CKKS Homomorphic Encryption Enables Private Multi-Agent Control

Cloud-based coordination of multi-agent systems — such as drone swarms or robot teams — requires sharing sensitive state information with a central server, creating a fundamental conflict between coordination and privacy. Fully homomorphic encryption (FHE) theoretically resolves this by allowing computation on encrypted data, but its severe arithmetic constraints force a complete redesign of every control loop stage. In a new paper, Sai Sandeep Damera and colleagues from the University of Maryland present the first end-to-end encrypted control pipeline for multi-agent coordination that operates entirely on CKKS-encrypted data using only addition, multiplication, and cyclic rotations.

The pipeline reimagines all four core stages — sensing, state estimation, state propagation, and consensus control — to work under encryption. To overcome FHE's computational demands, the authors use steady-state Kalman gains (avoiding online matrix inversion) and apply graph Laplacians via a diagonal method that scales with the number of nonzero cyclic diagonals, supporting ring, torus, and complete-graph topologies in a unified framework. They also derive a periodic bootstrapping bound, treating CKKS bootstrapping as an impulsive disturbance, and show that steady-state error depends on bootstrapping precision and the closed-loop spectral radius — providing a direct design equation for the privacy-accuracy tradeoff. The pipeline is validated on a multi-agent formation control scenario, confirming stable closed-loop operation with bounded tracking error. The work has been submitted to IEEE for publication.