It's the Network Stupid!

The 21st century is a story of decentralization and the network is the new supply lanes

Finally getting back into the swing of things here and while I’ll be taking some time off for Thanksgiving I’m hoping to get back to this on a more regular basis.

TL;DR

The uber theme of the 21st century so far, summarized by Martin Gurri’s The Revolt of the Public, has been the disruption of traditional top down institutions by decentralized bottoms up networks with diffuse and dynamic control. The battlefields of the 21st century have been no different: the massive combined arms forces with centralized command & control of the last century gave way to diffuse networks of non-nation actors like Al-Qaeda, ISIS or even Anonymous.

Recently, even conflict between nation-states has taken on a flavor of this, with Russia and Ukraine utilizing asymmetric systems like remotely piloted (FPV) drones and GPS jammers to offset the massive capital advantage of expensive GPS guided artillery and heavy armor.

Despite first being fielded at scale in the late 90s in the form of Predator surveillance drones in the Bosnia war, autonomous drone technology is still nascent. The main catalyst for this tech has been the Cambrian explosion of commercial technology for the software and hardware components enabled by cheap (mostly Chinese) global supply chains.

In 2013, DJI introduced the Phantom which quickly became the dominant affordable consumer drone as FPV drones were introduced by ImmersionRC, Fat Shark and others around the same time. The combination of cheap consumer scale drones combined with FPV control has created a weapon of war which has come to define the largest war in Europe in 80 years. Suddenly the same precision strike capabilities that only great nation states could afford to build in small numbers due to expense were widely available to any hobbyist (or terrorist) with an Amazon account.

As the conflict has dragged on, the FPV systems have become more and more hands off and automated. The introduction of terminal guidance systems incorporating elements of visual AI algorithms by Auterion, FoxFour and others was driven by the need to simultaneously defeat jamming and improve precision, particularly against moving targets. The massive scale of design iteration by two software savvy adversaries has driven a pace of product evolution that is reaching the limits of what is possible using 1v1 tactics and operations. This brings us to the swarm and the next stage of autonomous warfare.

First a word on jamming

Disrupting/jamming the command datalink and video downlink is the most popular non-kinetic way to defeat drones - resulting in hundreds of CUAS jammers proliferating on the market today. These range from the elaborate (like this Titan system from BlueHalo/now Aerovironment) which employ sophisticated AI/ML algorithms to match narrow band jamming to particular frequencies, to simpler barrage noise jammers such as those made by Unwave in Ukraine (pictured on the right). In jamming systems, the more precisely you can match your jamming waveform to the one you are trying to defeat, the more effective it will be. Narrow band jamming tries to match this as precisely as possible, but must be tuned to the band of interest and follow the signal as it hops around; barrage noise jamming in contrast floods the whole operating frequency range of the signal of interest with noise, resulting in a lower jam to signal (J/S) ratio but ensuring that at least some degradation of performance is achieved.

Figure 1: Titan CUAS jammer from BlueHalo (left); Unwave Barrage Noise Jammers made in Ukraine (right).

While US jamming systems such as Titan can run tens to even hundreds of thousands each, the Barrage noise jammers on the right retail for low thousands of dollars. While in theory the system on the left should work better, in practice it’s sophisticated response systems designed to tune the energy so precisely and optimally are fragile when confronted with rapidly changing waveforms, tactics and other real world issues commonly seen on the battlefield in Ukraine. They also require more operator training to use, which isn’t always realistic for 60 day conscripts. So the Ukrainians have mostly stopped using these US in systems in favor of simpler domestic solutions - primarily barrage noise jammers that are cheap and expendable.

Cheap and unsophisticated electronics favoring simplicity and brute force over exquisite and performance optimization wins on the battlefield in Ukraine

Unwave presently builds up to 2,500 jammers per month, making them the largest EW company in the world based on pure unit volume.

The network enables more elaborate drone tactics and maneuver

Why is this important? A key distinguishing feature of kamikaze drones, unlike the last generation of missiles, is that they are not fire and forget. This is an asset in the sense that they are dynamically retargetable, can be waived off from a strike at the last minute and can even conceivably return if not exercised).

However, inherent in this is that the user must stay engaged and connected to the drone to precisely command it. This creates a liability as shown in this clip from Starship Troopers (perhaps the most famous line in all of Robert Heinlein’s books by the way):

Attention and engagement from the operator in a 1v1 sense is integral to drone warfare today in a way that it wasn’t on past battlefields, where soldiers and units separated from command & control could still continue to fight on, albeit without the benefit of collective effectiveness.

Architectural robustness and new tactics bas the fight rages continue to make drones on both sides more robust against jamming. In addition to traditional anti-jam improvements to comm links like frequency hopping, spread spectrum, better filters and antennas and other techniques to buy the data link more robustness, we are also seeing the introduction of more automation to relieve the burden on the datalink.

The first and easiest improvement was the introduction of waypoint modes whereby drones can fly on a pre-set route without the benefit of ground communication. However, GPS jamming frustrates waypoint modes and prevents them from being used. This was then countered with the introduction of terrain following navigation (TFN) which enabled the drone to find its way by looking at terrain, using techniques similar to what was done in pre-GPS cruise missiles like the Tomahawk.

Automated terminal guidance systems were recently added to enable the drone to hit its target on final approach without the need for a user command link. Since jamming signals grow in strength with the square of the distance to the target, these help ensure that the drone will fly to it’s target even if it’s target is employing CUAS jammers itself.

A new capability we’ve started to see in the last twelve months or so is the addition of relay drones, such as the Baba Yaga and the Dovbush Kamikaze drone carrier (shown in video below), which are able to serve as both a signal booster for the actual attack drones and also enabling the comm data link to have line of sight over the jamming. The higher angles of attack these achieve also make the kamikaze drones even harder to destroy.

US suppliers of radios are moving a step beyond this with the introduction of mesh or so called Mobile Ad Hoc Networks. Multiple drones or ground nodes operating cooperatively can bypass regions of jamming by connecting in directions and through nodes that are less impacted. Silvus Mobile Networks with their MN-MIMO technology is considered the Cadillac of this space, with Cadillac prices ($10-$15k per radio). Persistent Systems is another MaNET radio worth mentioning in the same vein, though the addition of ground towers to their architecture adds an unnecessary vulnerability even if extending range to 75 miles or more. Until the radios themselves start to cost less than the rest of the drone (many large Ukrainian attack drones cost $15k each, smaller ones as little as $1000), it’s not clear to me these approaches will see widespread employment, at least not in Ukraine.

“What’s flying today is almost all Ukrainian. Also, there are still a lot of Chinese drones,”

US drones have really not been well received in Ukraine since the first year of the war, with the natives preferring their own systems as more battle tested and easier to adapt. There is also a prevalent feeling that many American drone companies are just there to check a marketing box and aren’t really there to help them win the war.

How is swarming different

As I previously mentioned, up until this point, drone warfare has been primarily 1 on 1. This is a form of combat organization called melee (left of Figure 2 below) and is the default of uncoordinated forces. The tunnel visioned approach of FPV operators executing a narrow mission to seek and destroy individual vehicles lacks the coordination and synchronization to do much else.

Figure 2: A comparison of the four canonical forms of military organization: melee, mass, maneuver and swarm (from Swarming & The Future of Conflict, Paul Arquilla, Don Ronfeldt )

However, recent reports from the Kursk offensive indicate that we may be seeing the beginnings of more complex organization and tactics occurring. During the offensive, it was reported to me that the Ukrainians took advantage of a lack of electronic warfare systems in the front on the Russian side (most had been deployed to Karkhiv, Donetsk or elsewhere) to put up a huge wall of kamikaze drones. These drones stayed in the air like loitering munitions and were used as a mass formation that was able to push forward and make room for advancing combined Ukrainian forces that were able to seize and hold territory (we’ve yet to build a drone that can replace infantry and armored vehicles for this). This proved enormously effective and was integral to them penetrating deep into Russian territory within the Kursk Oblast (state) as shown in Figure 3.

Figure 3: Ukrainians have seized thousands of square kilometers of territory in Russia’s Kursk Oblast aided by mass drone kamikaze strikes acting as a vanguard and cover operation.

These operations, while enormously effective, represent only a fraction of the maneuver potential of drones, in particular as we uncouple their precise command and control from their human operators. With so many systems dependent on remote operation, the network will become the key to sustained operations: without a datalink the drone is mostly just an expensive paperweight right now.

Meanwhile, the product people are busy figuring out how to launch drones at scale, as you can see in this (rather frightening) video from China:

What is clear is that regardless of what level of autonomy is required, these drone swarms will require a great deal of in-situ cooperation. Nearly 20 years ago, Paul Arquilla and Don Ronfeldt at Rand wrote a visionary paper on “Swarming & The Future of Conflict” which really outlined the key characteristics of the drone swarms we will soon be seeing as the next evolution of battlefield maneuver.

“The historical insight that information flows (or their disruption) provide a key, both to understanding and practicing swarming, should encourage U.S. military leaders to view information operations as a central element in the development of future military doctrine” - Paul Arquilla, Don Ronfeldt from Swarming & The Future of Conflict

Arquilla and Ronfeldt saw the key enabler of swarms as information systems and specifically networks: “…the rise of advanced information operations will bring swarming to the fore, establishing a new pattern in conflict.” These insights were ahead of their time and we are just now seeing them play out.

The Denied, Degraded, Intermittent, Low Bandwidth (DDIL) environment

A lot of proposals from DIU and others have been coming out lately asking for drones that can operate in Degraded, Disconnected, Intermittent and Low-bandwidth (DDIL) environments. This is simply the latest acronym/buzzword that the DoD has come up with to talk about a comms denied and GPS denied environment, something they’ve been thinking about for quite a while. However, what is perhaps different this time is the implicit assumption that the systems will have to operate or fight through the electronic interference rather than simply having a better network, more jam resistant GPS or radio to just overcome it.

Victory in future battles will depend more on who has the best “networks” than on who has the strongest “platforms.” - Arquilla & Ronfeldt

The important insight to draw from this is that in a world of wireless networks, the spectrum characteristics of each space on the grid is essential to fingerprint, characterize and monitor. Based on conversations with front line forces I’ve had in Ukraine, very elaborate electronic topography is being mapped and lanes free of friendly datalink and PNT jamming (yes the Ukrainians also jam GPS) are maintained so that their own drones can pass into the no man’s land unimpeded. These may change dynamically based on mission planning, times of day or stages of an operation.

The need to “fingerprint” the battlefield and find, fix and eventually finish hundreds of emitters will be essential to creating lanes of communication to command and monitor swarms of autonomous systems.

In traditional electronic warfare, we spent a lot of time leading up to the conflict in Intelligence Preparation of the Battlefield (IPB) trying to discern the Electronic Order of Battle (EOB) the enemy would follow so we could defeat their air defense systems. In this modern electronic war underway in Ukraine, we are seeing this taken to a level of complexity orders of magnitude beyond layups of maybe 200 radars along the Fulda Gap in Germany (the canonical battlefield for the clash of the Superpowers) or the Taiwan strait. The need to “fingerprint” the battlefield and find, fix and eventually finish hundreds of emitters - in real time - will be essential to creating lanes of communication to autonomous systems.

Putting this all together

The complexity of the imminent introduction of the swarm will make this all the more essential and put pressure on the resolution and latency of the systems that do this. It’s almost impossible to imagine that large exquisite systems like Compass Call and Rivet Joint will get anywhere near the fight to be able to do this given the density and sophistication of modern air defenses I discussed in the previous post.

EA-37B Compass Call, the Air Force’s newest Electronic Warfare Aircraft

If we want to employ swarms rather than people at scale to defeat our adversaries, it will also be incumbent on us to develop new ways to fingerprint the battlefield and build a new “electronic wall” that can hold and secure spectrum and electronic topography in the same way we hold territory today. The Suppression and Destruction of Enemy Emitters and Networks (SEEN/DEEN) I mentioned in the previous article must become part of our doctrine if we are to succeed on the battlefield in the swarm era.

I’ll leave you with these final words of wisdom from Arquilla & Ronfeldt, who truly were decades ahead of their time: “Today, the key form of organization on the rise is the network, especially the all-channel network. The new information technologies render an ability to connect and coordinate the actions of widely distributed “nodes” in almost unprecedented ways. Whoever masters this form will accrue advantages of a substantial nature. Yet, networking alone is not enough; just organizing into a network is no guarantee of success. There must also be principles and practices—a doctrine—to guide what a networked force should do and how it should behave.”