Quite a few Serbian anti-aircraft crews died in such strikes. To us, flying at a moderate height in an unarmed, converted passenger jet, the support was comforting. It was good to learn afterwards that NATO forces never lost a single airborne tanker as a result of enemy activity.

The refuelling system was basic. In the KC-135, the boom operator worked in a prone position. In the KC-10, by contrast, there was a small chamber at the rear with three seats and a large reinforced window, almost two by six feet in size, which the operator used to observe everything going on below. The boom would be lowered and he’d watch the receiving aircraft coming in, directing him by radio if necessary to make small adjustments in speed or altitude.

We’d observe the entire process carefully while a system of mirrors allowed the ‘boomer’ to monitor other aircraft in formation off the KC 10 wing tips. The boom was ten feet longer than the KC-135’s and featured ‘fly by wire’ flight controls and an increased fuel transfer rate of up to 1,500 gallons per minute.

Then followed our sortie on board one of the Joint-STARS Boeings, which almost didn’t happen for me because my security clearance hadn’t come through from Washington by the time that crew call was sounded. Five minutes later, one of the officers poked his head through a door and gave a shout. I was on my way, the first non-American civilian to have this experience.

But first, some background to what it’s all about.

One of the concepts that originally brought the Joint-STARS project to fruition was the fundamental truth about conflict: in war, he who achieves better movement on the battlefield invariably dictates its outcome. That little maxim is as apposite today as it was in the time of Napoleon’s swift attack at Austerlitz. It was equally valid in Vietnam and in Operation Desert Storm, which is one of the reasons why the system is referred to as ‘The Window to the Battlefield’.

The difference today is that Joint-STARS is designed to give those involved in a ground war – whether it be Kosovo, Iraq, the Philippines or Afghanistan (or Syria or Iran in the future) – the living dynamics of the battlefield. Most significant, it comes in real time.

In the same way that the E3 Sentry (AWACS) provides all-weather surveillance, command, control and communications in the air, the Joint-STARS concept is land-fixated. Operational procedures allow it to act in concert with ground-hugging AH-64 Longbow Apaches by digitally passing along targeting information. Co-ordinated with a good mix of manned and unmanned assets, it has the potential for a truly seamless data transition, either in peacekeeping or in the sort of debacle that the Allies faced in Kosovo.

When we were aloft over Kosovo there were four or five of these machines in full operation, several of them based in Okinawa to monitor developments in North Korea and, obviously, offshore China. The tally of converted Boeing 707-300 series aircraft for this work is ultimately likely to number more than two dozen, with a fair proportion designated for the Far East.

As explained by Lieutenant Colonel Kevin C. Peterson and Major Phillip G. Basinger in a post-Kosovo report, the Joint-STARS is an Army-Air Force system designed to provide immediate surveillance intelligence, targeting, and battle management to the land component commander:

The system is made to support a corps-size unit and consists of a USAF and Army aircrew, and what we like to call the ‘business end of the system’, the Ground Station Module (GSM), operated by the Army.

The E-8, using its chin-mounted multimode radar, collects moving target indicators (MTIs), fixed target indicators (FTIs) as well as synthetic aperture radar (SAR) imagery. It downlinks all this information to the GSM.

GSMs not in the footprint of the aircraft data-link might task another GSM to relay the data through a satellite at a reduced data rate using built-in satellite communications radio.

Once fielded to military intelligence (MI), aviation, and artillery units, the GSM will be the most numerous MI end-item in the Army, located from maneuver brigade up through echelons above corps (EAC).

Our crew on take-off from Frankfurt numbered 23. Two were US Army specialists, one of whom was the deputy mission crew commander (DMCC), and this pair operated at the rear of the plane in conjunction with the air force specialists onboard. It was their job, together with those manning the other 16 work stations, to provide intelligence and targeting information to the Combined Air Operations Centre (CAOC) at its Southern European headquarters in Italy.

Information requests such as area searches, SAR images, and a good deal else besides, were relayed through workstations on board. There were a number of variations: more requests for data that came in from the Joint Analysis Centre at Molesworth in England (one of the primary downlink points for Joint-STARS data) or an asset on the ground that needed clarification about something that was visual.

There was also a direct link with our aircraft’s airborne adjunct, the EC-130 Airborne Battlefield Command, Control and Communications (ABCCC) and with the Army’s Ground Station Module through a secure Surveillance Control Data Link. I was told that up to 15 ground stations could simultaneously pass text messages back to the Joint-STARS and our three VHF radios were primarily used to communicate with them.

Once ‘in orbit’ adjacent to Kosovo, the aircraft flew tight figure-ofeight circuits of classified length. If anything on the ground moved, the aircraft, in theory, would detect, locate and support attacks against it. Because visibility was not a factor, no matter what the weather, such operations could go on around the clock, though most missions are a bit more than half a day. Ours lasted 14 hours and included an in-flight refuelling session from a KC-125 tanker.

Long hours aloft should be truly boring. However, in reality it stayed interesting from the start because there was always something happening. On one of the earliest flights over Kosovo, before we got there, one plane hadn’t been on-station for an hour when it was thrown into an almost vertical dive. That left some crew members who hadn’t been strapped in clinging to the roof of the aircraft.

No reason for evasive action was given. However, the crew was aware that such things only happened if threatened by hostile missile or aircraft action. What was clear was that there were either enemy fighters scrambling or there had been a missile alert.

The threat factor stayed sobering while we were aloft. Crews have refused to remain on-station without a combat air patrol (CAP), though numbers and types of fighter escorts were secret. The crew was emphatic: ‘No CAP and we’re out of here’, they would state.

The Joint-STARS defensive strategy is simple. Should enemy aircraft suddenly become a threat, the crew wouldn’t waste time before heading in the opposite direction.

The issue of Yugoslavian ground-to-air missiles remained uppermost in all our minds while we were up there. From what I was able to observe on the consoles, mobile SAM-6 batteries were thick on the ground in Kosovo. There were also long-range SAM-3 batteries, some of which had been targeted by the time we arrived, according to a Pentagon briefing a month earlier.

Certainly, that destructive potential was raised often enough during the course of hostilities. Missile (and Triple-A) threats, while not taken lightly, were invariably balanced against intelligence reports from ground observers and the fact that any Joint-STARS orbit was always beyond the maximum missile target range of the latest version of ground-to-air missiles that Belgrade acquired over the years from Moscow. Almost everything in the Serbian armoury, we were aware, was former Soviet Union in origin.

It was interesting that during our flight there were some active concentrations of ground forces radar-tracked near the southern and eastern borders of Kosovo. There were also indications of a lot of activity along the Albanian and Macedonian frontiers and it quickly became clear that the Yugoslav Army was preparing for a major ground action.

No US spokesmen would comment on this aspect, even though in places the ground was cluttered with hardware. That included armour, which indicated a substantial military presence.

Meanwhile, the US Air Force has lifted some security restrictions surrounding the ability of its long-range, air-to-ground surveillance system to locate, classify and track in virtually any weather, on-line and in near real-time, a variety of enemy ground targets.

Although the concept is old hat – a prototype Joint-STARS test-bed was originally used to track, locate and target Iraqi divisions in Operation Desert Storm (especially during the Battle of Al-Khafji) and Airborne Reconaissance Low (ARL) and the Mohawk systems are also active – very little about this long-range detection weapon has been made public.

Tactically, the Joint-STARS system operates from the safety of friendly airspace. Our aircraft, for instance, was able to peer about 150 miles across borders and do a stand-off analysis of enemy ground assets. Each aircraft carried a phased-array, mechanically operated radar antenna housed in a 25-foot dome or ‘canoe’ under the forward fuselage area. It weighed tons and we could feel it thudding as it alternated its scanning role from port to starboard. That equipment provided the necessary data for pinpoint strikes by aircraft, missiles or artillery for fire support.

We were also told that Joint-STARS ground-tracking radar, in a single eight-hour sortie, could search an estimated 200,000 square miles of terrain, though the officer warned that the aircraft was unable to identify targets. Other assets such as Unmanned Aerial Vehicles (UAVs) or Human Intelligence (Humint) might have been used to clarify the information still further.

Most details regarding operational and acquisition procedures on board Joint-STARS aircraft remained classified. However, information made available in a briefing by the unit commander at the Rhein-Main Air Force Base, indicated that its fundamental operating modes included Wide Area Moving Target Indicator (MTI) surveillance and Synthetic Aperture Radar (SAR).

Covering a 60-square-mile block, the images brought up on the aircraft’s monitors could be split into eight grid areas. These were electronically scanned in azimuth with a rapid revisit rate of seconds. Or it could be narrowed down to search an area of a bit more than a square mile.

Depending on earlier intel reports – or something spotted on the screen – potential targets might have been subjected to intensive repetitive sector searches. Movement detection involved a Doppler shift system coupled to phased array radar.

For example, the system was sensitive enough to differentiate (but not specifically identify) between wheeled and tracked vehicles, picking up a reading bounced off the tracks of a moving tank or APC. This appeared on the monitor as a yellow dot: in the area over which we were orbiting there were stacks of them. What happened was that the MTI displayed moving vehicles or amour as moving imagery (dots) and SAR as still photos. Also, as we could see, MTI could be displayed on a variety of backgrounds from SARs to military maps to slope-shaded maps.

Much of the technology was space-age stuff. For instance, rudimentary analysis was provided by Joint-STARS’ Synthetic Aperture Radar/Fixed Target Indicator (SAR/FTI), which produced area images almost comparable to the negative of a photo, complete with shadows off surrounding mountains. In some areas the effect was almost 3-D.

By pushing out energy bursts in a succession of small grids (which, in turn, sent back images) it was able to identify clearly features such as runways, buildings, and aircraft on the ground, or even rail tracks. In turn, the system sounded out reflective energy, or, in the argot, ‘fine imagery’.

Another revolutionary development, and one more pointer to the future, was that each set of images could be automatically stored in electronic target folders and immediately downloaded into the aircraft’s database. Ground activity on a particular day might be called up on the aircraft’s computers an hour or a month later to examine before-andafter anomalies. Thus, our Joint-STARS platform had the ability to indicate immediately whether there had been any new troop arrivals, or, possibly, fresh construction sites in the area of operations.

In another area it might pinpoint repairs surreptitiously being made to a bridge that had been bombed a day or three months before. SAR data maps of Kosovo contained the precise locations of critical nonmoving targets such as bridges, secreted motor pools, harbours, factories, airports, buildings or camouflaged vehicles as well as armour.

Obviously, anything new-fangled always has detractors. From my own observations, the system appeared to be cumbersome. Following some serious tactical blunders that left innocents dead, follow-through instructions for any attack were always very carefully assessed (and sometimes re-assessed). Any strike needed to be filtered through half a dozen or more commands, having first been cleared from above. It could be a time-consuming process. Also, targets needed to be cleared by all NATO member nations, of which there were 19 at the time. That implied very specifically that this was the first war fought by ‘committee’. Even the original Korean War of the 1950s, a UN campaign from beginning to end, wasn’t hampered by anywhere near as much bureaucracy or obfuscation.

There are also those who reckon that the Joint-STARS system might have been too complex for the job, though this would be ignoring the basics of a quite remarkable system that was regarded as impractical a couple of decades ago. From personal observation, it was clear that there were time-delays in acquiring information and processing detail for a strike, which sometimes took a while. The sighting, for instance, of a tank or an enemy convoy on the ground, followed by a command to a fighter strike element nearby, might involve a ten-minute time lapse, sometimes longer. Independent verification, often from intelligence elements on the ground (an essential adjunct), involved still more time. By then the target might have disappeared into a mountain hideaway.

In an article titled ‘Too Much Data, Too Little Intelligence?’ the role of Joint-STARS in peacekeeping operations in Bosnia was analyzed by Lieutenant Colonel Colin Agee, deputy commander of an Airborne Military Intelligence Brigade out of Fort Bragg.

While conceding that Joint-STARS was lauded as a ‘star performer’ in Desert Storm, he wrote in Jane’s IDR Extra (May 1997) that as the IFOR mission proceeded, ‘the use of Joint-STARS to track military movement became harder’, in particular because of the difficult terrain. He went on:

Perhaps the biggest challenge to effective use of the system was data overload: the inability to distinguish significant MTIs from the voluminous data stream… The screen would quickly fill with MTIs, making it difficult to determine those of military significance.’ From personal observation (last May), it appeared to this observer that the system still needed some fine-tuning and on the face of it, still worked best with a good pair of eyes on the ground.

As he pointed out, during the first month of operations over the Balkans, the aircraft seemed to be hampered by a variety of technological problems, particularly in a flying electronic platform with so much potential for glitches. Any operational Joint-STARS flight has two electronics technicians onboard, but even then things happen.

In an early flight, one of the workstations had to be shut down because of smoke fumes detected by a crew member. Another report in Aviation Week (3 May 1999) said that their aircraft operated with a poor satellite communications antenna and a faulty vapour-cycle cooling system, which made for a cold working environment.

Also, trying to keep the aircraft in the air needs about 60 ground maintenance personnel and they seem to have done well. In the first month of Kosovo operations, the unit missed only one flight, significant because both planes were deployed over the Balkans every day. It was accepted logic that to be truly effective, three such aircraft were needed to provide NATO with around-the-clock surveillance.

The role of the Joint-STARS unit in southern Europe, we were initially briefed, was threefold.

In the Balkans war, Joint-STARS provided indications and warnings of friendly and enemy movement on the ground.

Its surveillance system allowed for intelligence assessment and, as a consequence, intensive preparation for future battlefield activities.

The system is totally uninhibited by cloud cover. A welldefined and powerful line-of-sight communications system, once on-station (and in conjunction with E-3 Sentry AWACS elements) becomes a factor in the further deployment of other airborne elements of the Theater Air Control Systems (AETACS). These can include assets such as the Air Force RC-135V/W River Joint (Sigint), an occasional U2, the Navy’s EP3, the Royal Navy’s Nimrod Electronic Intelligence Gathering aircraft, France’s RC12 surveillance system, Airborne Forward Air Control (ABFAC) as well as ABCCC along with ground TACS units.

Operating in a region as rugged and mountainous as the Balkans presents its own peculiar set of problems. Since the system is line-ofsight, the high, broken Balkan terrain tended to influence surveillance coverage through screening and disrupting communications. I was to observe that shadows caused by the mountainous terrain could be seen clearly on the SARs. Also, the shadow effect tended to assist the adversary in concealing assets.

Clearly, this affected radar images as much as it did contact with clandestine units on the ground – London sources indicated to this writer that ‘some’ British SBS (Special Boat Service) personnel were active inside Kosovo at the time and that American SEALS might have infiltrated Kosovo in small groups and were liaising with KLA forces active inside the disputed territory.

The SAR and FTI capability used in conjunction with MTI and MTIhistory displays allowed a limited post-attack assessment to be made by operators following a strike on hostile targets. Significantly, major advanced elements of the programme included software-intensive radar with several operating modes; the unique antenna with three receive ports; four high-speed processors capable of performing more than 600 million operations per second, as well as associated software.

In effect, though, while Joint-STARS is the magic bullet of the future and tends to reduce the need for visual armed reconnaissance sorties to locate targets, keeping friendly casualties to a minimum, it could never dispense with the human element altogether. The system works as follows.

While possibly monitoring a number of frequencies simultaneously, Joint-STARS – in conjunction with Combined Air Operations Centre – spots a convoy of vehicles moving somewhere within the target area of the future. First the specific area is electronically isolated. It is analyzed and assessed by ground staff at CAOC as well as by Joint-STARS personnel. Finally, the potential target is referenced to previous electronic imagery in the onboard database and graded according to priority. If the subject is confirmed hostile by other theatre assets, ABCCC might then order in a strike wing.

Prior to leaving Frankfurt, the pilot would receive pre-flight briefings that included details of up to 20 priority targets. All would have been pre-located by a combination of UAVs or the activity of infiltrated agents or something spotted on the ground by previous Joint-STARS flights.

Once in orbit I was shown the day’s target list that had been numbered according to priority. Meanwhile, the Mission Crew Commander (MCC) would start at the top and try to work his way down – in conjunction with a variety of strike assets that would do the actual hits – though the system was flexible. If something else of significance came up in the interim, it would have been targeted. Ultimately, the objective was that the Joint-STARS MCC would be working indirectly with AH-64 Apaches, though that concept still needed a measure of streamlining.

The choice of using sixties-vintage Boeing 707s as vehicles for the E-8C Joint-STARS programme – many of them acquired abroad from a variety of commercial airlines – was unusual. Most of the pilots complained about these Boeing 707s being underpowered. Others would have preferred more altitude.

Several factors for the choice prevailed, including budget constraints. Another was safety; the 707 passenger jet still holds the best commercial safety record. One of the officers involved disclosed that once the decision to proceed had been made, they scoured the world for them.

Size was of paramount importance. The 707 has the advantage of holding more of the operator workstations that are essential for onboard battle management. The need for more additional workstations has been seen in AWACS, simply because they influence the span of both intelligence and control, which, in the words of a spokesman for the manufacturers of the system, ‘are key factors in J-STARS’ ability to target large numbers of dynamic [moving] targets’.

The contract to deliver – initially at a cost of about $225m per machine, but these days, a good deal more – was awarded to the Northrop Grumman Corporation. The first four systems had an 11-hour endurance (22-hour flights with in-flight refuelling were routine).

The electronics on board any one of these aircraft were formidable and, since those early sorties in the Balkans, have been supplemented by a good deal more. Even then it was stressed that a single Joint-STARS Boeing 707 platform had a greater computer capacity than all the combined assets of America’s E-3 Sentry (AWACS) together: the USAF had 33 AWACS in its inventory at the time.

The main computers on board at that stage were 6 Mil Vax 866 mainframe computers. Two were systems monitoring, one was a hot spare. Three more were general-purpose computers, one of which handled local area network (LAN), another radar and the last was also a spare. Each of the 18 operational workstations on board was a Digital Equipment Corporation machine with a 133 Mhz processor, 512 MBs of RAM and two multi-GB hard drives (obviously this would be much advanced at the end of the first decade of the new millennium).

A point made by a technician on board was that they are prohibited from switching on the system over developed areas. The power activates every electronic garage door below the aircraft track.

Operating in the crowded air lanes over Southern Europe presented other problems. Balkan airspace was sometimes so crowded that it could be harrowing. Our own flight had a narrow miss with a fighter while onstation. A day later a Joint-STARS Boeing 707 en route to the Adriatic avoided a head-on collision with a passenger jet by going into a dive, pulling four or five Gs. Wrong data had apparently been given to the oncoming plane. Later the same day, there was another near miss in cloudy conditions when two of the escort fighters had to take evasive action within a couple of hundred yards of the Joint-STARS aircraft.

Several observers were of the opinion that allied aircraft proved to be a greater threat than any fighters that the Yugoslav Air Force might have been able to muster. Also, there was a lot of adverse comment about the ability of military ATC personnel to control the air movement associated with the campaign, which is possibly a lesson for the future about working over densely populated areas. Air traffic control was so bad sometimes that Joint-STARS aircrews were frequently ordered to operate under the ‘see and avoid’ rule.

As for the threat of enemy fighters, which anywhere else might have caused problems, the US Air Force had not yet funded a self-protection suite for Joint-STARS. NATO intelligence was only too well aware that a number of Yugoslav Air Force MiG-21s were stationed at Pristina, only minutes’ flying time from where we were airborne. They apparently resisted attack because all were secreted in deep mountainside hangers that made conventional attack impossible for the duration of the war.

It was interesting that the one time that Belgrade attempted to put Yugoslav Air Force MiGs in the air – possibly in a bid to stymie Joint-STARS activity – it took a handful of Coalition planes only 39 seconds to shoot down three of them.