S Lab
[Sibin Mohan]
Systems Security Research Group at GWU and University of Illinois
:
Swarm Security
Team Members
Collaborators
Overview
An autonomous swarm is a team of agents that each act on local information yet must coordinate toward a single mission. Its strength and its fragility come from the same source, since the interconnection that turns one agent’s decision into collective behavior is also what lets a single disturbance spread instead of staying where it began. In settings where resiliency is critical, from disaster response to distributed defense, that exposure is the whole problem. The threat that matters is rarely a random fault or a noisy sensor; it is a strategic adversary that captures, spoofs, or persuades a few of the agents, turns the swarm’s own coupling against it, and behaves like a teammate in order to break the mission. Underneath the security question is a question of scale: a swarm acts at two levels at once. Each agent makes its own local choices, including, once compromised, the choice to act against the team; whether those choices still add up to a working mission is settled only across the team as a whole. The line we care about runs between those levels: how much corruption a swarm can absorb before the mission fails, and when a local harm stays contained instead of cascading through a sparsely connected team. Finding that line is the heart of our work.
Publications
(TBD)