USA   F-111 Aardvark


GPS Part V - The USAF EDGE High Gear Program

by Carlo Kopp

published in Australian Aviation, 1997

(c) 1996-2001 Carlo Kopp

Last year Technology Explained provided a comprehensive four part discussion of the new generation of GPS guided munitions, and their implications for air warfare in the next two decades. In this follow-up article, we will take a look at a very important technology demonstration program, which was sponsored by the USAF's JDAM program office.

The purpose of the EDGE (Exploitation of DGPS for Guidance Enhancement) High Gear program was to demonstrate the military potential of wide area differential GPS techniques for weapon guidance, by achieving accuracies better than 3 metres. The program was a stunning success, yielding what should be regarded as remarkable results.

The EDGE program demonstrated that appropriate use of DGPS techniques can provide military aircraft and munitions with sub-metre positioning accuracies in all three axes, over areas of continental sizes. How this was achieved will be the subject of this article.

The EDGE Program

The USAF EDGE Program resulted from a series of Concept Exploration Studies which were sponsored by the JDAM program office. The purpose of these studies was to determine alternatives for providing the JDAM with a 3 metre CEP under adverse weather conditions. The baseline JDAM CEP is 13 metres, which places the weapon into the category of "accurate" rather than "precision" munitions. While an "accurate" JDAM is clearly a weapon of tremendous utility, a "precision" JDAM would allow the weapon to wholly supplant the existing Paveway II/III with an all weather fully autonomous replacement. Existing expectations are that 80,000 JDAM kits will be built.

While the JDAM is expected to become the principal all weather "bread and butter" munition for US services, it is not expected to wholly replace all seeker equipped weapons. This is because seeker equipped weapons, using millimetric wave, optical and Synthetic Aperture Radar techniques are becoming more cost competitive, and can operate even in environments where a sophisticated GPS jamming threat or poor GPS reception exist. Moreover, autonomous seeker equipped weapons can achieve high accuracies often with limited support from a launch aircraft. We can therefore expect that the US munitions inventory early in the next century will comprise mainly JDAMs, supplemented by seeker equipped weapons, to provide the diversity to deal with a wide range of delivery platforms, conditions and jamming threat environments.

Four studies were contracted for, and these focussed primarily on precision seekers for the JDAM, requirements being that the techniques are cheap, autonomous, allow for retargeting in flight and are all weather capable.

One of these studies, conducted by SRI International of Menlo Park, California (formerly Stanford Research Institute), identified the potential of Wide Area DGPS (WADGPS) techniques to fulfill this requirement. The USAF subsequently contracted SRI to conduct a proof of concept experiment. This experiment involved the testing of DGPS over a long (ie 2000 NMI) baseline (ie Florida to California), and then led to the construction of a four station WADGPS network, termed the EDGE Reference Receiver Network (RRN). Following the testing of the network, a number of GBU-15 glidebombs were modified for DGPS/inertial guidance and successfully tested at Eglin AFB in Florida, the USAF's equivalent to our ARDU.


The best starting point for a discussion of the EDGE RRN are the limitations of the existing GPS scheme and commercial DGPS schemes. Readers unfamiliar with GPS are advised to review the GPS fundamentals covered in the 1996 series.

There are three basic sources of error when delivering any munition, these are the target location error (TLE), navigation errors and guidance errors. In a GPS based system, the navigation error is produced primarily by three mechanisms, which are uncompensated atmospheric transmission delays in the satellite signals, errors in the satellite's onboard atomic clock and orbital ephemeris data transmissions, and GPS receiver errors caused by noise and multipath. If we are using the civilian GPS SPS, then a further error is produced by satellite clock "dithering", which intentionally limits accuracy. These errors appear in the pseudo-range measurement to each of the satellites in view and carry through to the navigation coordinates produced.

The conventional commercial DGPS schemes in use provide a "band-aid" fix to compensate for dither and atmospheric delays, and satellite orbital and clock errors by measuring pseudo-ranges to satellites from a precisely surveyed location, and using these to calculate a correction which is broadcast to aircraft by a radio beacon. To defeat (Selective Availability) clock dithering, the updates must be as frequent as one per second, to preserve satellite visibility relationships between the ground station and GPS user, the coverage is typically limited to about 300 NMI. A number of commercial DGPS schemes exist which use dedicated radio datalinks, and one which piggybacks the DGPS signal on to commercial FM radio transmissions. The latter is accurate to 1 metre at 75 NMI.

Military DGPS schemes can be somewhat more robust, jam resistance is inherently better with P-code systems, dither is not an issue and the receivers can further compensate ionospheric delays by comparing the L1 and L2 GPS carrier signals. Tropospheric delays can be reduced by using error models embedded in the GPS receiver firmware. At a minimum a military DGPS scheme need only compensate for satellite clock offset error and orbital position drift. As a result, a military DGPS scheme can cover wide areas and use fairly sedate update rates as slow as 1 update per 30-45 minutes.

The optimal solution for military WADGPS is to add the corrections to the existing GPS navigation message broadcast, however due to limitations of the existing satellites and their supporting ground network this is not a practical short term proposition. The USAF's follow-on and separate WAGE (Wide Area GPS Enhancement) program has demonstrated the insertion of encrypted DGPS corrections into Page 4 of the GPS broadcast almanac message, and has been used for trials of the Block II CALCM and a modified AGM-130. It is intended that the WAGE system eventually transition to an operational system, as the existing GPS satellites are replaced.

In the near term, any operationally deployed military WADGPS schemes will have to employ radio datalinks for this purpose, as were used in the EDGE program. In the longer term, late model GPS IIR "replenishment" satellites, which employ satellite-to-satellite radio crosslinks, as well as a higher baseline accuracy of 6 metres rather than 20 metres, and a higher power output for weather penetration and jam resistance, would be used. These will have the capability to robustly support a fully embedded WADGPS scheme such as WAGE. The twenty one GPS IIR sats will be progressively deployed between 1996 and 2006.

The datalink scheme used in the EDGE program comprised the USN developed Improved Data Modem (IDM), fitted to the F-16D test aircraft and the central ground station. An encrypted 64 byte correction message included satellite IDs, pseudorange corrections for 12 satellites, standard deviations for corrections and the orbital parameters (specifically data used to select the exact ephemeris pages from the respective GPS almanacs) used in generating the correction. This message was broadcast from the ground station to the test aircraft, decrypted on receipt, and used to improve the accuracy of the aircraft's GPS aided inertial navigation system. Orbital parameters (specifically ephemeris parameters from applicable GPS almanacs) and corrected aircraft position were then downloaded via the Mil-Std-1553B bus to the test munitions.

The EDGE RRN evolved from the long baseline WADGPS experiment, and included further design enhancements by SRI to enhance its accuracy. The network employed four ground stations, each no less than 1000 NM from the intended test range at Eglin in Florida. The stations were placed at Kirtland AFB in New Mexico, Ellsworth AFB in South Dakota, Hanscom AFB in Massachusetts and Roosevelt Roads NS in Puerto-Rico, at precisely surveyed locations. Each ground station comprised no more than a high quality military 12-channel GPS receiver and choke ring antenna, designed for very low multipath reception, a desktop computer and a modem. Software running on the computer would gather GPS measurements, calculate errors for the site, and via a modem communicate these to a central site. A computer at the central site would then calculate the proper correction values to be broadcast via radio modem for aircraft operating in the test area.

While the hardware requirements for the EDGE RRN were clearly trivial, the SRI developed software which calculated the corrections was certainly not. A number of rather clever techniques were used, requiring no less than 40,000 lines of code, to minimise the resulting DGPS error.

The first technique used was to compensate for carrier phase slips, which occur when a receiver loses a carrier cycle. This was achieved by integrating carrier phase. Ionospheric delay was compensated by comparing L1 and L2 P-code measurements with the integrated carrier phase, in turn multipath and noise errors were compensated by carrier smoothing. Carrier smoothing involves the continuous integration of the carrier phase with previous measurements. Thermal drift in the GPS receivers was compensated by placing them in temperature controlled enclosures. Once these errors were compensated, tropospheric delays were measured by long baseline techniques. Tropospheric delays fall into two categories, a "dry" delay due to pathlength (slant range to satellite) and a variable "wet" delay, which is a function of humidity, temperature and cloud cover. To calibrate the tropospheric model, the ground stations were equipped to measure ambient temperature and pressure. The atmospheric tropospheric delay was calculated using differences in satellite elevation angles from the physically separated reference receivers to yield tropospheric pathlength values.

The final major error source to be compensated was the solid earth tide error. This error results from the earth bulging due to gravitational tidal forces, and can be as large as 30 cm in altitude twice daily, across a 2000 NM distance.

These corrections were combined using a weighted mathematical model which repeated the calculation until an optimal set of correction values was produced for the region of coverage. The correction values were then merged to produce a single set of numbers for transmission to an aircraft, optimised for the lowest possible error at the centre of the theatre of operations, in this instance Eglin.

The result of these corrections was a position error which during the EDGE trials varied between 5 cm and 1.57 m, with an RMS value of 40 cm (15.7 in). On average, the position error was under 18 inches, in a network with reference stations of the order of 2000 NM apart, with WADGPS updates produced every 6 seconds and each deemed valid for 30-45 minutes. It is worth noting that accuracy in WADGPS schemes improves with geographical coverage, as more widely spaced reference stations can keep satellites in view longer and therefore determine their orbits more accurately. A continental network would do better than the existing EDGE, and a global network even better.

Clearly such results were outstanding, but to convince the sceptics there is no substitute for footage of bombs punching through targets. The next phase of the EDGE program therefore concentrated on demonstrating the utility of WADGPS for munition guidance.

The EDGE GBU-15 Munition

The baseline GBU-15, used by the USAF and RAAF, is a glidebomb equipped with a TV or thermal imaging seeker and two-way radio datalink. It was well suited for such a demonstration because it has both the volume to accommodate a GPS guidance package, once the existing seeker was removed, a highly reliable flight control section which simplified integration, and sufficient standoff glide range to guarantee a zero probability of hit should the DGPS system not perform and guidance default to inertial alone. Inertial errors increase with flight time, but GPS/DGPS errors do not. Six rounds were custom modified for the EDGE trials.

To fit the bombs on the diminutive F-16D fighter, the older long chord wing assembly was used in preference to the newer short chord wings, although the short chord control surfaces were used to provide safe clearance with the trailing edge flaps of the F-16. The standard GBU-15 optical seeker, mounted in the nosecone, was removed and replaced with a GPS/INS package. The existing analogue autopilot, gas bottle reservoir for control power, and control actuators were retained. The GPS/INS package was interfaced to the autopilot electrically, producing suitable steering commands.

The GPS/INS package comprised an Integrated Flight Management Unit (IFMU) and a GPS receiver. The Honeywell IFMU was based upon the off-the-shelf HG1700 (GG1308) Ring Laser Gyro IMU package, and provided the 1553B interface to the launch aircraft, a telemetry interface, the autopilot D/A interface and the interface to the GPS receiver. Software running on the IFMU executed navigation, pseudorange differential GPS corrections, weapon status and health monitoring and event sequencing. In effect the brains of the bomb, the IFMU weighed all up under 8 kg.

The Interstate Electronics Corporation (IEC) SEM-E GPS receiver was a five channel P/Y code capable military GPS receiver, designed for fast satellite acquisition. Built as a set of four SEM-E format circuit boards, the receiver was small enough to fit inside the IFMU cage. Two antennas were fitted, one on the top of the nose section and one on the tail. The receiver could select either antenna to get the best satellite visibility for any given geometry.

Before weapon release the Kalman filtering software running on the bomb IFMU was fed with position and velocity data from the launch aircraft via the 1553B bus, in effect slaving the bomb to the position of the aircraft, with an allowance for the moment arm between the aircraft INS and weapon IFMU. Once the bomb was released, the GPS receiver would acquire five satellites within 10 seconds and the Kalman filter mode adjusted to support no less than 17 states. The filter was designed to progressively blend in GPS receiver measurements with increasing weight, after release (technical readers will note that the channel noise or error was initially assumed high, and then progressively reduced to match the expected error of the differentially corrected solution). This was to ensure that the data provided by the receiver was stable and "trustworthy", as receivers often take several seconds to settle in once activated. Differential corrections downloaded before launch were then fed into the Kalman filter. The software was implemented in DoD ADA high level language.

Conventional proportional navigation was not employed in the EDGE scheme. A new guidance law was used which allowed the weapon to impact the target at any desired vertical angle and heading angle, to maximise lethality and flexibility. For fixed targets, proportional navigation essentially aligns the weapon velocity vector and the target line-of-sight vector (in the simplest of terms, the weapon just flies from the launch point straight to the target). The EDGE guidance law aligns the weapon velocity vector, target line of sight vector, and a target impact vector, in a manner devised to match the kinematic capability of the weapon (as is done by JSOW). This allows the targeteer/bombardier/pilot to specify the target surface to be hit with an optimum angle for penetration.

In the simplest of terms, the EDGE navigation scheme could be described as similar to that used in the F/RF-111C AUP nav-attack system, with the addition of a more complex Kalman filter which applied the differential corrections sent to the launch aircraft from the ground station, and with a sophisticated flightpath control algorithm designed to maximise lethality.

The Flight Tests

While the objective of the flight test program was to put bombs into test range targets, a series of tests had to be conducted before this could occur. These involved static ground testing and captive flight tests.

The static ground tests involved parking the Block 50 F-16D, carrying two bombs, on to a precisely surveyed point. One of the bombs was fed differential GPS corrections, the other used standard GPS. Twenty simulated launches were "flown", and the position measurements from both bombs recorded and compared. After 100 seconds of "flight", the nominal time from release to impact, the "differential" bomb produced a mean position error of 6.3 feet with a standard deviation of 3.6 feet, compared to 12.8/8.5 ft for the "standard" bomb. The figures were even better for the 3 dimensional error, with 8.8/5.5 ft vs 20.3/12.3 ft, respectively.

A similar series of tests were then conducted using a "differential" and "standard" bomb captive carried by the F-16 flying through an instrumented corridor over the test range, the aircraft flying twenty five passes to gather test data (who ever said a that a test pilot's life had to be exciting all the time !). These captive flight tests were complicated by the F-16 wing, body and tail blocking the line of sight to satellites on a number of runs. Antenna wing shadowing did not cause problems during the live drops, as the fast acquisition GPS receiver could acquire and lock up satellites very quickly, once the bomb was clear of the aircraft.

The best miss distance achieved by the "differential" bomb during the captive tests was about 3.5 feet, with an average of 12.03 ft. Under the same conditions, the "standard" bomb achieved about 20 ft in most tests.

The captive tests demonstrated that an error between one half and one third of that produced by standard GPS could be achieved. This was subsequently confirmed by the live flight tests.

With only six test rounds available for use, the USAF had to be very cautious in how they used their test articles, to achieve best effect. The drop flight tests were split into three categories. The first two flights involved an attack on a horizontal target, the second two a vertical billboard target, and the final two emulated an operational scenario. The tests were conducted during May and June, 1995.

The profile for the first two tests saw the bombs released at 30,000 ft from a distance of 12 NMI, the weapons flying a shallow 18 degree dive until close to the target, where they nosed over and dived at 84 degrees, impacting at about 300 m/s velocity. In both instances the bombs hit within a 5 metre distance from the programmed aimpoint. This consistent error was attributed to antenna multipath effects during the last 20 seconds of flight, a result of the satellite signal interacting with (ie reflecting off) the GBU-15's large tail surfaces.

The profile for the second pair of tests saw the bombs released at 26,000 ft from a distance of 14 NM, the weapons flying a shallow 20 degree dive to impact on the vertical target, with a velocity of about 290 m/s. In the third test the bomb hit within 1.9 metres (6.2 ft) of the aimpoint, which was indeed the highlight of the series. A useful comparison here is that the GBU-15 airframe is 12 ft (3.7 m) long, therefore the error was about one half the length of the bomb ! The fourth test impacted short, nine metres from the aimpoint. Published analyses of test results suggests that higher than expected humidity may have impaired the accuracy of the tropospheric model used.

The third flight test simulated an operational sortie. Unlike earlier tests, where target coordinates were produced by Defence Mapping Agency (DMA) survey, target coordinates were produced by the USAF Space Warfare Centre at Falcon AFB, in Colorado. The DGPS corrections were sent to the aircraft from an IDM ground station deployed to Tyndall AFB in Florida. Two 46th TW aircraft were flown in the test, each carrying one bomb, in radio silence, in a tactical formation. The weather was overcast with a base at 12 kft and thunderstorm activity. The easier horizontal target attack geometry was chosen for this test, release altitudes and ranges were similar to the previous tests.

The results were consistent with earlier testing, with one round hitting 3.9 metres from the aimpoint, the other failing to acquire a full set of satellites and impacting 11.4 metres from the target. During this final testing, the accuracy of the GPS corrections produced by the EDGE network was less than one-half metres in the horizontal and less than one metre in the vertical.

It must be noted that horizontal target attacks are easier than vertical targets. This is because attacking a horizontal target means that the bomb dives down in a near vertical trajectory, therefore in effect making vertical position errors irrelevant. As the vertical error in GPS is inherently greater than the horizontal errors, hitting a vertical target with a shallow dive trajectory is much more difficult. This makes the successful results of the EDGE trials all the more important.


The EDGE program demonstrated some very significant points. The first is that sub-metre positioning accuracies can be achieved using WADGPS schemes which exploit the full capabilities of military GPS receivers. The second was that substantial accuracy improvements can be achieved by using WADGPS schemes to augment the navigation solutions produced in GPS guided weapons. Because such WADGPS schemes allow for widely spaced ground stations, they are a viable proposition for operational deployment in any theatre where friendly territory can be accessed within 1000 NM of the intended area for weapon delivery.

Analysis of test results and telemetry from the EDGE tests suggests that the principal source of error were GPS receiver multipath effects, and limitations in the update rate of the Kalman filters used. Experience with the ground stations initially was that multipath corruption was a serious source of error in the navigational solutions produced . As funding for the EDGE project terminated after the final drop, the USAF has yet to perform a more comprehensive analysis on the gathered test data and validate the conclusions of the tests. Given that Paveway II class accuracy was achieved during the tests, the USAF was not under any great pressure to do so.

SRI and ASEI did carry out further company funded analyses of the drop results and applied the findings to further improve the accuracy of the EDGE RRN and to improve the performance of subsequent bomb tests. Most recent tests indicate that the EDGE RRN is achieving a 25 cm (9.8 in) horizontal position accuracy ! Much of what was learned during EDGE was also merged into the USAF WAGE program, resulting in a modified WAGE DGPS guided AGM-86C CALCM recently achieving a 3 metre miss distance in a vertical dive attack trial.

The project which has benefitted the most from the EDGE program is the USAF's Miniature Munition Technology Demonstration (MMTD - "Small Bomb") program, which has used the established EDGE network for trial drops. In all of five recent drop tests, miss distances below 1.5 m (4.9 ft) were consistently achieved. The MMTD tests have used Kalman filters with higher measurement rates than EDGE, and have placed a GPS antenna on the tail of the test aircraft, which suggests that two problems identified in EDGE have been since solved.

In the Australian context, the adoption of a dual mode PPS/SPS WADGPS network for regional use could offer significant dividends in improved navigational and weapon delivery accuracy for all ADF platforms, and should the JP 129 and SOI surveillance and recce projects proceed, significantly improved calibration accuracy of SAR generated radar images. Because update messages are both compact and can be infrequent for such networks, they would not strain the existing and limited HF and satellite resources. As the ground stations involve installations of relatively trivial cost and complexity, it would be feasible to place redundant ground stations at continental and remote Australian sites (eg Cocos Island, Christmas Island, Norfolk Island) as well as at bases in friendly regional countries. An existing telephone channel and encrypted modem would be adequate to carry the required ground station traffic. Such a project is easily with the capabilities of our DSTO researchers.

RAAF aircraft such as the F/A-18 and F/RF-111C/G if equipped comprehensively with suitable GPS and datalink receivers could then navigate and deliver weapons with positional accuracies of the order of a metre. In practical terms, this means that the delivery error for a dumb bomb becomes primarily the systematic delivery error for the weapon system, and the target location error. This could be as low as 30-50 ft, subject to delivery profile.

To fully exploit the capabilities of such a network, the RAAF would require suitable GPS guided weapons such as the GBU-31/32 JDAM or the BAeA AGW/Kerkanya, in variants which are equipped to handle WADGPS correction updates from the launch aircraft. In the instance of the JDAM, we will have to wait for the USAF to eventually introduce such a capability in a PIP upgrade, possibly using the WAGE scheme, or directly fund the required modifications to the existing JDAM hardware and software. When/if this takes place remains to be seen, as the current JDAM does not have such a requirement. In the instance of the BAeA AGW, it would be up to the designers to accommodate this capability in the current development design.

The EDGE program was the forerunner of bigger and better things to come. The combination of WADGPS techniques and the existing generation of GPS guided weapons promises autonomous, all weather precision bombs with accuracy equal or better to that of existing laser guided weapons. The force multiplication effects resulting from this will further elevate the primacy of modern air power as a power projection tool.


Special thanks to Earl G. Blackwell, SRI's Program Director for EDGE, Dr Don Kelly, and David Gaskill of ASEI, both formerly of the EDGE project team, and Lt.Col. Greg Teman, USAF, and the USAF JDAM program office for their assistance with the preparation of this article.

Author's Note (June, 2001):

The technology demonstrated in the EDGE trials has since migrated into a number of in service and new weapons. The Enhanced GBU-15 (EGBU-15) exploits the GPS aided navigation techniques, and the MMTD/SSB (Small Smart Bomb) relies extensively on EDGE experience. The RAAF's F-111G is the trial platform for supersonic drops of the SSB weapon. We can expect this technology to progressively migrate into all JDAM family weapons, including the Australian Boeing/HdH JDAM-ER glidebomb, based on the DSTO Kerkanya wing kit.

Pic.1 (F-16D + 2 bombs)

A Block 50 F-16D of the Eglin based 46th Test Wing carrying a pair of EDGE test weapons during captive carry trials. One bomb is measuring its position using standard GPS, the other is using differential GPS updates datalinked to the F-16 from a ground station, using an encrypted channel. The bomb using DGPS achieved accuracies as high as 3.5 ft in horizontal position (USAF).

Pic.2 (F-16D releasing bomb)

Bomb Away ! One of the six EDGE test weapons is released for a live test. In all six tests, both horizontal and vertical targets were attacked, in classical test range environments as well as a simulated two aircraft operational sortie. Typical bomb accuracy was 4 metres against a horizontal target, the DGPS solution being degraded by multipath effects resulting from antenna and bomb wing interaction (USAF).

Pic.3 (Billboard target)

History is made on the third flight test of the EDGE weapon, when the test round impacted within 1.9 metres from the intended aimpoint on the vertical billboard target. This test proved the potential of differential GPS techniques to replace conventional laser guided bombing technology (USAF). LOGO


bulletin board
| images | sounds | movies | other links | published articles | history | troops | comics | models
| external differences | patches and badges | books | AGM-142 Popeye | museums | memorabilia
Tail Numbers:
| EF / F-111A | F-111B | RF / F-111C | F-111D | F-111E | F-111F | FB-111A, F-111G | F-111K
F-111 Down-Under | airshows | RAAF Tail Numbers | ejection | artwork | feedback | Notice Board | MEMORIAL

Also visit' s companion site,   FB-111A Switchblade

Copyright '   F-111 Aardvark'.

Links to, and reviews of this site are welcome.

Disclaimer: This F-111 Aardvark Internet site does not represent the views of General Dynamics, Lockheed Martin, the United States Air Force, the Royal Australian Air Force or any other company or organisation which may be named herein.  Should any company, organisation or individual feel grieved that I am using their logo or product without permission, please contact me at the email above.

Flag Gifs from