The Nagorno-Karabakh war’s lessons are still being learned, but it is obvious that UAVs, or drones, played a huge role in the defeat of Armenian and Armenian-proxy forces.
Battlefield results show that Azerbaijani forces took out 241 Armenian main battle tanks (T-72s and T-90s), 50 BMD infantry fighting vehicles, 17 motorized artillery pieces, 9 radars, 2 SMERSH MLRS systems, 70 GRAD MLRS systems and a large number of vehicles including trucks.
But the most surprising result was the destruction of Armenian air defenses. Four S-300 air defense systems were knocked out, three TOR tracked air defense missile systems, 40 OSA 9K33 tactical air defense systems and five KUB 2K12 medium-range air defense systems were taken out.
As a result, virtually all of Armenia’s air defense systems were destroyed, leaving Armenian forces only with hand-held MANPADs for air defense.
In the many videos provided by Azerbaijan taken from surveillance drones or from the attack drone itself, there is no evidence that any Armenian radar or air defense system detected the attacking UAV or the accompanying observation drone before it was hit and destroyed. In the videos you actually see radar dishes turning as they are struck by UAVs.
This is not entirely new. In the September 2019 attack on two Saudi Arabian oil facilities at Abqaiq and Khurais, none of the Saudi air defenses detected the drone and cruise missile attacks.
In Saudi Arabia, three systems were guarding the oil fields and refining plants: the US-supplied Patriot, the Shashine Surface-to-Air Missile (SAM) system (France) and Oerlikon’s (Swiss) GDF radar-guided 35mm twin air defense cannons.
Shashine and Oerlikon systems were installed on the perimeter of Abqaiq and Khurais. The photos below were taken in 2019 and can be found in the Report of the UN-commissioned Panel of Experts.
As can be seen from the above images, Abqaiq from the north, east, southeast and south by air defenses. According to the UN Experts Team, it is now clear that the attacks came 27.00° N and 48.00° E, a waypoint location approximately 200 kilometers north-west of Abqaiq.
This has been established by exploiting the computer on a recovered drone that crashed before reaching its target. Its computer revealed the UAV was aimed at the Abqaiq facility. From this, it is clear that none of the Abqaiq or Khurais air defenses was able to detect either the drones or the cruise missiles used in the attack.
While the UAV strikes in the Nagorno-Karabakh war were mostly single-shot, discrete events, the strikes in Saudi Arabia in September 2019 were what amounts to swarming attacks.
For example, at Abqaiq, nine separator tanks were hit, three of them twice. These strikes all occurred in a period estimated to be about 20 seconds.
Here is a photo in the UN Report of one of the separator tanks that was hit twice:
These attacks were surprisingly accurate. Below is the diagram provided by the UN Yemen Panel:
UN Diagram of separator tanks at Abqaiq
Each separator tank is 28 meters in diameter and 9 meters high.
The central question raised by the attacks in Saudi Arabia and in Nagorno-Karabakh is why the air defense systems and radars did not detect the incoming drone threats. In the Saudi case, the radars also did not detect the cruise missiles.
The Russians experienced similar problems in Syria in attacks on the Khmeimim Air Base. On multiple occasions, the base was attacked by unmanned drones operating in swarms. The worst attack was in January 2018, when 10 drones rigged with explosive devices hit Russia’s Khmeimim airbase, while three targeted the Russian naval base in Tartus.
Russia claims it shot down seven of the drones using the Pantsir air defense system and was able to take control and land six others. Russia also said the airbase did not suffer any damage.
It is hard to say whether the Russian claims of success or their damage assessment is credible. The drones used in the attack were very low tech, made mainly of plywood and fabric with a small Chinese engine for power. Each drone carried 10 rocket grenades that were released by a simple, solenoid controlled mechanism.
Was the Pantsir as successful as the Russians claimed? What we do know is that a number of Pantsir air defense systems were destroyed by Turkish Bayraktar drones in Syria and in Lebanon.
On one day alone in Libya, nine Pantsirs were destroyed. The Bayraktar is a large drone. Its wingspan is 12 meters, considerably larger than the wingspan of an F-16 fighter jet at 9.96 meters.
The Pantsir was “improved” after the 2018 attacks, or at least the Russians said so. That’s because it was clear that the Pantsir operated poorly when Khmeimim was attacked. As a result, the Russian “success” claims should be dismissed.
Here is a photo from Khmeimim after the swarming drone attack in 2018. The aircraft looks like an Su-24.
The main issue is if conventional radars of air defense systems can detect drones? Israel has specialized in going after quite small rockets fired principally by Hamas in Gaza. Called Qassam, they do not escape Israel’s Iron Dome system.
Almost all these rockets have been hit while in flight – except those not targeted by Iron Dome – and once their engines are burning, they fly only at relatively low altitudes in an arc trajectory.
UAVs, on the contrary, do not have hot, burning rocket engines –they are typically powered by small internal combustion engines run under battery power. And many UAVs are constructed of plastic or composites, some home-built from wood such as balsam, so the only metal parts are the engines, which are typically quite small or not visible at all.
A good example is the Bayraktar TB2, which is made of composites and Kevlar, not metal. It uses a small Austrian internal combustion engine which is inside the fuselage and in the rear where it drives a pusher prop, also made of composite. Although it is large, it hardly has an IR signature and its radar signature is quite small, perhaps too small for easy detection.
There are radars that can detect small drones, but most conventional air defense systems do not have them. Such new radars work differently than conventional radars – they have very high-resolution scanning and they have computer algorithms that have signature data on different drone threats.
When there is enough radar imagery to run through the computer database, a drone sighting can be confirmed by the radar system and it can continue to track the object. The range is typically about one mile.
Looking beyond radars, there are other ways to detect a drone. It is possible to detect transmissions from a drone and locate it that way, through triangulation. In some cases, a drone can be detected by sophisticated optical sensors. And if the drone makes enough noise it can be tracked acoustically.
A modern drone detection system probably uses all these methods in combination and has elegant software to meld together all the information in near real-time to reach a solution on its track and how to eliminate the drone as a threat.
There was only one such system active in Nagorno-Karabakh that was brought in late in the war by the Russians. In Saudi Arabia, there was no drone detection system as far as known.
In Syria, there were air defenses, since augmented by improved Pantsir and by the S-400 at the Khmeimim base. The Russians rushed in their Krasukha jamming system to counter the successful use of both armed drones such as the Bayraktar and suicide drones like the Israel-made loitering munition known as Harop by Azerbaijan.
This would have been near the end of the war. So far as is known, the Krasukha jamming system was brought in to protect a Russian base near Yerevan. Even so, the Russians claim they knocked out 9 Bayraktar drones.
A question of accuracy
While the Nagorno-Karabakh war showed accurate hits on targets, all the drones were controlled remotely, sometimes called “man in the loop” operators. It also appears to be the case that the attack and suicide drones were linked through command centers to surveillance drones.
In practice, this meant that surveillance drones could pick out and track lucrative targets and the information would be used to call in the nearest attack drone.
This feature is seen in many of the videos that the Azeri defense ministry supplied online where the surveillance drone is viewing the target but where the rockets released by the drones arrive from a different angle and altitude. It is also clear in a video that tracks the destruction of two Smersh launchers.
In that example, the surveillance drone picks up the target, watches it launch rockets, follows it as it goes to a sheltered area and thereafter when an attack drone is available, the launchers are destroyed.
The Saudi situation was quite different. As the above diagram shows, the attacks were extremely accurate.
The question is, how come? The known communications range of the drones and cruise missiles that were used in the Saudi attack was too short to allow man in the loop operations. Unless, of course, the Iranians had new technology unknown to the west. If so, what was it?
There are four possibilities. These are: (1) completely autonomous drone operation not needing hot links back to a control center; (2) satellite links meaning that the drones could operate over very long distances; (3) a communications localizer of some kind, such as a transmitter-receiver located somewhere on Saudi Arabian territory or, alternatively, nearby; (4) or a more powerful onboard communications system.
Israel is convinced that Iran now has an autonomous system for its drones and cruise missiles, a system that is similar to the scene matching TERCOM that has long been part of the US Tomahawk cruise missile.
In fact, the Israelis are concerned that the Iranians are supplying Hezbollah in Lebanon and Syria with these new generation missiles, which is why Israel has been aggressively trying to destroy them in Syria and Lebanon.
Iran showed off a new missile called Mobin on August 21, 2019, at the MAKS airshow in Zhukovsky, Russia. The Iranians claimed that Mobin, by operating autonomously and being stealthy in design, could penetrate western air defenses.
According to the information they provided, Mobin was equipped with TERCOM and DSMAC, a scene matching artificial intelligence capability.
A second possibility is a satellite link for the UAV. This would require a satellite transmitter and receiver. None of the recovered cruise missiles or UAVs show a satellite link transmitter-receiver.
A third possibility is a communications localizer of some kind. None has been found.
Finally, a more powerful onboard transmitter-receiver is a possibility, but again there isn’t any evidence of one.
If there was a TERCOM-enabled autonomous capability in the drones and cruise missiles used in the Saudi Arabian attacks, then it is probable that surveillance drones were used some time ahead of the actual attack to carefully map the targets. That mapping information, including imagery, would then be loaded into the onboard TERCOM system.
At a minimum, the Iranian attack must have involved a long run-up to establish the targets and carry out the necessary weapon’s programming.
Since some of the drones and cruise missiles used in the attack have been recovered, and we know for sure that their computers have been exploited, then it is reasonable to assume that the US and other intelligence agencies know about Iran’s new capability, although they have not told the public.
Conventional air defense systems, whether US, European or Russian, do not reliably detect drones or cruise missiles.
In particular, the weakness of the US Patriot system and the failure of multiple Russian air defenses including the S-300 and Pantsir suggest that all of them have to be augmented by a new generation of air defense systems for the battlefield and for infrastructure protection.
A key problem for designers is that at present the range of high definition radars needed for identifying UAVs and small cruise missiles is limited. This means that many such systems are needed to defend borders, bases and high-value infrastructure. New technology is needed to significantly extend the range.
The rise of autonomous drones takes away one of the tools to defeat drones – namely communications jamming. While it still might be possible to blind a TERCOM system (for example with lasers), radio jamming of communications won’t work.
However, unless drones are equipped with accurate inertial guidance systems (for example ring laser gyros), the drones will need to rely on GPS. Efficient ways either to jam GPS or broadcast false parameters is a tool that has a future in combating drone attacks.
Today’s drones that are low powered and made of plastic or composites are naturally radar-evading. Optical scanning combined with thermal imaging offers a potentially better range than high-resolution radar and is a promising technology for future systems.
The US free and open policy on GPS needs re-evaluation, especially GPS availability in hot spot areas. While there are other GPS systems in the world such as Russia’s GLONASS and Europe’s Galileo, the possibility of a deal to control GPS accuracy in certain world hot spots may be an approach that could yield tangible benefits.
Thus a Satellite Security Arms Agreement (SSAA) ought to be on the agenda of US, European and Russian leaders.
Control of exotic, accurate gyroscopes should be made a policy priority. While these have proliferated in recent years, it may be possible to limit the supply and availability.