Aerospace demonstrated potential GPS accuracy enhancement techniques for the Joint Direct Attack Munition (JDAM), widely used in recent military engagements. JDAM uses GPS combined with an inertial system for navigation. Once released, the bomb's INS/GPS will take over and guide the bomb to its target regardless of weather. (The Boeing Company)

GPS/Inertial Navigation for Precise Weapon Delivery

Anthony Abbott

For centuries, military planners have sought to place a weapon exactly on an intended target. Such accuracy not only helps ensure destruction of the target, it helps prevent collateral damage. While systems have improved throughout the years, the advent of the Global Positioning System has brought a major advancement in precision weapon delivery.

In trying to hit a target with a weapon, the most basic approach is to launch it with the correct initial trajectory and let physics do the rest. This approach, however, is fraught with errors. Even with modern technology, the ability to direct such a weapon is limited by the accuracy of the target coordinates, uncertainties concerning aerodynamics and mass, and many other factors. The process is difficult enough for a stationary launch platform and even more difficult for a moving launch platform, such as a fighter or bomber aircraft. The problem gets even worse if the weapon must be powered for a portion of its flight, as is the case with ballistic or cruise missiles.

Navigation in Weapon Delivery

One of the early devices used to enhance missile accuracy is the Inertial Navigation System or INS. An INS calculates a vehicle's current position, velocity, and attitude by integrating the measurements from the system's inertial measurement unit—essentially a set of accelerometers and gyros. This information can then be used to steer the weapon toward the target. Although an INS improves weapon accuracy, the technique still leaves a significant margin of error, dominated by targeting errors and the buildup of instrument errors over time. Although many clever techniques have been invented to reduce these errors (including many developed at The Aerospace Corporation), the accuracy of weapon systems that rely solely on INS will always be limited.

During the 1960s and 1970s, Aerospace helped implement a new concept for navigation—the Global Positioning System (GPS). The system offers the user remarkable navigation accuracy simply through passive reception of satellite signals. The technology offers several advantages over INS in certain scenarios. For example, INS navigation errors tend to be cumulative, building up over time; but GPS errors tend to be bounded because the error sources (satellite position, velocity and signal propagation errors) are more easily modeled and mitigated. On the other hand, INS has good error performance in the short term, especially under high dynamics; GPS performs best with longer flight times and is less suitable for conditions of high dynamics. Not surprisingly, the two techniques are frequently combined to obtain robust, accurate navigation for demanding military applications (see sidebar, Absolute Coordinates).

Dumb Bombs

Efforts to calculate the correct release condition for an unguided or "dumb" weapon began in the late 1960s and early 1970s. Again, the basic idea was to launch the missile with the correct trajectory from a moving platform and let physics do the rest. Although the mathematical equations were readily available, the algorithms for the continuously computed release point and continuously computed impact point only became feasible with the advent of microprocessors that were powerful enough to perform such calculations using data from the launch vehicle's navigation system. The first such system used inertial navigation outputs to compute the impact point and provided steering and release cues to the pilot.

GPS offered a better approach. One of the main objectives of the GPS Phase I program was the precise delivery of dumb bombs. In early tests, the continuously computed release and impact point algorithms were implemented using a four-channel GPS receiver that was integrated with the launch vehicle's inertial measurement unit. By integrating the navigation and weapon delivery functions in the same computer, the two processes could be synchronized, with the weapon delivery function using the current best estimate of navigation parameters (position, velocity, and attitude as well as wind speed and direction) for accurate impact point prediction.

This chart shows the degree to which impact error is sensitive to errors in launch-vehicle velocity for dumb bombs without active guidance.

The weapon delivery software was so sophisticated that it actually calculated the impact and release points based on the extrapolated position and velocity. The software would predict the impact point of the weapon if it were dropped at any given moment and compare it to the desired impact point. The impact error would be projected into along-track and cross-track components. The cross-track error drove a deviation display in the cockpit to help the pilot adjust the ground-track angle. The pilot's job was to steer the aircraft to drive the cross-track deviation to zero. At the same time, the along-track error was displayed. This allowed the pilot to judge how close the plane was to the correct release point. As the plane approached the release point, the pilot would arm the automatic release mechanism. The computer issued a release command when the along-track impact error reached zero.

The Phase I weapon delivery system was remarkably successful. The equations of motion for the bombs were fairly complete, and most error sources were either modeled or mitigated. The primary accuracy limitation was the wind: Although wind speed and direction could be determined at the release point, the wind could change as the weapon fell. Another error source that was difficult to mitigate was the variability in the aerodynamics and mass properties of each bomb. The predicted impact point algorithm had to use average values because it would be impractical to enter the values of each bomb into the operational software. Nonetheless, the program was considered a great success.

Smart Weapons

As impressive as the GPS Phase I weapon delivery test results were, they made clear that accuracy would remain inherently limited unless some intelligence were placed in the bomb itself. The first attempt to make dumb bombs into "smart" guided weapons used an INS and associated fin-actuation system in a tail kit that replaced the normal tail section of the bomb. At the time, GPS receivers were too big to fit in a tail kit. It was thought that with proper initialization from the host vehicle, the bomb's INS could sustain navigation accuracy from the release time to impact. This strategy worked well as long as the launch vehicle had a GPS receiver to initialize the bomb's INS prior to release. Without a GPS receiver on the launch vehicle, the handoff errors were too great for precision bombing, simply because the host vehicle's INS would accrue errors that would be handed off to the weapon during ingress.

The conventional air-launched cruise missile is guided by GPS. GPS accuracy can be improved using a variety of augmentation systems, such as differential GPS and wide-area differential enhancement. (U.S. Air Force)

The addition of GPS to the launch vehicle allowed it to initialize the bomb's INS with great accuracy. The navigation error buildup during the relatively short descent of the weapon was reasonably good as long as it was initialized properly. This approach, however, would not work well for a standoff weapon because the integrated instrument errors would grow to unacceptable levels during the longer weapon flight time.

As GPS receivers became smaller, the prospect of placing one in the same tail kit with an inertial measurement unit became feasible. With this concept, as long as the target coordinates could be determined using GPS, the impact error could be driven quite low. The impact error would be dominated by the targeting error and the GPS error during the weapon's flight time. The error buildup of the INS would be essentially removed—which is particularly attractive for standoff weapons with long flight times.

Of course, if the weapon uses a GPS receiver, the launch vehicle must provide the necessary GPS handoff information (in addition to the INS transfer alignment information). This handoff information usually consists of initial position, time, and velocity, as well as the GPS satellite orbital data. This information is especially critical for a weapon with a short flight time because it must obtain a GPS fix well in advance of ground impact to steer out the residual INS error buildup. Typically, the weapon requires enough information from the launch vehicle to acquire and track the GPS satellites within several seconds after release.

For older weapons, the GPS receiver uses the coarse acquisition (C/A) code to acquire the signal and then switches to the military P(Y) code. Although the C/A code provides the easiest way to acquire the GPS signals, it is also more vulnerable to jamming. Most modern weapons use (or will use) direct P(Y) code acquisition for better antijam protection. Direct P(Y) acquisition requires more careful host-vehicle integration because the weapon must have knowledge of time to within several milliseconds in order to search the range and range-rate uncertainties through its direct-acquisition application-specific integrated circuit chip. A time transfer from the host vehicle's GPS receiver to the weapon's GPS receiver via the flight management and the weapons management subsystems usually accommodates this time initialization.

These bomb-damage assessment photos show a target in Afghanistan before (left) and after (right) a strike by a B-2 bomber using GATS/JDAM. GATS uses a synthetic aperture radar to determine relative target coordinates and downloads them to the JDAM prior to release. (View larger image.)

The difficulty in accurate time transfer to the weapon is more a matter of economics than technology. Without an appropriate direct-acquisition chip in the GPS receiver, direct P(Y) acquisition requires initial knowledge of time to an accuracy on the order of tens of microseconds. To send a time pulse with this accuracy requires a high-bandwidth line to each weapon. This can be expensive to install. On the other hand, if the weapon has a direct-acquisition chip, the receiver can tolerate a time uncertainty on the order of a few milliseconds for direct acquisition of the P(Y) code. Accuracy on this level can be accomplished with a standard serial interface, if care is taken during the design of the software message protocol. Hence, many contractors have chosen not to install the high-bandwidth line in favor of using a weapon with a direct acquisition capability. As integrated circuits improve, the time accuracy requirement should diminish further without sacrificing antijam performance.

Integrated Systems

As long as the targeting function is performed independently of the strike function, the associated error contributions from each of these functions will be additive. Moreover, the GPS contribution to the targeting error could be different than the GPS contribution to the strike system error if the time delay between targeting and strike is more than 10 or 20 minutes. With proper systems engineering, however, such concerns can be minimized.

For example, a complete system can be designed more optimally than a series of optimally designed individual subsystems all functioning together. Weapon delivery is no exception. In the case of GPS/INS weapons, integration of the targeting and strike functions at the system level enables certain design choices that are not possible when each subsystem is independently conceived.

Designers have known for years that errors in the GPS measurements are temporally and spatially correlated—partly due to the design of GPS, and partly due to errors in signal propagation through the ionosphere and troposphere. This error correlation could be exploited through a system-level design that uses the same satellites in the targeting and strike functions.

Thanks to the very long distance from the navigator to each of the GPS satellites, the relative geometry of all GPS navigators in the same vicinity (such as a vehicle on the ground and an airplane flying above it) is similar. Hence, they will all experience essentially the same range measurement errors. The correlation is especially high if all navigators use the same four satellites. In fact, even some of the atmospheric errors, such as ionospheric propagation delay, are spatially correlated over significant distances. Hence, using the same satellites in the targeting function and the strike function offers error correlations that can be exploited to improve accuracy.

Bomb-damage assessment photos showing a target in Afghanistan before (left) and after (right) a strike by a B-2 bomber using GATS/JDAM (GPS-Aided Target System/Joint Direct Attack Munition). Today, thanks to GPS, multiple targets can be destroyed in one pass. (View larger images.)

The temporal correlation of errors becomes important when the targeting and strike phases must remain separate. Again, the property of the GPS ephemeris errors is such that, except for ephemeris updates, the range error from an imperfect ephemeris is highly time correlated. This temporal correlation can be exploited to minimize errors if the time from the targeting phase to the strike phase is on the order of several minutes. The temporal correlation of atmospheric propagation errors also works in the favor of accuracy if their correlation properties are exploited.

Without a systems approach to the entire problem, targeting errors would be a major contributor to the weapon's overall impact error budget. Proper integration of the targeting function and the weapon navigation function is ultimately responsible for complete system accuracy. The key to achieving very small weapon impact error is to force the weapon's navigation system to incur nearly the same errors as the targeting vehicle. This is one of the few instances in life where two wrongs make a right.

Bomb-damage assessment photos. The left photo shows Krivovo support base in Serbia. The strike was performed by a single B-2 at night in complete cloud cover after flying from Whiteman Air Force Base (midway between St. Louis and Kansas City) to Kosovo nonstop. Eight weapons, two per building, were deployed, with offsets in targeted points on each building to spread the damage. Synthetic aperture radar targeting was used just before launching the weapons. The photo to the right shows Shindand airfield in Afghanistan. The strike was carried out by a single B-2 at night after flying from Whiteman Air Force Base to the region nonstop. (View larger image.)

The techniques used to ensure that the weapon makes the same errors are straightforward, but many practical design choices must be made to ensure this behavior under all circumstances. The first step is to force the weapon to use the same four satellites that the targeting vehicle used. The weapon should also use the same ephemerides and ionospheric compensation calculations as the targeting vehicle. Given these design choices, the weapon will achieve the same position biases as the targeting vehicle, and the impact error will be dictated by other error sources that are more random in character (see sidebar, Imaging Bias and Relative Targeting Errors).

Future Trends

Although GPS-based relative navigation systems are capable of impressive accuracy even with substantial GPS position biases, there's still room for improvement. In the future, absolute navigation using GPS will probably be so accurate that relative navigation will no longer be required. Future conflicts will probably rely on smaller munitions (250 to 500 pounds) to minimize collateral damage. As weapons are reduced in lethality, the accuracy of the impact point must become even greater to ensure target kill (see sidebar, GPS for Stealth Bombing).

GPS/INS-guided weapons are very effective against stationary targets, but many adversaries have adopted defensive strategies that involve constant movement. This challenge is being addressed by numerous studies, which have shown that GPS/INS delivery techniques can still work if some adjustments are made—specifically in terms of calculating the revised target coordinates and transmitting them to the weapon in flight. Several methods could be used to send updated coordinates to the weapon—for example, updated target coordinates could be sent to the weapon with a new signal. This approach would allow the weapon to receive and decode the updated target coordinates and send them to the guidance function within the weapon's computer. Aerospace is studying possibilities such as this.

Aerospace is also investigating a number of methods for detecting moving targets and estimating their coordinates. Some types of synthetic-aperture radar are already capable of indicating moving targets on the ground. Current methods cannot yet establish an unambiguous target track with high confidence and accuracy, but considerable research is under way to perfect this capability.

One of the most stringent requirements associated with moving targets is the latency of the updated targeting data. If the target is traveling in a predictable path—along a straight line or a digitally mapped road, for example—this problem can be resolved with a more relaxed latency requirement by using track filtering and prediction algorithms. If the target is in an open area and is capable of "jinking" maneuvers, the latency requirement becomes far more demanding. Hence, the need for frequent updates in the targeting data may drive the system architecture. For example, the same vehicle that launched the weapon might have to perform the targeting update function and send the data directly to the missile over a radio link.

Summary

GPS and INS technology work extremely well together to provide the high accuracy and robustness needed for modern weapon delivery systems. As targeting technology improves, the integration of the targeting function into the weapon delivery system should result in spectacular accuracy not only for stationary targets but for moving targets as well. By properly integrating timely, accurate targeting information with the postlaunch guidance and navigation functions, a major advance in future weapon delivery capabilities will be possible.

To Summer 2002 Table of Contents