Why can't we shoot down drones?
It's a political, not a technical problem
TL;DR
For several weeks now, the news has been filled with reports of US Navy ships, aircraft and anti-aircraft systems from the Gulf of Aden to Jordan and Iraq being inundated by wave after wave of Iranian drones and cruise missiles. We’ve had over 150 attacks on our bases alone and scores more in the gulf of Aden in the past 4 months. Several of these have penetrated our defensive systems and last week tragically left us with three dead US Army reservists.
While for many Americans this may be the first time that the drone threat has been brought to their attention, for those of us in the defense sphere, we’ve been living this dream for a decade plus. The existing fielded solutions like the SM-2 missile or SLAMRAAM require a $1-2 million missile to shoot down a $2000 drone or $20,000 cruise missile. As we’ve seen in the Ukraine war, our missile batteries designed to counter ballistic missiles and high speed aircraft lack the magazine depth to sustainably defeat these threats over long wars of attrition.
It took the DoD over seven years after UAS with bombs (2012-2014) were first encountered while fighting ISIS to set up the Joint Counter UAS Office in 2021. This is significantly slower than it took for us to set up the Joint Counter IED Office (JIEDDO) that took a year to set up during the second gulf war to address the IED threat. Meanwhile our tech sector made steady progress towards fielding solutions that can effectively counter the proliferated threat of drones and drone swarms.
We are being inundated by drones because the enemy is on a 30 day acquisition response loop and we are still playing with an acquisition bureaucracy on a 5-10 year timeline.
The DoD acquisition and sustainment bureaucracy has been very slow to incorporate CUAS systems into the force structure. I have some experience with this first-hand from having founded a CUAS company five years ago, Epirus, which has the only system in government evaluation currently that can effectively defeat massive drone swarms. Indeed, The Center for the Study of the Drone counter-UAS (CUAS) in 2019 listed over 537 products in the space back then. These products are finding their way into units but oftentimes not through traditional acquisition means - workarounds like RIFS, OTA and direct commercial sales (8277) contracts are being used by Special Operations Command SOCOM and others to field this systems quicker.
The enemy is on a 30 day acquisition OODA loop and we are still dealing with an acquisition bureaucracy on a 5 year timeline. The Pentagon has a bias towards its existing contractor base, a slow moving acquisition structure and an emphasis on a training and doctrine tail that often makes perfect the enemy of good enough.
In this post, I’m going to dig into how we got here and the doctrinal, technical and political reasons why we are in the predicament of having the greatest army in the history of the world be at the mercy of $2000 drones.
Obsolete Doctrine for counter-air
First things first, let’s do a quick review of the kill chain:
As Figure 2 shows, back in the 80s and 90s when many of our air defense systems were first fielded, each piece of the kill chain was functionally decomposed quite cleanly and generally handled by separate systems. Find platforms (like the E-3 Sentry Airborne Early Warning (AEW) aircraft) would search far and wide to find threats, then hand them off to other platforms (such as the cancelled E-10A program) to fix on, ID and track them. These planes would then hand them off to 4th (F-15, F-16) and 5th (F-22, F-35) generation fighters to target who would then launch missiles, fire guns or employ effects like jamming to engage the targets. Finally, the AEW assets and engaging fighter assets would assess the target to determine if they killed the target or not.
Simple enough, right? The only problem is that systems security compartmentalization and a number of false starts on the acquisition side (like the E-10A above) led to this always being fragmented with many work arounds and bottlenecks. For instance, F-35s and F-22s couldn’t talk to each other until recently due to incompatible datalinks and F-22s couldn’t even talk to the AWACS aircraft because it lacked a Link 16 antenna. In another infamous example, Close Air Support (CAS) was far more effective from the 50 year old A-10 than the F-35 because the F-35 lacked an inexpensive ROVER terminal for coordinating with forward air observers (FAO) on the ground. Thankfully, these changes have slowly been integrated over time, but the bias towards performance over function and integration in acquisition lingers to this day.
In today’s systems, the kill chain is much more monolithic with single systems compressing most of the traditional F2T2EA kill chain. Fifth generation fighters like the F-35 are so loaded with sensors they can basically handle the whole kill chain themselves or between wingmen. Modern missiles and bombs have the ability to be launched untargeted and then engaged once they form a link with a weapons controller (known as network enabled weapons). Loitering munitions like the Israeli Harpy or the Switchblade drone from AeroVironment literally hang out and wait for the enemy to rear his head - then fix, track, target and engage all happen within the same asset. This has led some to suggest that it’s time to simplify the kill chain to something more like what’s shown in figure 3:
“One could argue the military is in a similar position to as it was 18 years ago if it forces today’s solutions into yesterday’s processes.”- Maj. Mike “Pako” Benitez, F-15E Weapons Systems Officer and War on the Rocks contributor
So bureaucratic inertia continues to force our troops to fight wars using yesterday’s tactics, which are proliferated culturally through TTP (tactics, training & procedures) and doctrine. In the case of CUAS systems for stopping enemy drones, these iterative exercises have literally been underway for years in some cases, with no end in sight. Meanwhile, the OODA loop (Observe-Orient-Decide-Act) for new technologies and tactics in Ukraine is measured in days, not years.
“The OODA loop for introducing new tactics or technologies to the drone battlefield in Ukraine is 30 days.” - Dr. Steven Blank, Stanford National Security Conference, November, 2023
Detecting and Killing Drones is technically challenging, but effective systems exist to solve it (and are fieldable today)
Now let’s shift our focus to the technical challenge or detecting and neutralizing enemy drones, where thanks to a decade plus of experimentation, hundreds of millions in private and public investment, and significant trial by fire in the Middle East and Ukraine over the last several years, has produced a litany of relatively effective technical solutions.
Detection and Tracking
Group 1 and Group 2 drones (generally classified as being below 50 pounds of payload) fly low, slow and make very little noise compared to airplanes and helicopters, and have a low radar, visual and acoustic signature. They are extremely maneuverable and start and stop frequently, frustrating tracking sensors and those dependent on motion. This makes them extremely hard to detect and defeat by conventional systems designed to shoot down faster, larger and higher flying aircraft. In fact, one of the programs for defeating UAS is actually called the mobile-low, slow, small integrated defeat system (M-LIDS) for exactly this reason.
Since drones typically fly up to 100 miles per hour (~160 kph) for the smaller ones and 200 mph or more (~320 kph) for the larger fixed wing variety, if you’d like at least about two minutes warning, ideally you would like to be able to detect them at a range of at least three miles (~5 km), if not further for smaller drones and six miles (10 km) or more for faster fixed wing drones. That way you at least have ample time to tell people to duck and cover in the event that you aren’t able to engage and defeat them before they strike.
Detection systems designed to find, fix, ID and track drones rely on a variety of modalities to accommodate the difficulties of tracking them. For instance, radar tracking of drones can be frustrated by the fact that many radars intentionally notch out low velocity targets to mitigate false alarms from birds, swaying trees or trucks passing by on the horizon that are not in the air. This is a feature, not a bug - but can prevent older systems like the TPQ-50 (which the M-LIDS system above uses) from detecting drones sometimes that can try to literally fly underneath the radar. Other radar systems such as those from RADA (now a subsidiary of DRS), Echodyne and others use a mode set (collection of radar waveforms, signal processing and procedures executable as a discrete function) specifically designed to look for these drones (not to mention are much cheaper) and are quickly becoming more ubiquitous as a result.
EO/IR systems, such as those used in Anduril’s Sentry Tower Systems or HighPoint Technologies Razorback System, use visual and IR cameras to detect and track drones. These systems have the advantage of very precise angular resolution and the ability to easily ID targets, but suffer from low search rates due to narrow fields of view, limited line of sight due to smoke, weather, daytime night time or other obscuration. They are also bad at estimating range and velocity in the absence of other sensors or active means (Lidar). Panoptical solutions in the future and distributed optics may mitigate some of this by employing multiple optics at once (sort of like your Tesla does) to maximize situational awareness, but the obscuration issues of weather, smoke and other factors will linger.
RF detection systems that look for emissions from the drones themselves, such as those employed by Whitefox and Hidden Level, provide 1D or 2D (elevation) line of bearing (or even “passive ranging” through multistatic/triangulation means), to the drone or operator based on downlink signals transmitted by the drone. These signals are detectable nearly instantaneously from significant range, but are easily masked by programming the drone to operate using AI/ML for terrain following navigation, GPS waypoint modes or limiting transmission to avoid detection. Some drones are even programmed to use 5G cell towers for connectivity making them hard to distinguish from cell phones.
All of these signals are usually fused into a single integrated operational picture, provided to the warfighter by C4ISR (Command-Communications-Control-Computers-Intelligence-Surveillance-Reconnaissance) systems such as Anduril’s Lattice system, RTX’s KuRFs or others where the warfighter can then see the integrated operational picture and decide what countermeasures or kinetic solutions can be employed to disrupt or kill incoming drones. Since all of these detection and tracking systems have natural limitations driven by physics, sensor fusion is the most effective way to ensure that nothing gets past your defense by tactically exploiting the blind spots of a particular sensing modality.
Disrupting and Killing
When it comes to killing drones, we have plenty of options too: the most intuitive solution favored by traditional warfighters are kinetic: with machine guns like BAE’s AMPV solution, Raytheon’s Coyote Drone, or rifle augmentation like the Israel’s smart shooter system being just a few of the solutions currently employed or in the process of being fielded.
We also have reusable collision drones such as Anduril’s Anvil system or even crazier ideas like the Dutch using trained falcons to snatch small drones. All of these systems, while conventional and easier to provide assessment on effectiveness, are easily overwhelmed by large drone attacks en masse due to magazine, range and rate of fire issues. However, they do make a hell of a bang!
It’s important to underscore just how many drones we may expect to see in a drone swarm attack. Here’s a scene becoming all too familiar to us of a drone sky show in Dubai in lieu of fireworks which employed 1000+ synchronized drones. It doesn’t take much of a stretch of the imagination to imagine these critters simultaneously and synchronously attacking a forward operating base and overwhelming our kinetic solutions:
Electronic Warfare solutions such as communications jammers, GPS jammers and cyber attack systems which actually try to attack the drones communications protocols to trigger effects, like the Ninja system have also been around for a decade or more, with limited effectiveness. These systems are frustrated by the same factors that frustrate RF detection systems as well as reprogrammable waveforms, messaging protocols, the “near-far” problem that limits effectiveness against receivers closer to the transmitter, boosted and spread spectrum communication systems meant to defeat jamming. They also are of limited effectiveness against drone swarms like shown above which often operate on different codes, waveforms and frequencies and jammers come with their own risks of RF fratricide (jamming your own drones and other systems) that can limit effectiveness.
This leads me to my favorite category of solutions: Directed Energy weapons. These come in two flavors: High Power Microwave (HPM) such as Epirus’s Leonidas system shown in the video below and Lasers, such as RTX’s palletized 10-kw solution.
Lasers are pretty intuitive: heat burns a hole in the drone or burns out its electronics. HPM works by overwhelming the electronics of the drone with microwave energy, forcing it’s electronics to malfunction or fail and letting gravity do the rest of the job.
Lasers generally have the advantage of range and also the ability to do visible and irreversible damage to a target. However, they suffer from rate of fire limitations, high maintenance costs and have significantly degraded performance due to weather and other obscuration.
HPM on the other hand is the most effective solution to countering swarms- in general, if a target is in the beam when it’s activated, it dies. These systems create forcefields of energy that are rapidly redirected to new vectors as needed.
High Power Microwave (HPM) solutions are the most effective solution to countering swarms- if a target is in the beam, it dies.
Both types of directed energy systems are in varying levels of trials right now and could be coming to the battlefield this year to protect our soldiers, however bureaucratic and budget hurdles remain from an acquisition force more fearful of fielding an imperfect solution than leaving our forces unprotected.
The political reason - why it’s really not happening
The economics of the modern battlefield have shifted and the acquisition landscape is lined with the bodies of the fallen. During the cold war and up until the surge in the second Iraq war, we thought we were going to fight modern wars with increasingly expensive, Battlestar Galactica systems like the XM2001 Crusader Artillery system, elaborate ISR systems to target them like Space Radar and JSTARS recap, protected from above by F-22 stealth fighters all netted using the TSAT communications system and Future Combat Systems network backbone. What do all the systems I cited have in common? They were all cancelled or pared back - sometimes after spending tens of billions of dollars - as the modern battlefield rendered them too expensive, obsolete or irrelevant.
As always happens, the new enemy gets a vote and decided to sidestep our technical advantage rather than just run straight into it like the last enemy did. The rise in asymmetric warfare tactics and adversaries we saw in the last couple of decades has filled the battlefield with cheap, decentralized systems like drones (UAS), proliferated commercial communications and satcom systems, open source intel (OSINT) and improvised explosive devices (IEDs) which has leveled the playing field for actors lacking a military budget equal to the entire GDP of all but 20 countries.
Our acquisition system has failed to speed up to confront this threat. It was built up around a 10 year fielding cycle during the cold war - embracing the DoD acquisition framework memorialized in the Goldwater-Nichols Act of 1986 and development tools like the Cohen-Clinger DoD Architectural Framework - where oftentimes the frameworks for thinking about the problem become pursuits onto themselves where the cartography becomes so extensive it covers the terrain its meant to depict (thank you Juan Luis Borges). The lumbering bureaucracy created by these reforms inside the Pentagon has been slow to reform to confront the new reality of an enemy with an acquisition OODA loop of 30 days as mentioned above. Rather than streamlining and removing bureaucratic barriers, much of the defense “innovation theatre” of the past decade has only proliferated lots of new organizations that have their own prurient interests not aligned with putting new systems in the hands of the warfighter. A recent article from military and capital hill veteran John Noonan underscores this point:
“Despite the urgent need for new UAV defenses, the Pentagon is too bottlenecked, too paralyzed by its own internal dynamics, to get these new systems into the field. Dozens of new administration offices and joint cells and bureaucratic edifices have been erected to get at the problem, but despite the best-laid plans of mice and men, the problem persists. It is another MRAP moment for the Defense Department. Troops are in contact, dying and suffering terrible injury, but the Pentagon has struggled to expedite the solution.” - John Noonan
Recommendations
We need to streamline the acquisition bureaucracy and make it more tightly aligned with the warfighter, particularly to make it more aligned with the warfighter. Some recommendations based on what has worked successfully in the war in Ukraine which has caused the Ukrainian drone industry to grow exponentially from seven to over 200 manufacturers in the 2 years of the war (if you are in the acquisition bureaucracy and are saying “we’re already trying to do this!” please feel free to yell at me about it in the comments):
a) Remove the barrier between the warfighter and the war factory: create integrated warfighter and developer teams providing direct communication between the end user (the warfighter) and development cells to provide rapid iteration against the new threats.
b) Create competition amongst procurement authorities: charter multiple groups within DoD to create and field solutions and judge them by time to market and scalability as well as effectiveness. A 70% solution available today may beat a 90% solution five years from now - and can iterate to catch the 90% solution over time.
c) “Discord forums for defense”: Create more communities of practice (at the appropriate security levels) that encourage the sharing of ideas and battlefield feedback to battle system developers in days, rather than stove-piping innovation behind the traditional “working group” models that restricted feedback to development programs and took years for the field to talk to the factory. Encourage outside experts to get cleared into these forums and donate their time to introduce outside thinknig
d) Use executive order authority to waive some aspects of EAR and ITAR: both ITAR and EAR have aspects that are preventing us from talking to our allies about key technologies and fighting while people are dying. We should demand congress take action to amend these obsolete laws which have created embedded bureaucracies that operate until irrational rules that stifle innovation. However, this will take time and in the interim the executive may have authority to suspend or bypass aspects of these laws to allow better technical collaboration with our allies.
e) Make time to deploy a hard requirement: during WWII the US famously designed, built and tested the P-51 in 90 days. We need to restore that sense of urgency. Rather than letting systems sit in the lab forever as we make them perfect, embrace good enough and realize that no perfect system is perfect after the enemy makes first contact with it and waiting forever it becomes irrelevant. Use contract incentives for early delivery and reward contracts who embrace continuous integration/continuous development approaches to field systems faster.
While I realize much of this will be anathema to the mission assurance and industrial security crowd, dead Americans on the battlefield due to us not fielding the CUAS systems we have today and iterating on them more rapidly should sober us to the reality that we aren’t moving fast enough. We must take practical steps to speed up our acquisition OODA loop to confront the new drone - and broader battlefield - reality.