ACA’s journal, Arms Control Today, remains the best in the market. Well focused. Solidly researched. Prudent.

– Hans Blix,
former IAEA Director-General

Is Missile Defense on Target?
Share this

Philip Coyle

The clock is ticking. Last December, President George W. Bush announced plans to begin deployment of a strategic nationwide missile defense system at Fort Greely, Alaska, by September 30, 2004. With less than a year left before that deadline, it is clear that the president’s decision has drastically changed the priorities in the missile defense program and lowered the bar on the acceptable standards for an effective military system.

If the Bush administration’s now anemic testing schedule continues on track, the United States is set to deploy a missile defense system that is simply not up to the job. The ground-based midcourse defense (GMD) system, as it is now called, has not shown that it can hit anything other than missiles whose trajectory and targets have been preprogrammed by missile defense contractors to eliminate the surprise or uncertainty of battle. Nor has it proven that it can hit a tumbling target, perform at night, or find ways to counter the decoys and countermeasures that a real enemy would use to throw a defense off track. Tests so far have all been conducted at unrealistically low speeds and altitudes, and it is not clear that the system will be able to track and identify the warhead it is supposed to destroy.

Such criticism is not partisan in nature. Bush’s new testing schedule lags not only the comprehensive tests planned by the Clinton administration, but even the testing objectives of Bush’s first two years. Indeed, the Pentagon’s current missile defense plan marks a radical shift from a half-century of military testing carried out under Republican and Democratic administrations alike.

After Bush’s announcement, the missile defense program’s priorities immediately switched from challenging and necessary testing to building facilities at Fort Greely and hauling hardware and equipment to Alaska. Since construction began on June 15, 2002, 550 acres have been cleared, at least 620,000 cubic yards of dirt have been removed, 11 buildings have been built, and 25 others refurbished. Six missile silos are to be completed by next February, 10 more by the end of 2005, and as many as 40 in the years to come. Yet, the ability of the missile defense system to carry out its required tasks has barely inched forward.

Before the Deployment Decision

As envisioned, the GMD system is meant to consist of a set of silo-based interceptors, beginning with six at Fort Greely and four at Vandenberg Air Force Base in California. These interceptors are to carry infrared detectors capable of discriminating enemy warheads from decoys. The system is slated to include a mobile, sea-based X- band radar as well as fixed early warning radars at Shemya, at the end of the Aleutian chain, and at Beale Air Force Base near Sacramento, as well as early warning radars in England and Greenland. It also is to use satellites with infrared detectors capable of distinguishing between launches of peaceful rockets and ICBMs and discriminating enemy warheads from decoys. Finally, the GMD system is supposed to have a complex battle management command and control system that includes a network of satellites and ground elements extending from Washington, D.C., to Alaska, including Cheyenne Mountain in Colorado and sites in California.

In developing a schedule to develop and test the components needed for this system, the president began with a system inherited from his predecessor. The GMD system has more than a passing resemblance to the National Missile Defense (NMD) system planned by President Bill Clinton to protect the United States from attack by long-range ballistic missiles. However, the GMD system is now only the centerpiece of the larger Bush Ballistic Missile Defense System (BMDS), a “layered” system intended to be capable of shooting down missiles in all phases of their flight—boost, midcourse, and terminal—and from platforms based on land, at sea, in aircraft, and in space.

During the first two years of the Bush administration, the Pentagon carried out a testing program that did not depart radically from its predecessor. To be sure, there were some changes. The Bush administration has conducted five flight intercept tests of the GMD system as opposed to three flight intercept tests of the NMD system in the final two years of the Clinton administration. On the other hand, all of the flight intercept tests attempted in the first two years of the Bush administration were quite similar to tests during the Clinton years and did not push the state of the art as strongly as tests either planned or accomplished during the Clinton administration.

The Bush administration has shown some political wisdom in following a cautious script. Year after year, delays in the development program had stretched out the planned milestones, and mounting technical difficulties had shown that this program was no different than any other high-technology military development. It would not be surprising if it took a decade or more to develop an effective military capability. Only six days before the president’s deployment decision, the program had experienced yet another dramatic failure when an interceptor “kill vehicle” failed to separate from its rocket booster. To the people doing the actual work, the last thing they expected was an order from the president to move the schedule for deployment to the left.

Testing Not Accelerated

Bush administration officials such as Lt. Gen. Ronald Kadish, the head of the Missile Defense Agency (MDA), have sought to calm concerns expressed by Congress and the press by saying that the Pentagon would rev up the pace of testing to meet the president’s goals. Yet, overall the pace of flight intercept tests and, most importantly the rate of successful flight intercept tests, has stayed about the same. Since the inception of flight intercept tests in October 1999, five successful intercepts have been carried out in eight attempts. That is a rate of about one success every 10 months. At that pace, it could take 10 or 15 years before the GMD system could pass the 20 or 30 developmental tests required before realistic operational testing could be conducted. Developmental tests, especially in the early years of a program, may be heavily scripted with unrealistic or artificial limitations. Operational testing, on the other hand, must be realistic with the systems operated by real soldiers, sailors, airmen, or Marines, as they would be in battle.

Yet, intent on deploying the system in time for the 2004 presidential elections, the Bush administration has sought to act as if the necessary milestones were unnecessary obstacles. Just look at how the Pentagon dealt with problems caused by the unreliable surrogate booster rocket used in the first eight flight intercept tests, as well as delays in the operational, production version needed to launch the “kill vehicle” to collide with incoming missiles in space: it simply cancelled nearly half of the intercept flight tests it had initially outlined. Unable to make the system square with the usual Pentagon definitions of military capable programs, the Department of Defense has dumbed down the requirements for a militarily effective program. Incapable of having key components such as an eagle-eyed X-band radar and flight sensors in place for the “deployment date,” the Pentagon is ready to place the system on operational status even without the parts needed for it to be effective.

The problems began with the booster rockets. Booster development and testing alone has taken about three more years than planned. At one point in the schedule, booster development and testing were to have been completed in 2000, but that slipped to 2001 and now 2003. In the meantime, a surrogate rocket booster, a modified Minuteman ICBM used in all of the flight intercepts tests, has been the direct cause of three major failures. So, program officials saw little benefit in risking high profile future tests on that booster. Pentagon officials are now counting on new prototypes from Lockheed Martin and Orbital Sciences Corporation. Both booster designs are likely to have only one intercept attempt each before they are deployed next fall as part of the GMD system.

Pentagon officials were able to celebrate a rare success in August when Orbital Sciences successfully launched its booster carrying a mock kill vehicle. The Lockheed Martin design is awaiting a similar test as this article goes to press. But because of the absence of an effective booster, the testing of the overall GMD system has been set back.

Equally significant, some of the remaining flight intercept tests are gradually being downscaled from full flight intercept tests to “radar characterization” tests and other simulations that do not require the interceptor actually to hit its target. For example, two flight intercept tests planned to have been in the remaining months before deployment have been cancelled and replaced by non-intercept, radar characterization tests. Nor is much progress being demonstrated in the ability of the system to discriminate between actual enemy warheads and decoys and countermeasures.

Not only does the lack of stressing flight intercept tests undermine military effectiveness, it also weakens public accountability. Given MDA policy to classify information about these tests, it is difficult for Congress or the press to track, let alone confront, the agency’s claims based on non-flight intercept tests, ground tests, or other simulations that do not involve the clear test of whether an interceptor hits its target.

Effectiveness Standards

Last December, Secretary of Defense Donald Rumsfeld acknowledged that missile defenses would not be very good at first. The capability would not be defined by the classic military phrase “Interim Operational Capability”—namely something new with proven warfighting worth—but rather capability, as Rumsfeld put it, “with a small ‘c’”. Nevertheless, he said, even at first this new missile defense would be “better than nothing.”

The president’s decision to deploy missile defenses is a remarkable example of a new procurement philosophy at the Pentagon called “capability-based acquisition,” which means the opposite of what it sounds like. The priority is on acquisition and deployment, not demonstrated, effective war-fighting capability. In a sign of the times, last January, Joint Chiefs Chairman Gen. Richard Myers circulated a new draft Instruction on the Joint Capabilities Integration and Development System that officially eliminates military requirements, replaces them with “capabilities,” and talks about “crafting capabilities within the art of the possible.”

In addition to speaking of capability-based acquisition, U.S. defense officials also talk in terms of spiral development or evolutionary acquisition. These terms are used more or less interchangeably and, except for the fact that they all describe an interactive approach for building capability, no one in the Pentagon seems to know what they really mean or how to implement them in practice.

The term “spiral development” originates with Professor Barry Boehm, director of the University of Southern California Center for Software Engineering. In Boehm’s model, rigorous testing is needed at each loop in the developmental process.1 In the Defense Department, however, spiral development is seen as a way to avoid testing and to cut corners.

The traditional Defense Department approach—sometimes called “fly before buy”—is to wait to procure a new military system until it has successfully demonstrated that it can work in realistic operational tests designed to simulate real-world conditions. For major defense acquisition systems, the law prohibits full-rate production until the system has been through realistic operational testing and the results are reported to the secretary of defense and the U.S. Congress. If the added military utility turns out to be only marginal, such systems are usually cancelled.

But major military development programs can take decades and, in an attempt to speed the process, capability-based acquisition was conceived. Capability-based acquisition aims to streamline the process drastically by shortening development time and deciding to fund marginal improvements to military value that might not have been considered worthy of funding in the past.

However, as the president’s missile defense decision shows, capability-based acquisition can mean buying new equipment that has not been through realistic operational testing and that offers little or no demonstrated military utility. Neither the GMD system, to be deployed near Fort Greely in Alaska, nor its sea-based adjuncts, to be deployed on Navy ships, has gotten far in its developmental testing; and neither has begun, let alone completed, more stressing and realistic operational tests. That is why the development and testing for both these systems—over the next two years and both before and after initial deployment in 2004—will be so important.

The shift to capability-based acquisition leads to confusion all the way around.  For example, a recent General Accounting Office report2 on the readiness of technology to support missile defense evaluated those technologies against a lesser standard than would be required actually to defend the United States from a realistic threat.

Rumsfeld cites the Predator Unmanned Aerial Vehicle as an example of this “capability-based” approach. In one sense it is not, because the Predator went through realistic operational testing more than two years ago. In another sense, however, the Predator is an ideal example of capability-based acquisition in that it did not meet its military requirements in those tests, was found to be not effective and not suitable, and yet proved its worth in the wars in Afghanistan and in Iraq.3 The Predator brings a new type of reconnaissance to the battlefield, and although that capability falls short of what its military users wanted and still want, it is much “better than nothing.”

Of course, such judgments are relative. The current version of the Predator costs only about $3.5 million apiece, whereas the GMD system is estimated to cost at least $70 billion, although no one knows what the final bill will be.

The PAC-3 Missile

On the other hand, the Patriot Advanced Capability 3, or PAC-3, illustrates a more traditional defense procurement cycle. Finished with a first phase of both developmental and operational testing, it is in low- rate production and was deployed in limited numbers in the Persian Gulf for use in the war in Iraq. Although it appeared to be doing well in development tests—hitting 10 out of 11 targets—those early tests involved the usual artificialities of preplanned intercepts. In more realistic operational tests conducted last year, the PAC-3 hit only three targets out of seven tries, or less than 45 percent.

Even more critically, none of those tests included Scud missiles, its intended targets. Nor did any of the Patriot’s successes in the recent war come against Scuds. If Saddam Hussein had the faster and longer range Scud missiles, he never used them. So, the performance of the PAC-3 against Scuds has still not been demonstrated in combat and still has not been tested at home. Otherwise, the results in Iraq so far appear to be consistent with the results from operational testing.

Recognizing that the PAC-3 requires further testing, the president’s budget request for fiscal year 2004 lays out another 23 new flight intercept tests to be conducted beginning in 2003 through 2006. The first of these will be developmental tests with more realistic operational tests to follow.

Yet, proponents of missile defense have cited the “successes” of Patriot missiles in Iraq as proof that strategic missile defense can work as well. But if the Patriot turns out to have significant effectiveness, it will be mostly because of a firing doctrine that shoots two or three Patriot missiles at each incoming target. No tests of the GMD system have been conducted involving either multiple incoming targets nor multiple attempts to kill a single target. Also, an ICBM travels farther, faster, and at higher altitudes than short-range tactical missiles such as Scuds and may carry decoys and countermeasures.

Assuming the Bush administration goes ahead with deployment, following only one or two more flight intercept tests, what kind of system will we have?

At a Space and Missile Defense Conference last August, Major Gen. John Holly, GMD program director, reportedly said the GMD system would initially have 70 percent of its required capability. Such a claim misleads Congress and the American taxpayer. In 2004, 70 percent of the required engineering and testing will not have been completed, 70 percent of the major system elements required to find and discriminate targets will not be operational, and the GMD system will not have demonstrated the capability to shoot down 70 percent of enemy missiles launched toward the United States. For the GMD system to work in 2004, it requires the MDA getting advance notice from the enemy—say, North Korea. This is because the GMD system has never been tested without enemy target information being provided—and preprogrammed into the system—well in advance of interceptor launch and without an element of surprise. North Korea would probably not be so obliging.

Target Discrimination

Since the Union of Concerned Scientists Report on Countermeasures4 was published in April 2000, the most persistent criticism of the GMD program is that it has not demonstrated that it can deal with even relatively simple countermeasures. (See page 10 on boost-phase defense proposals to deal with this problem.) Early tests included between one and three balloons that did not resemble the target re-entry vehicle in signature, motion, or shape. Tests need to be done with decoys that resemble the target reentry vehicle in convincing ways. To be believable, the GMD program must demonstrate that when a decoy actually resembles the target re-entry vehicle in some way, the Exoatmospheric Kill Vehicle (EKV) can still tell the difference. To do this, at the very least the GMD program needs the combined capabilities of high-quality X-band radars, heat-sensing missile discriminating satellites, and interceptors with target discrimination capabilities as well. Problems continue in all three areas, meaning that if a “capability-based system” is deployed in 2004, it will have essentially no real capability.

X-band Radar

During the Clinton administration, Lieutenant General Kadish told Congress that establishing an X-band radar on Shemya in Alaska was “the long pole in the tent” for the then-NMD system—i.e., the X-band radar was pacing the overall deployment schedule. The X-band radar is needed to provide essential target tracking and countermeasures discrimination information to the interceptor missiles.

The Bush administration decided not to request funding for an X-band radar at Shemya and instead has let new contracts for a self-propelled, floating X-band radar to be deployed on a modified oil-drilling platform. The floating radar can engage a wider variety of missile flight test trajectories than a fixed radar at Shemya. But construction of the sea-based radar is not expected to be finished until 2005 and then must undergo seven months of testing and be towed around South America to the Pacific Ocean because the huge platform will not fit through the Panama Canal.

Space-Based Infrared Satellites

The Space-Based Infrared Satellite (SBIRS) program has had two parts: SBIRS-high and SBIRS-low. SBIRS-high is a replacement for the Defense Support Program missile launch warning satellites, the first of which were deployed in the early 1970s. Over the years, at least 23 DSP satellites have been launched. SBIRS-high has fallen years behind schedule, and its cost estimate has increased from $4.1 billion to $8.5 billion with no satellites as yet launched.

SBIRS-low, a constellation of up to 30 satellites in low-Earth orbit, was to track and characterize enemy missiles and discriminate decoys from the target re-entry vehicle in the target cluster for the GMD system. SBIRS-low has also fallen years behind schedule, and cost estimates have increased from about $10 billion to $23 billion. In 2002 the SBIRS-low program was restructured and renamed the Space Tracking and Surveillance System (STSS). The new name implies that the original discrimination objectives of SBIRS-low may not be met by STSS. Also, fewer satellites are being studied, and there is no official cost estimate for STSS. The MDA is studying STSS constellations of various sizes, for example nine, 18, or 27 satellites, and the best way to obtain early test results from the first one or two satellites in orbit. But no operational testing is now slated until at least 2007.

Ironically, Bush’s June 2002 abrogation of the 1972 U.S.-Russian Anti-Ballistic Missile (ABM) Treaty has raised questions over whether STSS should be built as planned. Without the constraints of the ABM Treaty, the United States can now deploy land-or sea-based missile-tracking radar systems close to perceived enemy countries such as North Korea. The ABM Treaty did not permit forward-based missile tracking radars of this sort.

At a Senate hearing on April 9, 2003, MDA head Lieutenant General Kadish said the Pentagon was “rethinking the overall sensor requirements for a system without a treaty restriction.” He further added that “there is a major debate inside the community over whether we should have space sensors or terrestrial-based radars or a combination of both, based on affordability reasons and a whole host of other technical issues.” Such a debate, although welcome technically, means that the basic architecture of which STSS might be a part is undetermined and that a combined network of radars and infrared sensors on land, at sea, and in space certainly will not be operational in 2004, if this decade.

Interceptor Target Discrimination Capability

Because tests have not been done with decoys that resemble the target re-entry vehicle in convincing ways, the interceptors have not yet been stressed to show that they can deal with such countermeasures. A tumbling target presents discrimination challenges in relation to other objects in the target cluster that may also be tumbling, such as the bus.

The last flight intercept attempt, IFT-10, which failed because the EKV did not separate from the booster, was to have been the first nighttime test, designed to show that GMD interceptors can discriminate objects under different lighting conditions. All of the previous flight intercept tests were conducted in daylight. Because IFT-10 failed for other reasons, nothing was learned about nighttime discrimination in this test. Surprisingly, the MDA may go ahead with deployment in 2004 without trying to repeat a nighttime test. Agency officials have said only that the possibility that one of two remaining flight intercept tests before deployment may be conducted at night is under consideration.

All of the flight intercept tests so far have included both a C-band beacon and a GPS transponder on the target re-entry vehicle. The interceptor’s flight path is formulated using data derived from the C-band beacon onboard the target because there is no radar to track the target early in its flight. To be credible, the GBI must eventually show that it can hit a target with no targeting aids onboard the target re-entry vehicle.

All of the flight intercept tests so far have provided target information, such as its velocity and trajectory, to the interceptor before the intercept attempt. In actual combat, all of this information probably would not be available before interceptor launch. Tests need to be done that show that, if some or all of this advanced information is missing before launch, the interceptor can still successfully perform its mission.

In a recent report, the GAO also stated that the upgrade of the early warning radar at Shemya, called Cobra Dane, as well as that of another early warning radar in California, will not be completed by October 2004. That will push even further into the future testing to demonstrate that these radars can process and communicate needed information in real time to other elements of the GMD system.

Without the discrimination capabilities of an X-band radar, SBIRS-high and STSS (SBIRS-low) or a combined land/sea/space radar network, along with interceptor missiles with demonstrated discrimination capability, the GMD system to be deployed in 2004 will not have the major elements needed to be operationally effective and would appear to be just for show.

Beyond Initial Defensive Operations

Even before the Bush administration has demonstrated that the initial GMD system works properly, it is planning even more ambitious deployments. According to the Pentagon, there will be significant “block upgrades” every two years through 2014.

In fact, in its budget request for fiscal year 2004, the MDA describes its GMD program as part of “an integrated and evolutionary [BMDS] of initial modest capability.” The agency explains that “while there is only one BMDS, there is no final or fixed missile defense architecture.”

Adding another complication, administration officials have also said that they intend to build a “layered” system, capable of intercepting enemy missiles in all phases of flight—boost, midcourse, and terminal—and from platforms on land, at sea, from aircraft, and from space. The idea is that, if the interceptors in one layer miss their targets, the interceptors in subsequent layers will not.

The first step would be the deployment of as many as 20 missile interceptors to be placed on three Navy ships. Initially, the interceptors would be used only to guard against short- and medium-range missiles, as their interceptors will be too slow to be effective against the faster long-range missiles.

Pentagon officials acknowledge that such a sea-based system will not initially play a role in defending against long-range missiles. But they will be putting elements of the system in place beginning in 2005. The interceptors have scored three hits in four early intercept tests to date. Although the development of more powerful ship-based interceptors will take many years, the Pentagon still sees a useful role for ships in protecting against long-range missile threats. MDA officials would like the ships to serve as platforms for high-power phased-array radar systems that can be moved close to trouble spots such as North Korea. That would allow them to help track an enemy ICBM early in its flight and relay that information to the battle command and control center or to a strategic missile interceptor in flight. Fifteen ships are to be upgraded with such advanced radar systems by the end of 2005.


Now, with only a year to go, the pressure is on. But difficulties in the development program and delays in the major elements of the GMD system have made it clear that, if anything is deployed next fall, it will be more of a scarecrow than a realistic or effective missile defense capability.

Accordingly, the president’s decision to deploy the GMD system in Alaska by the end of fiscal year 2004 has changed everything but changed nothing. To be sure, it has reordered the priorities for engineers and scientists working in the program, as well as curtailed realistic flight intercept testing and progress in target discrimination. It also has changed the standards of effectiveness that the program must achieve and has led to massive construction at Fort Greely.

The President’s decision has also served to illustrate the problems with a capability-based approach to testing. As it is being implemented for missile defense, the new emphasis on capability-based acquisition means buying new equipment that has not been through realistic operational testing and which will have little or no demonstrated military utility in 2004. The Pentagon’s most successful development programs, such as the satellite- and laser-guided precision weapons demonstrated in Iraq, continue to rely on rigorous testing. This includes the Patriot program, which is vastly improved since 1991 but is still imperfect. Moreover, the successes with the Patriot missile in Iraq against relatively slow, low-flying, short-range missiles do not mean that missile defense against ICBMs will quickly follow. The concepts may seem similar, but the missions and technical challenges are completely different.

So, a choice must be made: Rumsfeld can either meet a political imperative by October 2004 or build a missile defense system that works. But the technical and operational challenges of an effective missile defense system are such that the Pentagon cannot do both.

Some Future Flight Intercept Test Milestones

First intercept attempt with a booster that is to be part of the deployed system Originally scheduled for late 2000, but development of a three-stage booster has been significantly delayed. Two competing models are currently being evaluated. All intercept tests to date have used a less powerful, surrogate two-stage booster. Each of the models is to be tested against a target in early 2004.
First intercept at night A December 11, 2002, test was supposed to prove this capability, but the system’s interceptor failed to function properly minutes after its launch, causing the test to fail. Whether the system can hit a target at night remains undetermined. MDA may try to prove this capability in its upcoming intercept test in early 2004, but it has not confirmed any plans to do so.
First intercept without C-band beacon targeting aids In intercept tests to date, a C-band beacon is attached to the target. Data from this beacon is used to calculate the intercept point. This test artificiality has been necessesary because of delays in the X-band radar to track the target early in the flight tests. MDA may try to prove this capability in its upcoming intercept test in early 2004, but it has not confirmed any plans to do so.
First intercept with a tumbling re-entry vehicle (RV) A tumbling RV, as opposed to one that is spinning, could present the interceptor with a more difficult target to hit. Previous plans called for a tumbling RV to be used in an intercept test in early 2001, but such a target has yet to be used. Now unknown
First intercept with decoy
balloons that closely resemble the target
Earlier Pentagon plans called for more realistic decoys to start being used in intercept tests beginning in the summer of 2002. But balloon decoys used to date reportedly do not closely resemble the target. Now unknown
First intercept against threat- representative target Previously planned for the first few months of 2003, but yet to be carried out. Now unknown
First intercept test with multiple targets MDA informed Congress in March 2002 that the system should be able to handle multiple targets. In December 2002, MDA projected such tests could take place as early as 2005. Now unknown
First realistic operational test The Pentagon previously estimated that a real-world-type test would occur in late 2004. Now unknown

Sources: Fiscal 2004 Missile Defense Agency budget request.


1. B. W. Boehm, “A Spiral Model of Software Development and Enhancement,” IEEE Computer, May 1988, pp. 61-72.

2. General Accounting Office, “Additional Knowledge Needed in Developing Systems for Intercepting Long-Range Missiles,’’ AO 03-600, August 2003.

3. The military users wanted the Predator to be able to fly in less than ideal weather and at night, to provide reliable communications, to be able to complete flights without having to abort, and to meet its maintenance requirements so as not to add substantially to the burdens U.S. troops already face overseas. Also, the military users wanted better discrimination from the Predator’s sensors. For example, in operational tests, the infrared sensor on the Predator only achieved a five percent probability of recognizing a tank, such as a Russian T-72 or U.S. M1A1 tank, compared with the desired, objective probability of 90 percent. The Predator also was unable to locate targets as accurately as required.

4. Andrew M. Sessler, et al. “Countermeasures: A Technical Evaluation of the Operational Effectiveness of the Planned U.S. National Missile Defense System,” UCS/MIT, April 2000.


Philip E. Coyle is the former assistant secretary for test and evaluation at the Pentagon, and currently a senior adviser at the Center for
Defense Information.



Posted: October 1, 2003