System con guration of GPS showing the three fundamental segments. (Drawing courtesy of U.S. Air Force.)
becomes the Y (or P/Y) code. (There are provisions in the Federal Radio-Navigation Plan for users with critical national needs to gain access to the P code.) a) Selective availability. The military operators of the system have the capability to intentionally degrade the accuracy of the C/A signal by desynchronizing the satellite clock or by incorporating small errors into the broadcast ephemeris. This degradation is called selective availability (S/A). The magnitude of these ranging errors is typically 20 m, and results in rms horizontal position errors of about 50 m (one-sigma). The of cial DOD position is that SPS errors will be limited to 100 m (two-dimensional rms), which is about the 97th percentile. A technique known as differential GPS (DGPS) can overcome this limitation and potentially provide civilian receivers with accuracies suf cient for precision landing of aircraft. b) Data modulation. One additional feature of the ranging signal is a 50-bps modulation, which is used as a communications link. Through this link, each satellite transmits its location and the correction that should be applied to the spaceborne clock, as well as other information. (Although the atomic clocks are extremely stable, they are running in an uncorrected mode. The clock correction is an adjustment that synchronizes all clocks to GPS time.) 3.
a) Satellite orbital con guration. The orbital con guration approved at DSARC in 1973 was a total of 24 satellites, 8 in each of three circular rings with inclinations of 63 deg (see Fig. 7). The rings were equally spaced around the equator, and the orbital altitudes were 10,980 n mile. This altitude gave two orbital periods per sidereal day (known as semisynchronous) and produced repeating ground traces with satellites positioned 4 min earlier each day. The altitude was a compromise among user visibility, the need to periodically pass over the Continental U.S. ground/upload stations, and the cost of the U.S. Air Force spacecraft launch boosters. Three rings of satellites were initially selected because it would be easier to have orbital spares; having only three such spares (one in each ring) would allow easy replacement of any single failure in the whole constellation. This con guration provided a minimum of 6 satellites in view at any given time, with a maximum of 11. As a
Fig. 7 Original GPS orbital con guration of three rings of eight satellites each. The nal operational con guration has the same number of satellites arranged in six rings of four satellites. (Drawing courtesy of U.S. Air Force.)
result of this redundancy, the system was robust in the sense that it could tolerate occasional satellite outages. Two changes have since been made to the original constellation proposal. The inclinations have been reduced to 55 deg, and the number of orbital planes have been increased to six, with four satellites in each. The number of satellites, including spares, remains 24. This constellation still gives six or more satellites in view all the time, virtually everywhere in the world. This assumes that there are no outages, and all satellites more than 5 deg above the horizon can be seen.
Fig. 8 Breakaway view of the GPS phase one satellite design. Satellite characteristics: weight at booster-satellite separation = 1636 lb, weight at insertion into nal orbit = 982 lb, antenna span = 17.5 ft, design life = 5 years, and life of consumables = 7 years. (Drawing courtesy of U.S. Air Force.)
Fig. 9 Space-quali ed rubidium-cell frequency-standard performance. These units were developed by Rockwell as a derivative of a clock designed by Efratom Inc. under contract to the NRL. (Data courtesy of U.S. Air Force.)
In the U.S. typical coverage is 6– 8 or more in view. The minimum number of 4 satellites (to navigate) is therefore reasonably assured. Many users would like to have 6 or more to perform cross checks on the integrity of the positioning solution. To attain this at all times is obviously more dif cult. This so-called availability problem will be addressed later. b) Satellite design. All GPS satellites are attitude stabilized on all three axes and use solar panels as the primary power source. The ranging signal is radiated through a shaped beam antenna; by enhancing the received power at the limbs of the Earth, compensation is made for space losses. The user therefore receives fairly constant power for all local elevation angles. (The requirement for received power on L1 is ¡163 dBW into an isotropic, circularly polarized antenna on the primary frequency.) The satellite design is generally doubly or triply redundant, and the phase one satellites have demonstrated average lifetimes in excess of 5 years (and in some cases over 12). This satellite design is shown in Fig. 8. c) Satellite autonomy: atomic clocks. A key feature of the GPS design is that the satellites need not be continuously monitored and controlled. To achieve this autonomy, the satellites must be
predictable in four dimensions: three of position and one of time. Predictability in orbital position is improved because the highaltitude orbits are virtually unaffected by atmospheric drag. Many other factors that affect orbital position must also be considered. For example, lunar– solar perturbations, solar pressure, and outgassing can all have signi cant effects. When GPS was conceived, it was recognized that the most dif cult technology problem facing the developers was probably the need to y accurate long-lived timing standards to ensure that all of the satellites’ clocks remained synchronized. As mentioned, the TIMATION program had been developing frequency standards for space, and so this effort was continued and extended. GPS has traditionally used two types of atomic clocks: rubidium and cesium. Phase one test results for the rubidium cell standard are shown later. A key to outstanding satellite performance has been the stability of the space-quali ed atomic clocks, which have exceeded their speci cations. They have measured stabilities of one part in 1013 over periods of 1 – 10 days.9 Test results are shown in Fig. 9. d) Ionospheric errors and corrections. Free electrons in the ionosphere create a delay in the modulation signal (PRN code). It is
not unusual to nd delays of over 30 m at lower satellite elevation angles. There are two techniques for correcting this error. The rst is to use an ionosphere model. The model parameters are broadcast as part of the GPS 50-bps message. This model is typically accurate to a few meters of vertical ranging error. The second technique uses both broadcast frequencies and the inverse square law behavior to directly measure the delay. By differencing the code measurements on each frequency; the delay on L1 is approximately 1.546* (difference in delays on L1 and L2). (The delay at 1575 MHz is found as the difference in delay multiplied by [ f 22 =. f 12 ¡ f 22 /] because it is proportional to the inverse frequency squared. The frequency of L2 is 1227 MHz.) This technique is only available to a P/Y code receiver (since only the P code has L2 modulation) or to a codeless (or cross-correlating) receiver.
Incubation and Birth: GPS from 1973 to 1978
Approval to Proceed with GPS
To gain approval for the new concept, Parkinson began to contact all those with some stake in the decision. After interminable rounds of brie ngs 16 on the new approach were given to of ces in the Pentagon and to the operating armed forces, a successful DSARC was held on Dec. 17, 1973, only three months after GPS was conceived. [Gen. Kenneth Schultz was particularly incensed with the endless presentations that had to be made in the Washington arena. The situation with any bureaucracy is that many can say no and few (if any) can say yes. To bring the nay sayers to neutral, extended trips from Los Angeles to Washington were necessary for Parkinson.] Approval to proceed was granted in a memorandum dated Dec. 22, 1973. B.
Fast Start to Development
The rst phase of the program originally included four satellites (one was the refurbished quali cation model), the launch vehicles, three varieties of user equipment, a satellite control facility, and an extensive test program. By June of 1974, the satellite contractor, Rockwell International, had been selected, and the program was well underway. Magnavox, which had been a key participant in the user equipment for 621B (along with Hazeltine), was selected to develop the user equipment under subcontract to General Dynamics, which was also responsible for developing the satellite control segment and the pseudosatellites for the Yuma range. The initial types of user equipment included sequential (the Y set) and parallel (the X set) satellite-tracking military receivers, as well as a civilian-type set for utility use by the military (the Z set).
The development test and evaluation was extensive, with a laser tracking range set up at the Army’s Yuma Proving Ground. An independent evaluation was then performed by the Air Force’s Test and Evaluation Command. To maintain the focus of the program, the GPS JPO adopted the following simple and direct motto. The mission of this program is to: 1) drop 5 bombs in the same hole and 2) build a cheap set that navigates (<$10,000 ), and don’t you forget it! The $10,000 price goal was considered very ambitious! The program developed rapidly; the rst prototype operational satellite was launched in February 1978 (44 months after contract start). The design is shown in Fig. 10. By this time, the initial control segment was deployed and working, and ve types of user equipment were undergoing preliminary testing at the Yuma Proving Ground. The initial user equipment types had been expanded to include a ve-channel set developed by Texas Instruments and a highly jam-resistant set developed by Rockwell Collins. C.
Further Technology Development
Needed: A Few More Good Satellites
The NAVSTAR GPS Development Concept Paper 133, approved in 1973, called for NRL to continue the technology efforts begun in the TIMATION project under JPO direction. A cesium-standard satellite clock was selected as the most promising candidate for meeting the ultimate requirements of the program. The NRL program was to aid in demonstrating the feasibility of the system concept by constructing NTS and to advance the state of the art in navigation satellite technology. The second technology satellite built by NRL, NTS-II, was the rst satellite launched in the NAVSTAR GPS program that was speci cally built to the GPS concept. Shown being placed atop the Atlas F launch system at Vandenberg in Fig. 11, it was launched in June 1977. NTS-II contained the rst two prototype cesium-beam frequency standards to be own in space. An essential component was the navigation payload, which was provided by the JPO. While NTS-II was short lived, the performance of the NTS-II units was a frequency stability of 2 parts in 10¡13 per day, giving a time error of about 20 ns a day. Because only four satellites were initially approved by the DOD, including a refurbished quali cation model, there was a need for spare satellites. (Recall that the minimum number for
Fig. 10 Phase one GPS satellite is a three-axis stabilized design with double and triple redundancy where appropriate. (Drawing courtesy of U.S. Air Force.)
name. Brentnall passed this along as a good idea to Parkinson, noting that if Walsh were to name it, he would undoubtedly feel more protective toward it. Parkinson seized this opportunity, and since that time the program has been known as NAVSTAR, the Global Positioning System. While some have assumed that NAVSTAR was an acronym, in fact it was simply a pleasant name that enjoyed the support of a key DOD decision maker. [It is noted that TRW apparently had advocated a navigational system for which NAVSTAR was an acronym (Navigation system timing and ranging). This may have been in Walsh’s subliminal memory, but was not part of the process. It was never used as an acronym.]
Coming of Age: GPS from 1978 to the Present
In the early years, the GPS program enjoyed only lukewarm support from the U.S. Air Force because it was viewed as a DOD program. As a result, many attempts were made to cancel its development. This led to key actions to broaden its usefulness and, hence, its support. Initially, it was felt that the full system could be made operational by 1984. Principally due to funding restrictions and redirection, the system was not able to become operational until 1994. Had the phase one satellites simply gone into production, this could have occurred almost 10 years earlier. Fig. 11 NTS-II being placed on an Atlas-F booster at Vandenberg Air Force Base.
three-dimensional navigation is four.) Any launch or operational failure would have gravely impacted the phase one demonstration program. Authorization for additional GPS satellites was urgently needed! The Navy’s Transit program inadvertently solved this problem. This chain of events unfolded when Transit requested funds for upgrading certain Transit satellites to a PRN code similar to that used by GPS. The purpose was to provide accurate tracking of the Trident (submarine-launched ballistic missile) booster during test rings into the broad ocean areas. Robert Cooper of DDR&E requested a series of reviews to address whether GPS could ful ll this mission. The GPS solution was to use a signal translator on the Trident missile bus, which would relay the GPS modulations to the ground on another frequency. The central issues were whether the ionosphere could be adequately calibrated (because it was a single-frequency system, the ionosphere could not be directly measured) and whether the translated signal could be recorded with suf cient delity (it required digitizing at 60 MHz). During the third (and capstone) review for Cooper, Parkinson (supported by James Spilker and John Klobuchar) was able to present convincing arguments that a GPS solution could solve the Trident problem provided two additional satellites were authorized. Cooper immediately made the decision to use GPS. He directed the transfer of $60 million from the Navy to the Air Force. This approved two additional satellites and thereby greatly expanded the phase one test time as well as signi cantly reduced program risk. This littleknown event also eliminated the possibility of an upgraded Transit program competing with the edgling GPS. E.
Why Is It Called NAVSTAR or GPS?
There has been much speculation regarding the origin of the names GPS and NAVSTAR. The GPS title originated with Gen. “Hank” Stehling, who was the Director of Space for the Air Force Deputy Chief of Staff research and development in the early 1970s. He pointed out to Parkinson that navigation was an inadequate descriptor for the proposed concept. He suggested that Global Positioning System or GPS would be a better name. The JPO enjoyed his sponsorship, and this insightful description was immediately adopted. The title NAVSTAR came into being in a somewhat similar manner. John Walsh (an Associate Director of DDR&E) was a key decision maker when it came to the budget for strategic programs in general, including the proposed satellite navigation program. In the contention for funding, his support was not as fervent as the JPO would have liked. During a break in informal discussions between John Walsh and Col. Brent Brentnall (the program’s representative at DOD), Walsh suggested that NAVSTAR would be a nice sounding
Add Ons and Delays, the Perils of Innovation
To expand political support, additional operational payloads were incorporated into the baseline design. From its vantage point in space, the NAVSTAR satellites potentially could time nuclear explosions on the ground from many directions and pinpoint their locations. This capability was added along with various others. These innovations were essential for continuing support but also contributed to delays in the program because they required extensive modi cations to the satellites. Steady progress has been made in GPS satellite development, but the real revolution has been the diminishing cost and increased capabilities of GPS user equipment. To protect the U.S. against hostile use of GPS by enemies, the civilian signal is deliberately degraded in accuracy as described earlier. The name given this degradation is S/A. This action has been a source of international tension over the potential adaptation of GPS as an of cial navigation standard. Because these effects can be calibrated out by a receiver at a known location on the ground (see later discussions of DGPS), the effectiveness of this technique to deny accuracy has come under severe questioning.
Status of GPS: Performance and Test Results
GPS offers two standard services in terms of accuracy. As mentioned previously, the service for military users is known as the PPS and is not affected by the deliberate degradation of S/A. The civilian service is known as the SPS. Both of these standard applications rely on the tracking of the modulation (or code) with no ground-based corrections. More sophisticated techniques apply range corrections computed from ground receivers located at known positions. These so-called differential techniques can attain relative accuracies of better than a centimeter. The following sections brie y summarize the current situation. A.
Standard GPS Navigation Performance
The performance capabilities of standard GPS are primarily affected by two things: satellite geometry (which causes geometric dilution) and ranging errors. Under the assumption of uniform, uncorrelated, zero-mean ranging error statistics, this can be expressed as position error D .geometric dilution/ ¤ .ranging error/ 1.
Satellite Geometry: Geometric Dilution Effect
Geometric dilution can be calculated for any instantaneous satellite con guration as seen from a user at a particular location. For a 24-satellite constellation and a three-dimensional x, the world median value of the geometric dilution factor (for the nominal constellation) is about 2.5. This quantity is usually called position dilution of precision (PDOP). Typical dilution factors can range from 1.5 to 8. The variations in this dilution factor are typically much greater than the variations in ranging errors.
Ranging errors are generally grouped into six major causes: 1) satellite ephemeris, 2) satellite clock, 3) ionospheric group delay, 4) tropospheric delay, 5) multipath, and 6) receiver measurements. With S/A turned off, all errors for single-frequency SPS are nearly identical in magnitude to those for single-frequency PPS except for receiver measurement errors (which decrease with increasing bandwidth). Dual frequency, which is only available on PPS, can reduce the third error (due to the ionosphere) to about 1 m. This is summarized in Table 1. (Multipath errors are generally negligible for path delays that exceed one and one-half modulation chips, expressed as a range. Thus, P code receivers reject re ected signals whose path delay exceeds 150 ft. For the C/A signal, the number is 1500 ft, giving a slight advantage to the P code. However, it is usually re ections from very close objects that are the main sources of dif culty.) 3.
Expected Positioning Accuracies for PPS and SPS
The product of the rms PDOP and the ranging error for a single satellite gives the three-dimensional rms position error. Because the value of rms PDOP averages 2.5 – 3.0 (depending on user location and the assumptions on minimum visible satellite elevation angle, or mask angle), one can multiply 2.5 times the values in Table 1 to nd typical positioning errors. Note that PDOP is the same for all users, civilian and military. A second way to specify accuracy is by spherical error probable (SEP). The SEP is de ned to be the radius of the sphere that contains 50% of the errors. For horizontal errors only, similar concepts of circular error probable (CEP) and horizontal rms error can be used to specify accuracy. Typically, the horizontal rms error is about 1.2 times the CEP. Without the degradation of S/A, SPS would provide solutions with about 50% greater error than single-frequency PPS due to uncompensated ionospheric effects and somewhat greater receiver noise (due to the narrower band C/A signal). It is reasonable to expect that rms horizontal errors for SPS with S/A off would be less than 15 m. 4.
Positioning Accuracy Summary for Code Tracking Receivers
High Accuracy/Carrier Tracking Receivers
Table 2 summarizes the expected positioning accuracies for GPS. It includes a conservative allowance for ionospheric errors. A special feature of GPS, which initially was not generally understood, is the ability to obtain an extremely precise ranging signal by reproducing and tracking the rf carrier wave (1575.42 MHz). Because this signal has a wavelength of 19 cm (7.5 in.), tracking it to 1/100th of a wavelength provides a precision of about 2 mm. Generally, carrier tracking techniques can be used in two ways. For Table 1 Nominal ranging errors for various classes of service Ranging accuracy, m Single freq. Dual freq.
Class of service PPS SPS no S/A SPS with S/A
5 6 20
3 N/A N/A
Expected positioning accuracies for various GPS operating conditions PPS Spec. Measured value, m (static), m
Ranging accuracy CEP (Horiz.) SEP three-dimensional
SPSa Est. capability No With S/A, m S/A, m
a SPS results are believed to be conservative. b For dynamic PPS users reported to be less than
normal use, carrier tracking can smooth code tracking and greatly reduce the noise content of code ranging measurements. The other use of carrier tracking is in a differential mode. There are several variations of this, including surveying, direct measurement of vehicle attitude (with multiple antennas), and various forms of dynamic differential. Modern receivers can attain 2-mm tracking precisions for this second use, but unfortunately this is not accuracy. Re ected signals (multipath) and distortions of the ionosphere can be signi cant errors (i.e., on the order of a few centimeters). In addition, to provide centimeter accuracy, one must determine which carrier cycle is being tracked (relative to the start of modulation) and compare this with another carrier tracking receiver located at a known position. The technique used to do this depends on the application. Surveyors, averaging over time for a static position, use techniques known as double or triple differencing to resolve this cycle ambiguity. For dynamic users (who require real-time positioning), resolving cycle ambiguities is a bit harder. C.
Worldwide Test Results for the PPS
Because each of the ve worldwide GPS monitoring stations is continuously measuring the ranging errors to all satellites in view, these measurements are a convenient statistic of the basic, static accuracy of GPS. Table 3 summarizes over 11,000 measurements taken from Jan. 15 to March 3, 1991 during the Desert Storm operation of the Gulf War.10 The S/A feature was not activated during this period. Note that the PPS results presumably would not be affected by S/A at all. During this period, one satellite (PRN 9) was ailing but is included in the solution, making the results somewhat worse than they would otherwise be. By dividing the overall SEP by the rms PDOP, an estimate of the effective ranging error can be formed. The average of these results is 2.65 m. [This number is probably somewhat better than an average receiver would measure for several reasons. Monitor station receivers are carefully sited to avoid multipath. The receivers are of excellent quality and are not moving. Also, since the monitor station measurements are used to update the ephemeris, there may be some tuning to make the predictions match any peculiarities (e.g., survey errors) at the monitor station locations. Nonetheless, an average ranging error of 2.3 m is an impressive result.] This should be compared to the speci cation of 6 m. D.
Local Area DGPS
The ultimate level of accuracy for GPS is attained by calibrating ranging errors with direct, real-time measurements from a calibrating receiver at a known location. These corrections are then applied in the user’s receiver to eliminate correlated errors. Errors correlated between user and reference are the dominant sources, particularly when S/A is activated. These techniques are called DGPS since the resulting position is effectively a measurement that is relative to the assumed position of the calibrating receiver. For a detailed explanation of DGPS see Ref. 11. Not addressed here are the various techniques for transmitting corrections to the user. This is an important consideration in the design of a system and, due to delays or bit errors, can be the largest source of DGPS error. The next sections will summarize some of the various DGPS techniques and provide estimates of their accuracy. This is the simplest and most widely used technique. A receiver is placed in a known location and solves for the corrections (as ranges) to each of the satellites in view. These are then transmitted in some convenient way to the user, who applies them to more accurately solve for his location. This is illustrated in Fig. 12. A well-designed DGPS system using code measurements can correct ranges down to 0.5 – 2 m of residual error. When coupled with geometric dilution (note that it still applies), the resulting navigation or positioning errors are on the order of 1 – 5 m, provided that the user is within 50 km of the calibrating receiver. The ef cacy of corrections at longer ranges is discussed in the next section. The U.S. Coast Guard has deployed a local area DGPS system which uses marine radiobeacons [nondirectional beacons (NDBs)] as data links. These links have added a modulation to the beacon
PPS accuracies and implied ranging errors, measured by the GPS monitor stations during desert storm; S/A is off
SEP three-dimensional , m CEPa two-dimensional , m RMS PDOP Est. range error, mb
a CEP equals the radius of a circle that would contain 50% of the errors. It is the two-dimensiona l analog of SEP. b This row is formed as follows. Horizontal dilution of precision (HDOP) is approximatel y PDOP divided by 1.7,
and CEP is rms horizontal error divided by 1.2. Because ranging error is horizontal rms error divided by HDOP, this is approximate d as 2.04 times CEP divided by PDOP. c Note that this is slightly larger than the result in Table 2, perhaps because of the ailing satellite 9.
Fig. 13 Carrier DGPS using the Stanford IBLS; accuracies of 2 – 4 cm are routinely attained.
Local area DGPS, the simplest form of DGPS.
signal, which provides the corrections in a standardized format. These can be used by marine and other users. Accuracies of 5 m or better have been attained. Even greater accuracy can be attained by user carrier-tracking receivers provided that the cycle ambiguity can be resolved. There are several ways to do this. A technique that has been demonstrated at Stanford University uses simple, low-power ground transmitters (called integrity beacons) to resolve cycles with very high success rates. If the cycles have been resolved, the ranging errors are reduced to a few centimeters, and dynamic position xes of better than 10 cm have been demonstrated. This technique [called carriertracking DGPS (CDGPS)] is shown as part of the Stanford integrity beacon landing system (IBLS) in Fig. 13. 2.
Wide Area DGPS
To increase the operating area of DGPS, a technique called wide-area DGPS (WADGPS) is used. This concept uses multiple calibrating (or reference) receivers to develop vector corrections for the various error sources. Existing systems cover the Gulf of Mexico and the South China Sea. Accuracies of 3 – 5 m have been reported using WADGPS. The U.S. Federal Aviation Administration (FAA) is currently embarked on a program to eld a WADGPS system called wide area augmentation system (WAAS). It will cover the U.S. and provide corrections through a satellite data link to aircraft and other users. 3.
Time and Frequency Transfer
The pioneering use of GPS was probably time and frequency transfer, because a single satellite within the common view of two receivers can easily resolve time to the microsecond or better. This is not surprising since TIMATION was initially focused on providing precise time transfer. The current time transfer capability is reported by the U.S. Naval Observatory to be 20 ns or better.
Fig. 14 Range of accuracies attainable with various forms of GPS: horizontal accuracy (1¾). 4.
Summary of GPS Accuracies
Surveying was the rst commercially signi cant market for GPS because periodic daily coverage provided important economic bene ts. By 1986, commercial application in this eld was in full swing, and surveying pioneered many of the techniques later employed by dynamic DGPS users. Surveyors rely on carrier tracking and resolve ambiguities by observing satellite motion over times of 30 min to an hour. Commercially advertised accuracies are 1 – 2 mm plus one part in 106 of range, and this performance is routinely attained. A newer technique called kinematic survey promises much more rapid resolution of integers using dual-frequency receivers. A summary of the full range of GPS capabilities is shown in Fig. 14.
Fig. 15 Antenna con guration for measuring aircraft attitude using CDGPS.
operation of the Soviet low-orbital SRNS Tsikada described above. Results of fundamental research on high-precision orbit prediction, general relativity effects, increasing long-term stability of spacequali ed atomic clocks, refraction of radio waves in the troposphere and ionosphere, and digital signal processing techniques were incorporated into the design of the new system. After a large-scale study, the research, development, and experimental work on elements of the system were completed and deployment began. The rst space vehicles of the GLONASS series (Cosmos-1413, Cosmos-1414, and Cosmos-1415) were launched into orbit on Oct. 12, 1982. The size of the space segment of GLONASS has continually increased, with one to two launches each year at rst and even more launches in the last few years. By the end of 1988, there were 6 completely functioning satellites in orbit, which were enough to begin full-scale testing of the system. By 1991, there were 12 functioning satellites, enough to give continuous global two-dimensional position xes. The GLONASS is intended to provide location, velocity, and precise time for naval, air, land, and other types of users. Like GPS, it was designed for unlimited use by military and civilian users. Like GPS, the GLONASS is open for use by foreign users as well. It has been pledged that the system will keep its basic characteristics unchanged and will be free to the world for at least the next 15 years. These proposals were given by Soviet (Russian) representatives to International Civil Aviation Organization (ICAO) on May 9, 1988, along with simultaneous disclosure of all technical characteristics necessary for development, manufacturing, and operation of user equipment. GLONASS has now been declared operational. It is planned that by the year 2000 GLONASS will be the basic navigation and precise time aid for all vehicles in Russia.
VII. Fig. 16 E.
Accuracies of vehicle attitude measurements using CDGPS.
A variation of DGPS uses carrier tracking and multiple receiver antennas attached to a single receiver to dynamically measure vehicle attitude. A typical con guration is shown in Fig. 15. This application was pioneered and demonstrated by Clark Cohen of Stanford University. The capability to measure attitude is particularly important for vehicle control applications. The accuracy of the measurement is a function of several parameters, including baseline length and sampling rate. Figure 16 shows these tradeoffs. Typical aircraft accuracies have been demonstrated to be 0.1 deg (one sigma) for all three axes. Attitude rate accuracies of 0.5 deg/s (one sigma) are also attainable.
Cold War Responses: Crickets and GLONASS
Because Transit and GPS both were principally developed for U.S. military purposes, a response from the (then) U.S.S.R. was not surprising. The Soviet systems will now be brie y described. Since the breakup of the U.S.S.R., these navigation systems have continued to be supported by Russia. A.
Russia’s Cicada (Tsikada)
Russia’s Global Navigation Satellite System
Tsikada (Cicada) is a passive Doppler satellite navigation system similar to the U.S. Transit system. The motivation for its development was similar as well: its principal use is for warship navigation. The orbiting Tsikada system usually consists of 4 active satellites of the Cosmos-1000 type. These satellites have an orbital inclination of 83 deg with an orbital period of 105 min. Like Transit, they broadcast signals on two frequencies: 150 and 400 MHz. The Tsikada operational concept is very similar to Transit, with typical horizontal position accuracies of 0.2 miles (two-dimensional rms). Tsikada provides all-weather global coverage for an unlimited number of users. The system continues in operational status: a rocket launched from Plesetsk on Jan. 24, 1995, carried a new Tsikada satellite as one of its payloads. Development of the Russian (Soviet) satellite radionavigation system (SRNS), the global navigation satellite system (GLONASS), started in the 1970s on the basis of the development and successful
Selected GPS Applications
Acceptance of GPS has been accelerating. For example, commercial GPS sets are currently being manufactured at a rate of over 60,000 per month. Almost half of these are going into Japanese automobiles. Commercial DGPS is also increasingly available. The U.S. Coast Guard is completing its coverage of all major waterways, including the Great Lakes and the Mississippi River, with the NDB (radiobeacon) version of DGPS. In the air, the FAA has begun the certi cation of GPS avionics. As the number of GPS receivers accelerates, so have the applications. Rather than repeat the usual applications, let us consider some of the more unusual uses that have been reported. A.
Unique and Unusual Applications
There have been many published lists of GPS applications. Most obvious are the usual navigation uses. Aircraft, ships, trucks, and backpackers are included in virtually all phases of motion or location. As fertile imaginations grasp the potential of the ninth utility, additional innovative applications appear. The following are some examples. A major commercial application is the use of FM stations to inexpensively broadcast DGPS corrections as an additional modulation that is invisible to the normal listener. With coverage extending across virtually the whole U.S., the additional utility of 2 – 5 m accuracy is enormous. At least two companies are pursuing this. The magazine GPS World has an annual contest for unique and unusual applications of GPS from which these examples are extracted. Here is a sample of some of the recent winners. 1) Tracking sheep with GPS is used to correlate their eating habits with the radioactive fallout from the Chernobyl accident. One x per minute is taken that is accurate to within 5 m over an extended period. 2) DGPS is used to map malaria outbreaks in Kenya. Research should indicate areas and circumstances to be avoided, hopefully reducing the incidence of the disease. 3) Tracking and coordinating the movements of large, parallel overhead cranes, which are used to move lumber, are used to prevent adjacent cranes from crashing into each other. 4) Oil spills are tracked using buoys that are equipped with GPS and a radio system to notify an oil spill response team of the location. Of course, the buoy will tend to drift with the spill. 5) The location and evaluation of the health of electrical power poles are tabulated. This replaces an error-prone manual
data-capture process and is an example of the extensive use of GPS for geographic information systems. B.
Worldwide Humanitarian Use
Throughou t the world, con icts of today and yesterday have in icted a terrible legacy on the landscape. That legacy is over 20 million buried land mines. The current situation in Bosnia is only the latest; Cambodia, Kuwait, and Somalia are still in our recent memory. According to the Public Broadcasting System, one person in 280 in Cambodia has been injured by a land mine. A potential application of GPS is the clearing and identi cation of safe corridors through these mine elds. The submeter positioning capability of DGPS can ensure identi cation of cleared areas. Robotic devices for clearing mines could be operated under closed-loop control with DGPS. It may be ironic that a system conceived for war could be used for such an important peacekeeping application.
Challenges for GPS
Although GPS was designed to be robust, expanded expectations have illuminated aspects that probably call for increased capabilities or resilience. The following sections outline these challenges. A.
Air: Integrity Challenge
Integrity is the technical term used by the FAA to describe the con dence in a measurement of aircraft position. It is usually measured as integrity risk, which is the probability that the error of the indicated position exceeds some threshold error value. There is great sensitivity to this aspect of navigation, because a signi cant position error can lead to substantial loss of life and erosion of con dence in airline travel. 1.
While autolanding an aircraft (FAA category III), the positioning errors cannot exceed a maximum safe error limit more than once in a billion landings. Whereas this is the most stringent stated requirement, virtually all civil uses have some implied or stated integrity speci cation. Therefore, the challenge is to provide a positioning signal that meets these dif cult requirements. GPS satellites will eventually fail and create holes in the constellation. These outages could be a major cause of reduced system integrity until they can be replaced. A rash of generic satellite failures would be dif cult to immediately replace.
2) Local geography may mask the satellites. This includes mountains, buildings, and vegetation. 3) Vehicle attitude may cause the user antenna pattern to have insuf cient gain in certain directions. 4) Local radio interference may prevent the user from receiving the GPS signal. 2.
Potential Integrity Solutions
Potential solutions to this challenge are being explored by the FAA and others. They include the following. 1) Ground monitor: all DGPS systems are naturally integrity monitors. 2) Cross check in the user receiver [called receiver autonomous integrity monitoring (RAIM)] using redundant ranging measurements from more than the minimal set of satellites. Usually, at least six measurements are needed for a high con dence in integrity. 3) Additional or supplemental navigation satellites can be of enormous bene t, especially when possible outages are considered. Strong arguments can be made for civil supplements. B.
All Users: Availability Challenge
For most users, four satellites must be available for a navigation solution. If the user is to determine integrity using cross checks with redundant satellites (RAIM), generally 6 satellites or more must be available. Availability is de ned to mean that the necessary ranging signals are available to commence an operation requiring positioning at a speci ed performance level. Availability requirements are determined by the particular application. For some aircraft applications, better than 99% availability is required. Less than the minimum required number of satellites (4 – 6) may be available for a variety of reasons. 1) Satellite outages produce a hole in the usual coverage. As the satellites continue their orbits, this hole will move, so that the local outage will not be permanent.
Ground: Continuity Challenge
Continuity for positioning systems means that an operation is not interrupted because of a lapse in measurements after the operation begins. This is somewhat different from availability, which requires that positioning measurements are available at the beginning of an operation. Continuity and integrity may be, to different degrees, safety issues, whereas availability tends to address the economy or ef ciency of a system. 1.
Continuity is particularly a problem for ground users due to intermittent shading of the satellite signals. Travel through cities exposes the ground user to urban canyons, which may limit the number of satellites to one or two. Tilting of the ground vehicle can aggravate the continuity problem because of modi ed antenna coverage. Complicating the reduced coverage problem is the need to reacquire GPS after an outage. 2.
Potential Availability Solutions
A user can reduce the availability problem by increasing his antenna coverage (e.g., dual antennas), or by using other measurements (such as precise time or altitude) to reduce the requirement for four satellites in view. A more universal solution to this problem is to supplement GPS with additional ranging sources. Supplementary satellites could be placed in GPS type orbits or at geosynchronous altitude. By making the satellite coverage more dense, both outages and local shading impacts would be lessened. Another solution, particularly useful for the landing of aircraft, is the use of GPS type transmitters from the ground. These are generally called pseudosatellites or pseudolites. One particular type of pseudolite is the Stanford integrity beacon, which also provides a means of resolving integers for the highly accurate CDGPS aircraft landing system shown in Fig. 13.
Potential Continuity Solutions
Ground users can supplement GPS with wheel counters and magnetic compasses to automatically ywheel through low-visibility periods. Additional GPS satellites (or supplementary payloads on other satellites as already described) would provide substantial help. Ground transmitters will only be useful in a local area because the GPS signal is strictly line of sight.
Next Wave: Coupling Precise GPS Positioning to Vehicle Guidance and Control
Most initial uses for GPS were as replacement technology for applications that were already established. DGPS has rapidly improved accuracy, while improved receiver technology has expanded the set of measurable quantities. A single GPS receiver can now measure 13 dimensions of position (more properly states) for an airplane or other vehicle. This is summarized in Table 4. These 13 simultaneous GPS measurements of vehicle state represent opportunities for expanded use. This next wave will probably include closed-loop control of a wide variety of vehicles using the power of the 13 GPS dimensions. Accurate control using navigation satellites (ACUNS) is the name given by Parkinson to this set of applications. The value of this class of uses includes greater utility, safety, and productivity. This section will brie y explore these opportunities. B.
Vehicle Control Status
A number of examples of automatic control using GPS are currently being developed. Their status and prospects are summarized in the following.
Capability of a single GPS aviation receiver: the 13 dimensions
State Three dimensions of position Non-DGPS Local DGPS Local CDGPS Three dimensions of velocity Non-DGPS Local DGPS Local CDGPS Three dimensions of attitude Three dimensions of attitude rate Precise time
20 – 50 m 1– 5 m 5 – 10 cm
Nonprecision approach Precision approach Automatic landing
0.3 m/s 0.05 m/s 0.02 m/s 0.1 deg
Improved guidance Precision approach Automatic landing Improved and automatic guidance Improved and automatic guidance Time coordinated operations
0.5 deg/s <1¹s
Fig. 18 Fully autonomous ight of a model aircraft with GPS-based navigation and control, including takeoff and landing. 3.
Autonomous Farm Tractor
The operation of farm tractors is very manpower intensive. It is exacting yet very repetitive: some elds in California require over 2 h to make a single pass around them. A current program at Stanford is aimed at providing closed-loop guidance of such tractors, initially supervised by an onboard operator. It is hoped that full robotic operation will be also feasible with proper attention to safety and integrity. With CDGPS, potential positioning accuracies of a few centimeters should be suf cient for the most stringent farm tractor operations. The Stanford program has already demonstrated closed-loop control of large John Deere tractor (on a rough eld) to an accuracy of 3– 5 cm. C. Some Further Guidance Opportunities 1. Aircraft Guidance With or Without a Pilot in the Control Loop Fig. 17 Vertical GPS sensor error for 110 autolandings of a United Boeing 737. 1.
Autolanding of Aircraft
Enormous FAA attention has focused on this application since the announcement that the U.S. would no longer support the development of the microwave landing system. Major FAA-sponsored contractors are working to establish GPS-based landing feasibility and to select the best concept. In October of 1994, Stanford University teamed up with United Airlines, under FAA sponsorship, to demonstrate full category III autolanding with a commercial transport, a Boeing 737-300. The resulting vertical accuracy is shown in Fig. 17. The estimated 2 – 4 cm position accuracy was obtained using the variety of pseudolites called integrity beacons. The GPS system certainly was more accurate than the laser tracker, which calibrated the 110 resulting landings. Current analysis shows that the landing system based on integrity beacons is the only con guration that provides the integrity required of the FAA category III landing systems. Surprisingly, the issue for autolanding is not accuracy but is instead the high demand for integrity. As mentioned, there must be a negligible (<10¡9 ) probability of the system indicating a position outside the speci ed protection limit. A number of organizations are working on solutions. There is high con dence that this aspect of the next wave of GPS development will be successful. It should lead to a reliable autolanding system that will be available at hundreds of airports that do not yet have this capability. 2.
Autonomous Model Aircraft and Helicopters
Research has also explored GPS-based guidance for unmanned air vehicles. Two examples at Stanford have been a large model airplane and a model helicopter. Figure 18 shows a ight in which a model aircraft took off, ew a square pattern, and returned to a landing (within less than half a meter of the designated path) without any human intervention. The winds were large relative to aircraft velocity, yet the model plane held to its ight path within 0.5 m except in the turns.
a) Parallel runways. Key airports such as San Francisco cannot use both of their main runways for blind landings during periods of low visibility. The main reason is the navigation inaccuracy of the instrument landing system (ILS) during the approach at long distances (5 – 10 mile) from the air eld. DGPS accuracy does not signi cantly degrade with distance. In addition, GPS can support gradually curved approaches whose separation distances are large at the longer distances and only gradually converge down to touchdown, thereby minimizing risk and user hazard. Beam-riding systems such as ILS cannot do this. b) Collision avoidance. With the precise three-dimensional position and velocity that GPS provides to an airplane, a generalized broadcast of these quantities to other aircraft in the vicinity allows all users to calculate the probability of collision. If necessary, evasive maneuvers can be undertaken and closely watched using the same information. c) Autonomous cargo aircraft. Whereas the potential of larger unmanned aircraft may seem farfetched, there are economic incentives for use, particularly for cargo aircraft. One could foresee an evolution in which large aircraft y with only one pilot as an emergency backup operator for cargo or overnight mail. Of course, issues of safety, reliability, and integrity would have to be resolved. The military has irted with remotely piloted vehicles (RPVs), most recently in Bosnia. As our society increasingly expects minimal risk to humans, the military will be correspondingly motivated to avoid risks to human life through use of GPS-guided RPVs. An analogous civilian application is autonomous crop spraying. CDGPS not only has the accuracy for this mission but, when coupled to atmospheric data, the wind vector (and chemical dispersion patterns) can be calculated in real time. This would reduce the hazard of misapplication and would probably increase the ef ciency of dispersion. 2.
Automotive Guidance Opportunities
The intelligent transportation systems program is a major new Congressional effort that will have a substantial impact over the
next 20 years. One of its key technologies is mobile positioning, which has GPS as a key ingredient. Uses will span everything from vehicle surveillance to emergency noti cation. 3.
Equipment Guidance: Construction and Agriculture
The use of GPS for farm tractors has already been mentioned, but this can be expanded into the precise control of virtually all forms of heavy construction equipment using CDGPS. Examples include mining, road building, and pile driving. Some pioneers are now starting to experiment with these applications.
Navigation Satellites, an International Resource: Need for Cooperation
Like it or not, GPS is truly an international resource. The U.S. DOD views this with concern, since GPS navigation can potentially be used by an enemy. This legitimate fear has led to the continuation of S/A long after it has been shown to be ineffective against a user who employs the crudest form of DGPS. The major issue is this: GPS has shown a capability vastly superior to any competing technology, yet it will not be established as a standard until the international community is comfortable with the U.S. commitment to uninterrupted service and to some form of shared control. The U.S. will not be comfortable with shared control unless they have some means of ensuring that the system will not be used against them by an enemy and will not relinquish control without others sharing the economic burden. All non-U.S. countries do not feel so uncomfortable with the current situation, but enough members of ICAO are concerned (particularly in Europe) that GPS alone probably cannot yet be established as an international standard. The details are many, but the essence of the problem is the perception that they cannot depend on GPS. Lurking in the background is a world community of GPS and DGPS users that is expanding at 1 million per year and accelerating in rate of growth. Any major changes or additions to GPS that are not compatible with the pre-existing user equipment will cause howls of discontent. Any attempt to legislate or otherwise rule that their equipment is unusable will be met with great resistance. In essence, they have voted with their pocketbooks. B.
2) All countries that have contributed in proportion to their gross national product would be deemed authorized users and would be entitled to a proportional (to their contribution) vote in setting international standards for GNSS. 3) An international oversight board for GNSS would be set up with powers as determined by common agreement. Operation of individual elements would be delegated to contributing states but would be monitored to ensure that minimum standards of quality are met. GPS is truly an instrument to unify locations and peoples throughout the world. When global was selected as its rst name, this was the intent. Hopefully this uni cation will be fostered with an international spirit of trust and cooperation.
Path to a Solution
To date, this issue has mostly been considered by ICAO, although it is broader than aviation. Perhaps it is time for the U.S. to propose a solution that addresses these international concerns. Such an agreement would be the basis for a truly international GPS [or, as it is called, the Global Navigation Satellite System (GNSS)]. It might include the following elements. 1) Regions and individual countries would be encouraged to supplement GPS with their own (probably geosynchronous) satellites. Such satellites would broadcast, as a minimum, a GPS SPS signal using an on-board atomic clock. (The stable clock would ensure that the resulting ranging accuracy would be consistent with the existing GPS system; hence, it would not degrade accuracy.) They would also include DGPS corrections as part of an integrity message. Prototyping of this has already begun in the U.S., and Japan has proceeded with a prototype satellite system called MTSAT.
Acknowledgments In closing, I would like to acknowledge and thank the people who worked for and with me as part of the original JPO. They overcame adversity, ignorance, and uncertainty. They steadfastly adhered to the vision. The destiny of engineers and builders is to be rapidly forgotten by the public. I, for one, will not forget either their sacri ces or their achievements. The historical discussion draws heavily from the rst volume of Theory and Applications of GPS. It also makes use of “A History of Satellite Navigation,” for the non-GPS material. I thank my coauthors of that article for their contributions.
References 1 Parkinson,
B. W., “Introduction and Heritage of NAVSTAR, the Global Positioning System,” Global Positioning System: Theory and Applications, Vol. 1, Progress in Aeronautics and Astronautics, edited by B. Parkinson, J. Spilker, P. Axelrad, and P. Enge, AIAA, Washington, DC, 1996, Chap. 1. 2 Parkinson, B. W., Stansell, T., Beard, R., and Gromov, K., “A History of Satellite Navigation,” Navigation, Journal of the Institute of Navigation, Vol. 42, No. 1, 1995, pp. 109– 164. 3 Anon., “The TIMATION I Satellite,” Naval Research Lab., NRL Rept. 6781, Space Applications Branch, Washington, DC, Nov. 1968. 4 Anon., “Navigation Satellite Constellation Study, Final Report,” RCA Astro Electronics Div., RCA Rept. AED R-3632F, Princeton, NJ, Jan. 1971. 5 Anon., “Medium-Altitude Navigation Satellite System De nition Study,” RCA Astro Electronics Div., RCA Rept. AED R-3665-F, Princeton, NJ, April 1971. 6 McCaskill, T., and Buisson, J., “NTS-1 (TIMATION III) Quartz-andRubidium Oscillator Frequency Stability Results,” Naval Research Lab., NRL Rept. 7932, Washington, DC, Dec. 1975. 7 Parkinson, B. W., and Gilbert, S. W., “NAVSTAR: Global Positioning System—Ten Years Later,” Proceedings of the IEEE, Vol. 71, No. 10, 1983, pp. 1177 – 1186. 8 Parkinson, B. W., “The Global Positioning System,” Bulletin Geodesique, Paris, No. 53, 1979, pp. 89 – 108. 9 Bowen, R., Swanson, P. L., Winn, F. B., Rhodus, N. W., and Fees, W. A., “GPS Control System Accuracies,” Global Positioning System, Vol. III, Inst. of Navigation, Washington, DC, 1986, p. 250. 10 Sharrett, Wysocki, Freeland, Brown, and Netherland, “GPS Performance: An Initial Assessment,” Proceedings, ION GPS-91, Institute of Navigation, Washington, DC, 1991. 11 Parkinson, B. W., and Enge, P., “Differential GPS,” Global Positioning System: Theory and Applications, Vol. 2, Progress in Aeronautics and Astronautics, edited by B. Parkinson, J. Spilker, P. Axelrad, and P. Enge, AIAA, Washington, DC, 1996, Chap. 1.