Differentially Private Formation Control: Privacy and Network Co-Design

TL;DR

This paper introduces a co-design framework for multi-agent systems that jointly optimizes network topology and differential privacy parameters to balance privacy protection with formation control performance.

Contribution

It presents a novel joint design approach for network topology and privacy levels in multi-agent formation control, incorporating differential privacy and tradeoff analysis.

Findings

Bounded steady-state error under differential privacy.

Tradeoff analysis between privacy, performance, and connectivity.

Successful simulation validation of the co-design framework.

Abstract

Privacy in multi-agent control is receiving increased attention, though often a networked system and privacy protections are designed separately, which can harm performance. Therefore, this paper presents a co-design framework for networks and private controllers, and we apply it to private formation control. Agents' state trajectories are protected using differential privacy, and we quantify its impact by bounding the steady-state error for private formations. Then, we analyze tradeoffs between privacy level, system performance, and connectedness of the network's communication topology. These tradeoffs are used to formulate a co-design optimization framework to jointly design agents' communication topology and their privacy levels. Simulation results illustrate the success of this framework.