The Coming Swarm Might Be Dead on Arrival


In the past several years, there has been much wailing and rending of garments about major-power competition. History is back, so people say, and with it, the specter of a possible large scale, state-on-state war.

Upon waking up to a future of warfare that requires more than blowing up individuals from safe distances, the U.S. military and defense establishment noticed that possible adversaries — from China and Russia to North Korea — understand the American way of war and have generated forces and methods to counter it. These concepts are generally described as anti-access and area denial (or in Pentagonese, A2/AD) and include various methods and systems designed to prevent U.S. forces from entering an area of operations and to interfere with their employment once they enter that area.

Mainly, this is to be accomplished by use of successive defensive belts consisting of precision weapons (such as long- and short-range missiles and rockets), cheap area-effect weapons (such as mines, both naval and land, and other explosive devices, improvised or otherwise) and attacks against the main advantage of modern western powers — the military network (via electronic warfare, cyber operations, and information operations).

To counter these trends, the U.S. military has embarked on a quest to regain its superiority, by developing “offsetting” technologies. The main technologies that are the focus of this endeavor have been described by the Pentagon as robotics, automation, big data, additive manufacturing and miniaturization, and human-machine teaming.

Though it is not yet clear how these technologies would be used in the real world, more and more voices — both within the Pentagon and outside — are talking about the “autonomous swarm.” For the purpose of this article, we will define swarm as a form of attack that is based on a convergence of multiple autonomous units (formations or individuals) from various directions on a single target or location. According to proponents of the autonomous swarm, we will soon see mass production of autonomous and cheap machines. These machines will be employed en masse to the enemy’s A2/AD defensive belts, overwhelm them, and enable U.S. forces to enter the area, operate in it, and fight and win America’s wars. Proponents of the “autonomous swarm” are talking about changing the cost equation, about how unsustainable it is to shoot down a $200 drone with a $3 million missile and about flipping this entire balance of power on its head.

And in the current state of affairs they are right. Modern war has become very discriminating, almost “surgical,” with missiles coming through windows, ever smaller munitions, increasingly expensive (and thus fewer) platforms and militaries (voluntarily) giving up their most destructive, wide-effect systems. While some argue that the price of precision munitions is low and keeps getting lower, looking at the actual, up-to-date prices proves otherwise. As can be seen from Figure 1 below, the simplest precision guided munition — the Joint Direct Attack Munition (JDAM) — costs tens of thousands of dollars. This is not cheap by any means and it only gets more expensive from there on. Since the underlying technology is well-known and in mass production, the chances of massive reductions in costs of ammunition (or the platforms that carry them) are low. With rising prices and falling inventories, it seems that reintroduction of cheap ordnance to the field en masse, will, indeed, change the character of war.

(Based on data from Program Acquisition Cost by Weapon System – Fiscal Year 2019 Budget Request, Washington, DC: Department of Defense, Office of the Undersecretary of Defense – Comptroller, February 2019)

However, with their enthusiasm for looking to future technologies, proponents of the swarm transformation sometimes neglect (again, one must say) to look to the past. The sad truth is that the technology to cause widespread destruction on the cheap is readily available. In fact, it is a technology so old, it was invented before militaries had started talking in terms of “offsets” or “revolutions.” It is industrial warfare 1.0, which can now be employed with an information technology twist.

As an example to explain the rule, one can take the idea of the autonomous aerial swarm — waves of cheap flying robots, coordinated with artificial intelligence, overwhelming future, helpless, ground forces. The success of this concept is predicated on the target’s air defense being based on high performance anti-air missiles, designed to shoot down fast, high-flying jet fighters. This requires agile missiles, with terminal guidance, which are not cheap, as can be seen in Figure 2 below.

(Based on data from Program Acquisition Cost by Weapon System – Fiscal Year 2019 Budget Request, Washington, DC: Department of Defense, Office of the Undersecretary of Defense – Comptroller, February 2019)

However, cheap, small drones exhibit none of the advantages of fast, more expensive, jets. They are neither nimble, nor fast, nor high-flying. They are more akin to early airplanes — fragile, slow, and low-altitude. To make them better would require making them more expensive. Making the individual platform more expensive (and complex to manufacture) would limit the number of individual platforms purchased and operated.

Some say that the autonomous swarm itself will be simple, but the munition it carries could be sophisticated and have “stand-off” range. Though, as can be seen in Figure 1 above, making a munition long-range, or even guided, causes the price of the munition to skyrocket.

To understand the premium of stand-off range, just compare the price of the Joint Air-to-Surface Standoff Missile ($1.4 million) with that of the JDAM. To understand the premium of even the simplest guidance system, just compare the price of a simple JDAM guidance kit (roughly $25,000) with that of the unguided BLU-117 General Purpose Bomb (roughly $5,400).

As can be seen in Figures 3 and 4 below, price has a direct correlation to the number of items purchased, with the numbers of simple and cheap JDAM being purchased in the tens of thousands, and more complex munitions being purchased in much lower numbers, down to the low hundreds. Furthermore, if we look at the absolute simplest munitions — such as M-795 high explosive 155 mm shells — we would see that each shell costs merely $1,375 and that in Fiscal Year 2019 the Army is on track to purchase 203,876 such shells.

The differences are so extreme that two figures had to be included to accurately illustrate the gap between more complex and less complex precision munitions purchased. Figure 3 shows the Department of Defense projected purchases of most munitions considered to be a “Major Defense Acquisition Programs;” Figure 4 shows shows those same programs, without the largest four (in terms of number of items purchased) — which are, incidentally, also the cheapest and simplest ones.

(Based on data from Program Acquisition Cost by Weapon System – Fiscal Year 2019 Budget Request, Washington, DC: Department of Defense, Office of the Undersecretary of Defense – Comptroller, February 2019)

There is no way to break this iron law of capability. Nothing can simultaneously be cheap, fast, maneuverable, and have long range. Truly large swarms will have to be made of simple machines, with either limited speed, limited range, limited protection (both physical and electronic), limited or cheap payload — and probably all of the above. If either the platform or its payload is not cheap and simple to mass produce, the very idea of the swarm — the multitude — will be undermined.

Militaries already developed a solution for the problem of many “small, slow and low” things – be they platforms or munitions – in the form of the anti-air artillery. The most known, though certainly not only, weapon of such type is probably the German 88 mm cannon. This technology posed a significant challenge to World War II air raids that contained hundreds of airplanes at a time — a veritable “swarm” (before it was fashionable to call it that). And that was before the information and communications technology revolution. Back then, anti-aircraft gunners had to aim the gun, using crude aids from optical or mechanical sights to some form of early radar assistance, and fire volleys of shells to hit these relatively small targets. Against high-flying targets, effects of such anti-aircraft artillery could be more limited. However, against low-flying targets, such fires were devastating — causing between 13 percent and 44 percent of all downed aircraft between February and May 1944.

Hitting a target with a projectile requires knowing its azimuth, range, timing, and altitude. At the time of World War II, measuring and calculating these was not an easy feat. However, the advent of information technology has cut both ways. The same technology that will enable the autonomous swarm already enables radars and electro-optical sensors to identify smaller and larger numbers of targets, and calculate the optimal points to aim cheap shells for maximum effect.

A German 88 mm flak gun had a maximum ceiling of around 30,000 feet and an effective kill radius of around 18 meters. Let’s assume, based on ground-level incapacitation probabilities, that a 105 mm high explosive, high-fragmentation shell has a “kill radius” against drones of 50 meters, or a kill sphere of roughly 523,600 cubic meters. Such a round costs $3000 and will probably cost less if produced and procured on massive scales — akin to the 155 mm shell mentioned above. Hence, simple cannons, similar to ones currently in service, can cover vast areas within minutes, severely degrading, if not destroying, a large part of any swarm that attacks it. If fired against targets on land, such cannons can reach a range of up to 13 km (PDF). Depending on the altitude of the target, this range will decrease, but such a cannon will probably be able to engage aerial targets several kilometers away, given proper sensors and targeting computers. Since there is no requirement to hit a very precise target in the case of a a slow aerial target, let alone a swarm of them, there is no need for terminal guidance — only a relatively simple timer or altitude fuse, and a large fragmentation warhead.

Existing cannons, with existing shells, can reach altitudes of tens of thousands of feet. Existing counterbattery radar systems boast the ability to detect not only artillery shells, but also drones. Every tank gunner has seen, firsthand, what even a 1980s and 1990s-era fire control system (and manual control) can do with ballistic fires. Trained gunners can hit, moving (even aerial) targets, from several kilometers away, with technology that is decades old. The laws of ballistics are well-known — so well, that in our quest for ever more accurate terminal guidance, we might have forgotten how accurate one can shoot even without expensive terminal guidance.

If operated as single platforms, such cannons will be vulnerable to attack by various kinds of platforms (simple drones; stand-off munitions) attacking from various directions. However, one must remember that no single air defense platform operates on its own. The relative cheapness of tube artillery and ammunition will enable an adversary on the defense to deploy a large network of dispersed, mutually supporting systems connected with each other and with sensors by a wide network of fiber optics to defend both fixed installations and deployed forces. On the offensive, said adversary will be able to deploy self-propelled systems, with integral sensors and the ability to connect to external sensors, to protect the maneuvering forces against swarming aerial attacks. This is, in effect, similar to systems deployed (both currently and previously) by the Russian military — from the venerable ZSU-23-4 Shilka anti-aircraft gun to the more modern Pantsyr-S1 (SA-22) air defense system — to protect the maneuvering forces.

Nor is the effectiveness of air defense artillery against modern aviation a theoretical matter either. During Operation Desert Storm, air defense artillery levied costs on coalition air operations when those airplanes had to go down to lower altitudes. During the 2003 battle of Karbala, an entire AH-64 Apache regiment (29 out of 31 helicopters) was taken out of the fight by small- and medium-caliber fires. What if these assets were manned and managed by a force better trained and equipped than the Iraqi army of 1991 or 2003?

One should note that the utility of air defense artillery in the modern war is not lost on current adversaries. Both Russia and China still operate air defense artillery systems — and indeed, deployed Russian forces have already intercepted several drone swarm attacks, with current equipment. Furthermore, the Russians are even introducing new air defense artillery systems, with larger caliber guns.

All this will cheapen air defense by orders of magnitude, enabling even a backwards military to defend against the most sophisticated military gear not yet available.

Some have speculated that delivering a large enough warhead to an estimated position of a target in the sky might be adequate even against stealth airplanes. If that is indeed so, large caliber air defense artillery, might be able to do exactly that, delivering large numbers of large “warheads” to high altitudes, without the requirements to accurately fix the target.

Meanwhile, current swarming drones and loitering munitions are anything but cheap. Military drones tend to cost anywhere from $28,000 to several million dollars. The price will have to drop by orders of magnitude just to reach the cost of a single shell, let alone to offset the cost effectiveness of one shell taking out several drones. All this before we factor in the massive research and development costs required to develop the level of autonomy that autonomous swarm proponents seek. So, we can see that despite dreams of massive, cost-effective, swarms that will engulf the enemy, the technology to counter such swarms already exists, even before we talk about future technologies. These technologies, of course, are not limited to an integrated network of air defense artillery, as described above, but can range from experimental counterswarm energy weapons to electronic warfare and experimental munitions.

Air defense artillery is, of course, not a panacea and does not fit all scenarios and every threat. Any good defense is composed of a variety of systems deployed across the entire battlespace. As such, to be successful, air defense artillery must be integrated with other measures such as electronic warfare, surface-to-air missile air defense, and an extensive network of sensors of multiple types. However, air defense artillery has an enduring and substantial utility, which many Western militaries have ignored for many years. The characteristics of these systems might make them a good solution for countering swarming attacks — whether like those seen in Syria today or those predicted by advocates of the “autonomous swarm” concept.

The discussion above about aerial autonomous swarms is also applicable to swarms on other domains. Land swarms can be countered just as well with large volleys of thermobaric or cluster munitions, both currently in service in China and Russia. In fact, World War I saw hundreds of thousands of soldiers cut down with manually aimed artillery and machine gun fire, causing such large-scale destruction that it can still be seen today.

The point of this article is that no form of autonomous swarm proposed or feasible today would be larger than the massed formations on land and air already seen in the major battles of either world war — if only for the challenges of physical space in the field. Yet the ability to accurately engage targets with dumb, cheap, munitions has only improved since — courtesy of the same information technology that will enable the autonomous drone swarm. The fabled increase in computing power that allows for the remote or autonomous employment of multiple drones will also enable a defender to use multiple sensors to detect and track more, smaller, targets at longer ranges and calculate the firing solutions to engage them with large volumes of dumb and cheap munitions. However, while the technology of the autonomous swarm is still under development, the technology required for massive fires and networked, multi-sensored defenses is already on the shelves — indeed, has been for decades.

To sum everything up, this article does not to say that drones, as individual platforms or as swarms, will not have their place on the battlefield, nor does it seek to advance one weapon system over the other. However, it is clear that commentators and technologists should caveat their current discussions and maybe temper some of their enthusiasm. Also, it is time to rethink some commonly held beliefs about the so-called third offset and its technologies. A real offset might not lie in inventing new technologies (let alone doubling down on those that are merely more of the same). Rather, much might be achieved with the combination of tested and true technologies (and concepts), such as information technology, initial guidance, and “dumb” ballistic munitions. Introduction of advanced munitions, such as the Hyper Velocity Projectile, might serve to further reinforce the capabilities of ballistic munitions to cause large-scale destruction on the cheap, against targets at even greater ranges or altitudes. The emerging preference of the U.S. Defense Department for hypersonic over automation and of the Hyper Velocity Projectile over the railgun might signal a recognition of the values of ballistic fires and the vulnerabilities inherent in the concept of the swarm.

From the vantage point of this author, it seems that the swarm might be coming, but it might turn out to be dead on arrival.


Shmuel Shmuel (yes, that is a real name) is a military analyst from the Dado Center for Interdisciplinary Military Studies. You can follow him on Twitter @Samdavaham, or at the War Hall. The views expressed above are those of the author and do not represent those of his employer.

Image: Flickr/Ars Electronica