to emulate the missile GPS anten-na network. This antenna network simulator was con-nected to a missile translator hardware simulator thatproduced the translated GPS signals at S-band. An S-band antenna hardware simulator produced the out-puts, which were recorded by prototype telemetry sta-tion receiver and recording equipment. The hardwaresimulator drivers were conditioned to encompass allanticipated effects, including signal propagationthrough the ionosphere and troposphere. The recordeddata produced by this simulation capability wereequivalent to the data that would be received from atelemetry site. These data, too, were sent to the post-processing facility.

The postprocessing facility now had all the inputsexpected for an actual flight test: GPS ephemeridesand clock files from NSWC, telemetry data, and thetranslated signal data tape. The facility then processedthese data as if they had come from an actual flighttest and produced an estimate of the underlying modelerrors that could be compared to the model errors usedto produce the simulated data. The process provideda complete test of the processing system to the extentthat the simulations were valid representations of thereal world. Segments of the simulation capability wereused in many test support activities associated withdeveloping the processing hardware and software atthe postprocessing facility. Two complete formal runsof the simulator were used to conclude this element ofthe SATRACK system validation.

A second very important test and validation ele-ment, conceived by the APL SATRACK design team,was based on tracking a satellite configured to act likea C4 missile in its coast phase. Transat was producedby modifying a Transit satellite that was in standbystorage at APL. The modifications included the addi-tion of a structural extension that comprised a C4translator prototype (actually two for redundancy) andan antenna array that matched the performance char-acteristics of the C4 configuration. Once in orbit,Transat provided regularly available missile-like testopportunities for SATRACK system and hardware val-idation tests. Transit navigation system capabilitieswere maintained in Transat so that it could serve asan additional operational navigation satellite when itwas not being used for SATRACK test purposes. TheTransit capabilities were also the basis for independent

NS HOPKINS APL TECHNICAL DIGEST, VOLUME 19, NUMBER 4 (1998)

Page 8: 1 Thompson

THE SATRACK SYSTEM

satellite trajectory measurements that could be com-pared with translator-derived trajectory measurementsfor validation purposes.

Transat was launched in October 1977 before anyGPS satellites were available. In early November, thesatellite was checked out in the Transit mode to verifythat it was ready to support operational navigationusers. It was then switched to the Transat mode, andinitial tests were conducted with the APL pseudosat-ellite. Through May 1978, Transat was primarily usedto test equipment at the eastern and western test rang-es. These tests provided the capability to check realsystem data interfaces using pseudosatellite signalswith an emphasis on range safety system testing. Tran-sat also proved to be an effective tool for checkout ofthe TDMMS. However, validation of GPS trackingconcepts would have to wait until at least two satelliteswere available.

In June 1978, the first missile translator flight test,C4X-17, occurred. Only one GPS satellite was avail-able for this test, and the translator failed after a shortperiod of operation. However, the test was significantto the SATRACK developers because it was the firsttime that we were able to demonstrate that actualtranslated GPS signals could be successfully tracked atthe APL postflight tracking facility using data recordedby one of the range telemetry sites. The C4X-18 flighttest in August 1978 was supported by two GPS satel-lites, and the missile translator worked perfectlythroughout the flight. Since then, translators haveperformed successfully on all flight tests. This testprovided the first opportunity to complete a full missileevaluation with SATRACK and initiated the firstopportunities for Transat evaluations of the GPScapabilities.

The SATRACK processing facility was now oper-ating at an intense level. Transat validation activitieswere just getting started, radar/TDMMS C4 processingwas continuing, normal SATRACK C4 processing wasbeginning, system validation runs were concluding,and we were first beginning to evaluate the actualephemeris accuracy for GPS satellites. The originalestimates for these errors reflected into range and rangerate uncertainties for each satellite line of sight were12 ft and 0.005 ft/s, respectively.

To directly assess the value of these errors duringeach Transat or missile flight test period, data werecollected at an accurately surveyed location. We re-ferred to this data collecting as “static missile tests.”These tests provided a direct observation of the linkrange and range rate errors relative to those computedfrom the satellite ephemeris. The formal simulationvalidation runs were completed in January 1979, anda four-satellite Transat validation test was conductedin March 1979. That Transat test and the static missileresults indicated that the GPS ephemeris was not

JOHNS HOPKINS APL TECHNICAL DIGEST, VOLUME 19, NUMBER 4 (

providing the expected accuracy. In the early days, wecross-checked ephemeris data from NSWC with datafrom the GPS Master Control Station (the officialsystem source) and with data provided by AerospaceCorporation. A careful analysis of all the available dataindicated that the GPS ephemeris errors were about 3times larger than expected for the period coveringC4X-18 (two GPS satellites), C4X-19 (three GPSsatellites), and C4X-21 (four GPS satellites) flighttests.7 The limited number of satellites and the larger-than-expected ephemeris errors were the major diffi-culties; all other aspects of the system performed asexpected.

The process of evaluating sources of ephemeriserrors continued through the end of 1981; 8 to 10different software configurations were evaluated.While this evaluation was being conducted, the initialbaseline ephemeris generation software was beingmaintained through all flight tests to that time. How-ever, in January 1982, a new baseline was selected thatproduced an improvement of about a factor of 2. Tomaintain consistent processing results, all previouslyprocessed C4 DASO and operational test flights(about 20) were reprocessed with the new baselineephemeris generation software. On the basis of thelimitations of the early C4X flight tests, SATRACKprocessing results were compared to radar/TDMMSresults into the early operational tests.8 After the base-line was adjusted, C4 trajectory accuracy achievedwith the SATRACK I system, based on the first 31DASO and operational flight tests, had a positionuncertainty of 35 ft and a velocity uncertainty of 0.09ft/s at body deployment.9

Some finer-grain improvements in ephemeris weresubsequently achieved by adding an additional GPSdata collection site at APL, and Transat supportedrange safety testing as late as May 1982. For the mostpart, the SATRACK I project was complete, althoughtranslators continue to support all C4 flight tests. Inall, translators have successfully supported range safetyand accuracy processing requirements for 154 C4 flighttests through the end of 1997.

SATRACK IIThe development of the Trident II (D5) missile

began in 1981. To support this development, APLconducted the Accuracy Evaluation System study.That study supported development of the technicalobjectives and guidelines document that defined theaccuracy evaluation requirements for the D5 MissileFlight Test Program. The SATRACK II requirements,defined by the Accuracy Evaluation System study,included significant performance enhancements.

It was clear that the D5 requirements could notbe met without a dual-frequency GPS signal tracking

1998) 443

Page 9: 1 Thompson

T. THOMPSON, L. J. LEVY, AND E. E. WESTERFIELD

capability to permit ionospheric corrections. The GPSsignal structures were reviewed in a series of meetingswith the GPS joint program staff. Unlike the L1 signal,the L2 signal could only be modulated by either thenarrowband C/A code or the wideband P code. Thepossibility of temporarily switching to the C/A codemodulation on L2 to support Trident tests was consid-ered impractical because of the effect on other users.Future satellites might have been modified to includethe dual-code capability on both frequencies. Howev-er, the joint program staff suggested an alternative thatwould use a third GPS frequency (L3). This transmis-sion served a nonpositioning purpose. It was used onlyintermittently, it was derived from the same frequencysource as the positioning signals, and it could have theC/A code modulation applied when needed forTrident test support. This approach was agreed to, andit became the baseline GPS signal concept forSATRACK II.

The implementation of a dual-frequency GPS ca-pability naturally affected all other aspects of the sys-tem hardware configuration. The addition of anothersignal channel to the missile translator could increasethe output bandwidth requirement and affect the te-lemetry station recording requirement. However, werecognized that the two GPS signals could be overlaidin a common translator channel. The signals could beseparated during the signal tracking operation by vir-tue of the differences in their code structures, but atthe expense of increased noise in each signal. Anotherbenefit of the overlay was that the SATRACK I re-corder would not need modification. We also realizedthat the range safety tracking capability could be basedon the translated GPS signals, that is, this choicewould eliminate the need for the L1/4 translated signalchannel. The initial system design was based on thecommon channel GPS approach, which we tested atAPL. However, at the completion of the preliminarydesign phase, the SATRACK II baseline was estab-lished with a separate channel L1/L3 configuration, andthe range safety system was based on the GPSL1 C/A signal alone.

Another major effort in establishing the SA-TRACK II baseline concerned the missile antennaconfiguration. A careful study of the phase noise char-acteristics of several candidate configurations eventu-ally led to the choice of a wraparound antenna array.This design was selected because it minimized phasevariations in the missile roll plane, apart from smallangular regions in the nose and tail directions. Thischoice was also important to the range safety config-uration, not because of accuracy considerations butbecause it allowed for smoother, more continuousperformance of the real-time signal tracking loops.

These were important but relatively straightforwardupgrades to the SATRACK I system. All the missile

444 JOH

and telemetry station hardware requirements wereupdated to include the modifications, and the APLsimulation facility was updated to support develop-ment of the postprocessing facilities. The developmentof the upgraded postprocessing facility was then begun.The new missile hardware test equipment configura-tion was again provided within the Lockheed Missilesand Space Company development contract, and thetelemetry station upgrades were provided within theInterstate Electronics Corporation contract.

The major hardware development effort at APLfocused on a redesigned SATRACK postflight trackingconfiguration to enhance accuracy and throughput.The tracking system was upgraded to track 12 dual-frequency links from right- and left-hand circularlypolarized translated signals (i.e., 48 range-code andcarrier tracking loops). Similarly, the tracked data cor-rection and editing software was enhanced to includeautomatic editing and analyst-interactive capabilities.The postflight tracking software was substantiallyimproved using a modular architecture and bettermodeling techniques. However, the most significantdevelopment activity of SATRACK II was the imple-mentation of the cumulative flight test accuracy eval-uation capability. Although conceptualized duringSATRACK I development, a formal theoretical basisfor its design and its subsequent implementation werecompleted as part of the SATRACK II developmentprogram.

The pad-launched developmental D5 flight testsbegan in January 1987 (D5X-1) and ended in January1989 (D5X-20). During that time, GPS satellite sup-port was not yet continuous and not all tests could beconducted within the time period providing maximumsatellite coverage. Furthermore, only the Block II sat-ellites had the dual-frequency signal capability. Theselimitations were mitigated by new dual-frequency pseu-dosatellites, but they, too, were being introduced on therange during the D5X test series. Despite the coveragelimitations, missile trajectory uncertainties at bodydeployment for the first five tests were from 13 to 20ft in position and 0.03 to 0.12 ft/s in velocity. However,from the seventh test on, the uncertainties improvedto 4 to 11 ft in position and 0.006 to 0.014 ft/s invelocity.10 The position and velocity requirements setfor the SATRACK II system were 20 ft and 0.01 ft/s,respectively. Whereas these requirements were met atdeployment, they were being exceeded in the boostregions. Currently, the system is providing velocity ac-curacy below 0.01 ft/s in all flight regions; the positionaccuracy is now less than 3 ft. This performance gainwas primarily due to GPS improvements (more satel-lites and more accurate ephemerides).

The first cumulative evaluation of the D5 MissileFlight Test Program was based on 19 tests that had usedrepresentative production guidance systems. This set

NS HOPKINS APL TECHNICAL DIGEST, VOLUME 19, NUMBER 4 (1998)

Page 10: 1 Thompson

included 4 of the described D5X flights and 15 laterPEM, DASO, and Commander-in-Chief EvaluationTest flights.11 Excellent results were obtained with thisvery early test sample, and subsequent evaluationshave contributed significantly to improvements in theunderlying weapon system error models. A companionarticle (Coleman and Simkins, this issue) describes thesignificant contributions achieved by the cumulativeprocessing methodology.

Other ApplicationsAt about the same time as SATRACK II develop-

ment was beginning, the Range Applications JointProgram Office (RAJPO) initiated development of atranslator for general missile test applications. APLprovided technical support to RAJPO, and they even-tually initiated a contract with Interstate ElectronicsCorporation to produce a ballistic missile translator(BMT) system for general-purpose range applications.The BMT provided an important test capability fornumerous National Missile Defense (NMD) test flightsthat we have been supporting since the early 1990s.A special adaptation of the BMT was also used for twoAir Force Peacekeeper ICBM flight tests that theLaboratory supported.

The first NMD test support we provided was for twoexo-atmospheric reentry intercept subsystem (ERIS)test flights in January 1991 and March 1992. Thesetests demonstrated that differential GPS measurementsbetween an interceptor and target, each equipped witha translator, could resolve the intercept vector geom-etry with submeter accuracy. The direct follow-onproject of postflight tracking and analysis support toNMD is continuing, with the most recent intercept testseries planned into 1999. The ERIS tests acted as aspringboard to a series of independent research anddevelopment (IR&D) activities directed at achievingintercept vector geometry accuracy of less than 2 cm.The ability to achieve that level of accuracy was dem-onstrated with an IR&D project based on the use of arecently developed SATRACK instrumentation.

The U.S. Air Force Peacekeeper test support was adirect spin-off of the Navy Trident work. The post-flight tracking and analysis work had the same generalevaluation objectives for individual tests. However,with only two Peacekeeper flights evaluated usingSATRACK, no basis for cumulative analysis existed.APL and the Charles Stark Draper Laboratory support-ed Peacekeeper contractors (TRW and Rockwell In-ternational) in a successful technology transfer of theNavy guidance evaluation capabilities.12,13

All objectives of the program were met. Of partic-ular importance, the two tests showed that the GPSestimation uncertainties were better than the bestavailable radar-based evaluation approach. Specifically,

JOHNS HOPKINS APL TECHNICAL DIGEST, VOLUME 19, NUMBER 4 (1

THE SATRACK SYSTEM

this test experience demonstrated the superiority ofGPS over radar for evaluating inertial measurementunit hardware errors. In addition, it was concluded thatGPS would provide highly accurate instantaneous im-pact point data for range safety.12,13 Although successwas achieved with the SATRACK approach, thePeacekeeper program was meeting its objectives withthe available radar instrumentation. Therefore, theAir Force had no motivation to change its evaluationapproach. However, Air Force reports recognized thatthe SATRACK methodology uncovered a previouslyundetected initial condition error. This finding led toa reassessment of the gravity model in the launch areaand its influence on the Peacekeeper guidance model.

SATRACK IIIThe Navy will continue to test and evaluate the

Trident Weapon System with the primary goals of de-tecting changes to system performance caused by agingcomponents and assessing system modifications need-ed to extend its lifetime. In this regard, we recognizethat SATRACK evaluation is only one part of systemaccuracy assessment, and accuracy assessment is onlya part of the total weapon system evaluation. Equaldiligence is needed in all aspects of monitoring andmaintaining the Trident system.

Continued D5 accuracy evaluation support willremain the primary objective for SATRACK. Howev-er, in parallel with this activity, we have identified anatural extension of SATRACK capabilities that cansupport precision intercept evaluations for nationaland theater ballistic missile defense flight tests. Thisrealization grew out of our ERIS flight test experienceand our support to the Strategic Defense InitiativeOrganization for the development of a precision inter-cept test capability for the Brilliant Pebbles Program.Both of these projects and the continued support ofNMD test objectives have, with Navy concurrence,taken advantage of the unique APL facilities devel-oped for Trident. As noted in a companion article(Thompson, this issue) on a high-precision sled test,we successfully completed an IR&D project devoted todemonstrating the measurement capability needed forprecision intercept test evaluations. An earlier IR&Dproject developed a translator design for this purposethat was the basis for a new Trident translator systemused for supporting special reentry body tests.

The upgraded SATRACK postflight tracking facil-ity will support existing C4 and D5 translators andreentry body translators as well as the replacementtranslator to be selected for the new D5 test missilekit. When completed, we will refer to this configura-tion as SATRACK III, which will take full advantageof technology growth in processing hardware andsoftware to produce a workstation-based facility that

998) 445

Page 11: 1 Thompson

T. THOMPSON, L. J. LEVY, AND E. E. WESTERFIELD

is easily reconfigured to support a wide range of tests.The capabilities of the new configuration and theexpected reduced test tempo of Trident flight testsnaturally produce a processing capacity that can beextended to support critical ballistic intercept testingof other high-priority defense programs. Our studiesindicate that this capability is required to adequatelysupport model validation of precision missile interceptsystems, and we have configured the postflight track-ing subsystem architecture to be easily expanded toeventually support such tests.

The office responsible for general range applica-tions, RAJPO, is developing a translator-based GPSrange system (TGRS) that includes a capability forintercept support missions. This system is intended toserve both range safety and postflight evaluation ob-jectives for a variety of range users. Many new appli-cations are expected to use TGRS digital translatorsand their ground station recording equipment. A newupgrade to the APL postflight tracking facility willprovide an interface for TGRS data processing in thenear future.

SUMMARYSATRACK has been a significant contributor to

the successful development and operational success ofthe Trident system. It continues to provide a uniquemonitoring function that is critical to the mainte-nance of the Navy’s strategic weapon system. As ournation moves toward a higher reliance on missiledefense, the need to establish equivalent levels ofassurance in those capabilities will require similar skillsand facilities. Although Trident test and evaluation isour primary focus in the evolution of the SATRACKsystem, we are attempting to accommodate emergingrequirements associated with missile defense systemtest and evaluation, as appropriate. SATRACK willcontinue to support the Navy’s strategic weapon sys-tem operations and evolution for the foreseeable fu-ture. The Laboratory looks forward to continued sup-port of the NMD program, and we hope to extend the

446 JOH

use of this technology to a wider range of weaponsystem evaluations.

REFERENCES1Kershner, R. B., “Low Altitude Navigation Satellite System,” in Electronic

and Aerospace Systems EASTCON ’69 Convention Record, Washington, DC,pp. 174–179 (1969).

2Easton, R., “Mid-Altitude Navigation Satellites,” in Electronic and AerospaceSystems EASTCON ’69 Convention Record, Washington, DC, pp. 180–183(1969).

3Woodford, J. B., Melton, W. C., and Dutcher, R. L., “Satellite Systems forNavigation Using 24-Hour Orbits,” in Electronic and Aerospace SystemsEASTCON ’69 Convention Record, Washington, DC, pp.184–189 (1969).

4Levy, L. J., “SATRACK Development,” in APL Developments in Science andTechnology, JHU/APL DST-6, Laurel, MD, pp. 46–49 (1978).

5Levy, L. J., “Requirements Study for a Trident II Accuracy EvaluationSystem,” in APL Developments in Science and Technology, JHU/APL DST-10,Laurel, MD, pp. 104–107 (1982).

6Duven, D. J., “Validation of the SATRACK Processing Facility,” in APLDevelopments in Science and Technology, JHU/APL DST-6, Laurel, MD,pp. 49–52 (1978).

7Anderson, R. J., Levy, L. J., Thompson, T., and Hester, R. B., “Evaluationof SATRACK System Performance,” in APL Developments in Science andTechnology, JHU/APL DST-7, 46–49 (1979).

8Thompson, T., “Performance of the SATRACK/Global Positioning SystemTrident I Missile Tracking System,” in Proc. IEEE Position Location andNavigation Symp., Atlantic City, NJ, pp. 445-449 (1980).

9Thompson, T., “SATRACK—Review and Update,” Johns Hopkins APLTech. Dig., 4(2), 118–126 (1983).

10Vetter, J. R., Schwenk, V. L., and Hattox, T. M., “An Improved GPS-BasedTracking System for High Accuracy Trident II Missile Navigation andGuidance Evaluation,” in Fourteenth Biennial Guidance Test Symp.,Holloman AFB, NM, pp. 67–86 (1989).

11Huneycutt, J. E., Payne, D. A., and Simkins, L. S., “Ensemble BallisticMissile Guidance Accuracy Evaluation Using GPS,” in Fifteenth BiennialGuidance Test Symp., Holloman AFB, NM, pp. 233–247 (1991).

12Peacekeeper/GPS Executive Summary Report, GT06PA, TRW Space andTechnology Group, San Bernardino, CA (1991).

13Peacekeeper/GPS Executive Summary Report, GT07PA, TRW Space andTechnology Group, San Bernardino, CA (1992).

ACKNOWLEDGMENTS: Since its inception about 25 years ago, much ofSATRACK and its applications have been a joint effort of the Space and StrategicSystems Departments. Many APL staff members, Navy sponsors, and commercialfirms have made significant contributions to these activities. We would especiallylike to acknowledge some of the early, and perhaps overlooked, contributors.P. Samuel Sugg of APL’s Strategic Systems Department is the single person mostresponsible for the original vision. He recognized that work supporting advancedsatellite navigation system studies in the Space Department had an importantapplication to Trident test and evaluation objectives. He suggested the conceptin early high-level Navy meetings that led to the original SATRACK proposal,and he initiated the interdepartmental discussions that produced that proposal.R. B. Hester was the program manager who skillfully led the SATRACK programthrough its critical early development phase. M. Brodski and CommanderO. Everett (Strategic Systems Programs Instrumentation Branch, SP25) providedoutstanding leadership and considerable encouragement throughout SATRACK’sformative years. In retrospect, above all others, these four should be recognized asthe prime movers for all that followed. We are sincerely grateful for their uniqueand important contributions.

NS HOPKINS APL TECHNICAL DIGEST, VOLUME 19, NUMBER 4 (1998)

Page 12: 1 Thompson

THE SATRACK SYSTEM

THE AUTHORS

THOMAS THOMPSON received a B.S.E.E. degree from Lafayette College in1960 and an M.S.E.E. degree from The Johns Hopkins University in 1968. Heis a member of APL’s Principal Professional Staff in the Strategic SystemsDepartment. He joined APL’s Space Department in 1960 and served as a groupsupervisor, a branch supervisor, and the chief engineer. Mr. Thompsonparticipated in a wide range of satellite development activities before becomingthe lead system engineer for SATRACK development in 1973. In 1983, hejoined System Planning Corporation, where he led the development ofspecialized radar instrumentation before returning to APL in 1992. At APL, hehas led the development of important upgrades to the SATRACK system forTrident reentry body flight test evaluations, and IR&D activities to demonstratenew GPS measurement capability for missile intercept evaluations. His e-mailaddress is [email protected].

LARRY J. LEVY is the Chief Scientist of APL’s Strategic Systems Department.He received his Ph.D. in electrical engineering from Iowa State University in1971. Dr. Levy has worked on applied Kalman filtering for over 30 years. He wasthe co-developer of the GPS translator concept in SATRACK (a GPS-basedmissile tracking instrumentation system) and was instrumental in developingthe end-to-end methodology for Trident II accuracy evaluation. He teachesgraduate courses in Kalman filtering and system identification at The JohnsHopkins University Whiting School of Engineering and Kalman filteringseminars for Navtech Seminars. His e-mail address is [email protected].

EDWIN E. WESTERFIELD attended the University of Maryland where hereceived a B.S.E.E. degree in 1952 and an M.S.E.E. degree in 1963. He is amember of APL’s Principal Professional Staff in the Strategic SystemsDepartment. He joined APL in 1954 and was involved with development of theTalos guidance system and led the development of a large telemetry facility. In1960 he transferred to the Space Department, where he led the development ofa relative navigation system using signals from the Navy Navigation SatelliteSystem. On initiation of the SATRACK program, he became responsible for thedevelopment of much of the hardware and also served as a group supervisor. Hetransferred to the Strategic Systems Department in 1990, where he is GroupSupervisor of the SATRACK/GPS Systems Group. Mr. Westerfield serves as theProgram Manager of the National Missile Defense Precision Missile TrackingProgram and of the GPS Range Applications Joint Program. His e-mail addressis [email protected].

JOHNS HOPKINS APL TECHNICAL DIGEST, VOLUME 19, NUMBER 4 (1998) 447