Raytheon Ate My Terrier

I opened my news feed yesterday and saw that Raytheon had eaten my dog.

A couple months ago I published Vermin, Terriers and Transformers, arguing that the only sane response to cheap drone swarms is a layered immune system: cognitive kill layers to crash their software, and autonomous interceptor drones (terriers) to hunt whatever vermin shitdrone survives.

Yesterday, Raytheon announced that Coyote Block 3NK, a lovable, loitering, recoverable, non-kinetic counter-drone effector, just defeated multiple drone swarms in a U.S. Army demo. Impressive.

And yep, that’s my dog, my terrier. It loiters, it hunts swarms, it uses a non-kinetic kill mechanism that minimizes collateral damage, and it’s recoverable for redeployment. Ok, yes, that’s actually better than what I proposed: I described kamikazes, and Raytheon built something reusable. (::shakes fist::)

LOL even better - they’re selling it on the cost asymmetry argument. Tom Laliberty, president of Raytheon’s Land & Air Defense Systems, called it a “cost-effective defense for individual drones and swarms.” That’s PR-speak for “the math finally maths.” And the Army program it feeds into is called LIDS: Low, slow, small-unmanned aircraft Integrated Defeat System. That’s literally the vermin profile. My vermin profile.

I’d like my consulting fee now, please.

Unboxing the Coyote

The press release says “non-kinetic payload” and then shuts up. So we get no specifics, but the operational parameters tell us exactly what we need to know.

It’s recoverable, it defeats swarms, and it minimizes collateral damage.

No doubt: that’s a high-powered microwave (HPM) or directed RF energy burst. So ya fly close and dump enough RF energy to fry whatever cheap unshielded electronics are in the vicinity. The Coyote doesn’t need to understand the target. It just needs to get close, turn up the heat and cook it.

For a lot of targets, cooking works great. A $500 drone with exposed PCBs and no shielding? Dead in two clock cycles. The HPM burst doesn’t care what firmware it’s running or whether it has a command link at all. Brute force, dump enough energy into the airframe to upset something critical, and gravity is your friend.

Shielding Wut?

Small problemo. Electromagnetic shielding is a function of geometry.

Any opening in a conductive enclosure acts as a receiving antenna at frequencies where the opening’s dimensions approach a half-wavelength. A 10cm seam resonates at 1.5 GHz. A 3cm vent resonates at 5 GHz. Smaller openings, higher frequencies, harder to exploit.

Now think about a Group 1 shitdrone, which is under 20 pounds, often much less. Maybe 15cm across. You can injection-mold the entire housing in one piece using conductive-filled polymer (nickel-coated carbon fiber in resin, off the shelf from SABIC or RTP). That’s one piece, with no seams at all. The camera lens is 6mm and motor wire pass-throughs are 2-3mm. At those dimensions, you don’t get resonant coupling below 25-50 GHz. Generating kilowatts at those frequencies from a loitering platform is a very different engineering problem than generating kilowatts at 2 GHz. (I am not a drone builder and not an RF engineer, so I don’t know this in my bones, but I did enough reading, and my LLM said it confidently, so we’ll just roll with that statement.)

Total added cost for this shielding? Fifteen bucks. Maybe thirty. If you add ferrite chokes on the motor leads, and a fine conductive mesh over the camera aperture, you’ve got a Faraday-hardened shitdrone that shrugs off the Coyote’s kill mechanism like John Wick’s magic bulletproof suit jacket.

Now scale up. A Group 2 drone (21-55 pounds, meter-plus wingspan) can’t be molded in one piece. Multi-part housing means seams. Long seams. A 30cm mating surface resonates at 500 MHz, which is trivially easy to blast at high power. Bigger sensors mean bigger apertures. More electronics mean more heat, which means vents, which means more coupling paths. The thermal management problem and the shielding problem fight each other, and you can’t fix both.

Group 2+ drones stay vulnerable to HPM for a long time. The physics favors the defender.

Group 1 shitdrones resist the Coyote for pocket change.

That’s what I’m calling the shielding cliff: Coyote’s kill mechanism has a geometric shelf life inversely proportional to target size. The smaller and cheaper the attacker, the easier it is to harden against brute-force EM kill. The vermin are the ones that escape the Coyote.

(If you’re an adversary reading this, you already knew this.)

Missing Layers

Map Coyote against the layered defense stack from my original essay:

Detection — not Coyote’s job
Classification — not Coyote’s job
Cognitive Kill (RF protocol exploitation) — not what Coyote does
Cognitive Kill (visual/optical) — not what Coyote does
Terriers — Coyote Block 3NK lives here
Hard kinetic — last resort

Coyote is Layer 5 with a Layer 3 kill mechanism bolted on. A mobile EM-fry platform. Effective, but blunt.

The surgical layers are still unbuilt. RF protocol exploitation that punishes cheap firmware’s missing bounds checking. Adversarial optical attacks that crash vision-navigating autonomous drones. Those are the layers that address the real nightmare: the hardened Group 1 HOLO swarm with a $15 Faraday shell, no RF command link to exploit, navigating by sight.

That drone is the scenario that keeps me up at night. Coyote can’t touch it.

Commercial Gap

Oh one more thing. Coyote is Army procurement, meaning theater-of-war stuff.

But, but… Who protects the substation? The hospital? The water treatment plant?

Duke Energy still doesn’t have a counter-drone warfare division. The “Digital Dome as SaaS” market is still wide frickin’ open.

Raytheon validated half my architecture with a live demo. The other half (cognitive kill layers that target software rather than hardware, and a commercial delivery model that puts it in the hands of everyone who needs it) remains a blue ocean.

Somebody should build that. You know where to find me.

Colin Steele is a former F500VP, CTO, inveterate dilettante, and shockingly unpaid defense consultant who blogs at colinsteele.org. He has no clearance, no contracts, and no equity in Raytheon. His previous credentials include FDA-regulated algorithms that actually help people, which he leans on excessively. He has unjustifiably strong opinions about aperture coupling.