Counterswarming Is Imperative

Imagine a Russian drone swarm comprising eight Lancet loitering munitions and four first-person view (FPV) quadrotors over Ukraine. The fixed-wing Lancets have limited maneuverability but can carry deadly payloads, making them too big a threat to ignore. Suppose two Ukrainian Flakpanzer Gepard self-propelled antiaircraft guns (SPAAGs) shoot down seven of the Lancets. The feint sacrificed the Lancets but forced the Gepards to expose their position, creating a perfect opportunity for the FPV operators to coordinate an attack on one of the Gepards from four different directions. 

This example raises the question of how to counter cooperative and hetero-geneous swarms. The ubiquity of cheap drones imposes on defenders a cost dilemma regarding high-end systems such as surface-to-air missiles (SAMs). Commercial drones with cameras can cost a few hundred dollars. Each Iranian Shahed-136 costs roughly $20,000. For comparison, a short-range SAM such as the IRIS-T (supplied to Ukraine) costs $450,000. Further, any individual SAM launch system can carry at most a handful of missiles. Using scarce, expensive, and highly capable missiles against large quantities of small, cheap, simple drones means defenders will run out of deployed and stockpiled missiles quickly and at a high cost.

Some proposed economical solutions involve electronic warfare, high-energy lasers, and microwave systems such as the Marine Corps’ Light Marine Air Defense Integrated System (L-MADIS). While these technologies are promising, they are still mostly in testing. Guns, on the other hand have been important to air defense since World War I. And some interesting concepts adopt an “it takes a drone swarm to catch a drone swarm” approach.

Defensive Swarms

DARPA’s 2017 Service Academy Swarm Challenge addressed that idea by asking participants to suggest defensive tactics for friendly drones to protect a high-value unit from an attacking swarm.

The participants developed several algorithms. The first instructed defensive drones to target the closest attacker. The problem with this “greedy shooter” algorithm was that multiple defenders would assign themselves to the same unit. If a single attacker were far ahead of the rest, all defending drones would target it, leaving the rest unengaged. 

An improved algorithm—“smart shooter”—included target deconfliction. Defending drones would start by selecting the attacker closest to them, but then check if another defender was closer to it. If so, the first defender would select the next closest attacker, then check it, repeating the process until it found an unengaged attacker.

The final enhancement used point-of- intercept navigation. Rather than trying to fly directly toward the attacker (which often became a tail chase), under the “intercept shooter” algorithm, the defender would determine the attacker’s velocity and direction and fly a course to intercept based on its own velocity. 

The smart shooter algorithms work best for slow-moving drones, simple attacker behavior, and relatively close ranges. But more complicated and capable attacking swarms present problems.

Intercept shooter is superior, but it is far from ideal. Any sort of complex flight behavior by attackers can require potentially large course corrections to meet the new intercept point. This was solved using proportional navigation (sometimes known as “constant bearing, decreasing range”)—what many SAMs use. However, an attacker also can fly toward a high-value unit until a defender closes, then draw away the defender by changing course. But this can leave the defended unit vulnerable. 

Recent student research at the Naval Postgraduate School (NPS) has built on this intercept-shooter algorithm, though still using fairly simplistic engagement scenarios. While this has helped determine performance tradeoffs, it has not painted a realistic picture of offensive swarm behavior. Still, some useful insights were drawn from the research. 

NPS researchers compared the intercept-shooter algorithm with the “Hungarian algorithm,” which assigns targets by minimizing the total distance between the pairs of defenders and their assigned attackers. In testing, the Hungarian algorithm was more successful when defenders were slower than attackers, while intercept-shooter had the upper hand when the defender drones were faster. 

The efficiency of each target assignment algorithm was also analyzed. In this analysis, the attacker-to-defender ratio was used as the independent variable. The dependent variable was an efficiency metric. This was created by normalizing the number of attackers against the number of defenders, then dividing that number by the length of the engagement, which was defined as time from first kill to the end of the simulation. The efficiency metric, too, favored the Hungarian algorithm for slower defenders and intercept-shooter for faster defenders.

The researchers’ concluded that the effectiveness of defensive swarms “increases linearly with weapon range, increases non-linearly with number of defenders, and increases exponentially with higher acceleration and velocity.” This suggests the highest return on investment for counter–unmanned aerial system (cUAS) drones can be earned by fielding faster platforms, followed by increasing the number of drones in the defensive swarm, and last, by improving effector range.

The growing variety of defensive drones includes fixed-wing ones, such as Lockheed Martin’s MORFIUS, Raytheon’s Coyote, and Anduril’s Roadrunner, as well as (often slower) rotary-wing platforms, such as Fortem Technologies’ DroneHunter. Some of these—e.g., Coyote and Roadrunner—are single-shot kinetic-kill weapons, destroying themselves in the intercept. DroneHunter and MORFIUS use ranged weapons. Thus, a heterogenous defensive swarm could rely on high-speed, high-g maneuvering Roadrunners to intercept the highest threat attackers in a swarm, while MORFIUS drones could disable comparatively low threat attackers.

It should be noted that researchers excluded data derived from a failure. An iteration counted as a failure if even a single drone reached the high-value unit. That makes sense for this research, but in the real world, while defensive swarms will undoubtedly be useful, they cannot be relied on exclusively to eliminate 100 percent of offensive drone swarms. Thus, the Marine Corps would be wise to stick to its guns, as it were.

Guns

A comprehensive cUAS approach would include drone swarms and air-defense guns. The latter would offer distinct advantages: Guns are widely produced, their ammunition is cheap, they are combat proven, and they can be paired with radar or electro-optical sensors for target search and fire control. Even though SPAAGs in isolation are an imperfect solution, Major General Borys Kremenetsky, the Ukrainian defense attaché to the United States, says Soviet-era ZSU-23-4s and Gepards nevertheless have been effective at destroying Russian unmanned aerial vehicles. 

How might the imagined engagement between Gepards and FPVs have played out if the Ukrainian vehicles had been assisted by a defensive drone swarm? 

A Raytheon Coyote kinetic-kill aerial drone costs $15,000. Air-defense guns typically destroy their targets with bursts of fire, but 7.62-mm NATO ammunition can be bought commercially for about $1 per round. This means that, for the cost of a single Coyote, some 10 to 15 thousand rounds of ammunition could
be purchased. That ammunition is also the effector for the MADIS’s Mk 2 M134 minigun.

Similarly, the Navy’s surface fleet would also be wise to bolster the gun armaments of surface combatants. A World War II–era Fletcher-class destroyer had five 5-inch guns and numerous 40-mm and 20-mm cannon. While Aegis destroyers and cruisers still have a 5-inch gun and Phalanx close-in weapon systems, their overall gun armament comes nowhere close to that of the Fletchers. Arleigh Burke–class destroyers carry two Mk 38 25-mm gun systems. Increasing that number would greatly improve their defensive capabilities against USVs. Bolting on cUAS systems derived from the MADIS Mk 2—which is already in full production—would also improve warships’ survivability against aerial drone swarms.

Individual drones have demonstrated their lethality on the modern battlefield. The upper bound of swarming has yet to be fully appreciated. Offensive and defensive swarm types and tactics will continue to iterate, with each gaining at least temporary advantages. Integrating defensive swarms into a combined-arms system with other time-tested defensive systems will enhance the effectiveness of each and make it more difficult for offensive swarms to achieve their objectives. The Navy and Marine Corps must continue to develop more capable defensive swarms and ground-based cUAS capabilities while it learns how best to integrate them.