Understanding 'Drone Swarming'
Why and how swarming can also be a vulnerability.
Whenever someone mentions the future of Unmanned Aerial Systems (UAS), they never fail to mention the necessity of drone swarms. They sometimes also mention the spectacular shows that can be performed by large numbers of drones, as if you were looking at some fireworks show. Drone swarms are said to be the future of warfare. Yet, as I will argue in this post, we often focus on the swarm or flocking of many drones, instead of the actual advantage this behaviour supposedly provides.1
I asked ChatGPT to give me a definition of drone swarming and it came up with a pretty good and complete one: Drone swarming refers to the coordinated use of multiple unmanned aerial systems, or drones, that operate together autonomously to achieve a common objective. In a swarm, drones communicate with each other, adapt to changing environments, and perform complex tasks as a unified group, mimicking natural swarming behaviour seen in animals like birds or insects.
The main difference with the current deployment of drones is the sheer numbers, or what we like to call in the armed forces: mass. There are a couple of advantages in having large numbers. First of all, with greater numbers you can overwhelm the enemy target acquisitioning, since it has too many targets to identify, track and target. A second advantage is that you can overwhelm the enemy by being faster than that the enemy weapon system is capable of reloading. This tactic is however not intended to deplete the entire ammunition supply of the weapon system, unless the system has low or expensive amounts of ammunition like the PATRIOT missile system. A third advantage is that large numbers can assist each other in target detection. One of the current ideas that is being worked on involves using a sensor mesh network. Harsberger explains it like this:
The sensor mesh concept uses simultaneous observations from multiple independent platforms to identify the target. "Mesh" refers to a network of connected, relatively simple sensors that can take advantage of mass—large numbers of vehicles and sensors working in concert—to make up for the relatively poor performance of each individual sensor. - But a mesh requires many, many sensors. The simulation shown here (video below) reflects the application of several thousand unmanned aerial vehicles (UAVs) to a Taiwan Strait scenario. The vehicles use simple rules to move into the area and toward detected ships, providing targeting solutions for weapons that can be fired from various platforms. Many UAVs are lost, but the sheer numbers in the mesh overwhelm the defensive response.2
This short video also gives you an impression of what it means to ‘overwhelm’ the enemy by large numbers. Also note that this means that many drones are being destroyed. This is why it is preferable to make these drones expendable. The cheaper and simpler they are, the more likely you are able to create large numbers and actually expend them. This is what I mean when I wrote about the future of drones in the Deadly Triad: many, autonomous and unmanned.
A fourth advantage is that large numbers can allow you to diversify the functions of these drones. Whereas some can do surveillance and identify targets, others can have other functions like striking the target with different weapon systems.
The most promising advantage is what I have earlier written about, swarm intelligence. This means that by simple rules within a swarm, collective intelligence can emerge that is capable of solving problems, without a single decision-maker directing this effort. We could combine these lessons with the sensory mesh network explained above.
The problem with ‘swarming’.
Current drones operating in swarms use simple rules that determine their behaviour. These rules make sure that large groups of drones can maintain a minimum distance to each other (ie. separate), have the same heading and speed (ie. align) and remain in a coherent group (ie. cohere).3 These rules will make sure that you see an actual swarm, as in a flock. However, precisely this mimicking of flocks of birds seems unnecessary when you look at the advantages mentioned above. They are nice to look at and beautiful for a show. Why create an easy target for the adversary? Swarms are predictable too some degree because of these rules and therefore easier to identify, track and target. Note that the individual drone behaviour might not be easy to predict, as is the case with birds, but their proximity makes them an easy target. They need to converge on the same target at (approximately) the same time in order to overwhelm, but from as many directions as possible, preferably not as a swarm. The short video above concerning a sensor mesh network also does not display a swarm. It may seem as such at a first glance because you are seeing large numbers converge on the same targets. The trick is that they need some proximity to communicate and coordinate, but that does not necessarily neccesitate swarming in the sense of ‘staying together as a group’.
When we now return to the definition of Drone Swarming, than actually the last part of the definition “mimicking natural swarming behaviour seen in animals like birds or insects” is unnecessary for achieving the desired effects. Moreover, bees do not flock, they do work towards the same goal and are able to converge on the same target such as a food source (see swarm intelligence).
To conclude, large numbers of drones can give you a clear advantage. However, we should not be misled by drone swarms that move in a beautifully synchronized pattern or flock. Instead, we should try to create large numbers of drones that at a first glance seem to be chaotic but which are in fact cooperation in target detection, acquisitioning and destruction. Because one thing is clear, if we can write an algorithm that can make a groups of drones swarm in unison, as in a flock, than our enemies can also develop an algorithm that can detect the patterns of these swarms and take them out.