The Army's answer to the swarm-drone problem may not be another interceptor missile or a bigger jammer. On September 2, 2026, Duke University announced it will lead a new six-year research effort, funded at $24 million by the Army Research Office, that aims to take enemy drones out of the fight through cyber and radio-frequency manipulation — hacking them, spoofing them, and in some cases wresting control of them away from their operators, all without a single kinetic round fired.
The project is called ICE-CEM, shorthand for "Inducing Cyber and Physical Effects in Unmanned Aerial Systems via Cyber-Electromagnetic Convergence." It runs from August 31, 2026 through August 30, 2032 under the Army's DEVCOM Army Research Office, and it is structured less like a single lab project and more like a distributed research consortium: Duke as the lead institution, six additional academic partners, and defense contractor Lockheed Martin rounding out the team.
A Research Hub With No Address
Unlike a traditional Army research center built around a physical facility, ICE-CEM exists as a network. As Hoodline described it, the program amounts to "a new kind of research hub — one that will not exist as a physical building at all." Instead, the work will be distributed across Duke's campus and the labs of its six partner universities, coordinated toward a common technical mandate rather than housed under one roof.
That mandate, according to Duke Today, is to develop fundamental tools and automated procedures capable of detecting, thwarting, and taking control of enemy unmanned aerial systems. Two constraints define the technical challenge, per CBS 17's Durham-area coverage: the tools must require minimal system interaction — meaning they should work with limited manual operator input in the field — and they must function with only limited prior knowledge of the specific drone being targeted. In practice, that means building counter-drone techniques that generalize across unfamiliar hardware and protocols rather than requiring a pre-built signature for every adversary system, a persistent problem for counter-UAS operators facing drones assembled from commercial parts or built to unfamiliar specs.
From Spoofing Radars to Seizing Control
ICE-CEM is led by Miroslav Pajic, a professor in Duke's Department of Electrical and Computer Engineering, alongside a bench of co-investigators spanning multiple technical disciplines: David R. Smith, Yiran Chen, Hai "Helen" Li, Tingjun Chen, and Neil Gong. That mix reflects the interdisciplinary nature of the problem — RF hardware, machine learning, network security, and control systems all have to work together for a system to reliably identify a hostile drone, disrupt its sensors or communications, and in some scenarios take command of its flight control.
The project builds directly on prior work out of Pajic's group known as "MadRadar," a 2024 research effort into radar-spoofing techniques. Where MadRadar explored how radar systems could be deceived by crafted signals into detecting fake targets or ignoring real obstacles, ICE-CEM extends that line of research toward a wartime application: manipulating the sensing and control loops of unmanned aircraft rather than merely defeating detection.
Why Cyber Over Kinetic
The pitch behind ICE-CEM is fundamentally about cost and scale. Kinetic counter-drone systems — whether missiles, guns, or directed-energy weapons — each carry a per-engagement cost and a logistics tail: ammunition, batteries, cooling systems, maintenance. A swarm of dozens or hundreds of cheap drones can overwhelm those systems simply by being numerous. Cyber and RF-based counter-UAS techniques, by contrast, promise to scale differently: a software-defined radio broadcasting a spoofed GPS signal or a crafted command packet doesn't run out of ammunition the way an interceptor does.
Seizing control of a hostile drone — rather than simply jamming or destroying it — also opens up possibilities that purely kinetic or purely disruptive electronic warfare cannot: an aircraft that has been hijacked mid-flight could in principle be redirected, forced to land intact, or exploited for intelligence before its handlers realize anything is wrong. None of the source reporting details specific techniques ICE-CEM will pursue to achieve that kind of control, but the stated goal — detect, disrupt, and take control, with minimal human intervention and minimal prior intelligence on the target system — sets a high technical bar distinct from simpler denial-based countermeasures already in wide military use.
Why It Matters
Cheap, mass-produced drones have become one of the defining threats on modern battlefields, and counter-UAS spending across the Pentagon has increasingly chased kinetic solutions — interceptor drones, directed-energy weapons, guns firing programmable airburst rounds — that all carry finite ammunition and significant unit costs. A six-year, multi-university research commitment aimed instead at software and radio-frequency exploitation signals that the Army sees cyber-electromagnetic techniques as a necessary complement, not just a niche tool, particularly against swarms where per-unit kinetic engagement costs don't scale.
The emphasis on tools that need "limited prior knowledge" of the target drone is also notable. Much of today's electronic warfare and counter-UAS tooling depends on having a known signature or protocol library for a given drone model — a brittle approach when adversaries are flying modified commercial quadcopters or novel designs built specifically to evade known countermeasures. Research aimed at generalized detection and exploitation, rather than system-specific playbooks, would address a real operational gap if it succeeds. That said, none of the available reporting specifies deliverables, milestones, or how quickly any resulting capability might transition from university labs to fielded systems — a six-year academic research timeline is a long way from a program of record, and Lockheed Martin's involvement as an industry partner suggests eventual transition is a goal, though no contract or acquisition pathway has been disclosed.
Given Duke's involvement of a defense-industry partner and the multi-institution structure, ICE-CEM is worth tracking as one indicator of where the Pentagon's counter-drone research dollars are heading over the back half of this decade — toward the electromagnetic spectrum and the software stack, not just the interceptor magazine.
