Online Poisoning Attacks Against Data-Driven Predictive Control

TL;DR

This paper introduces a numerical method to craft poisoning attacks on data-driven predictive control systems, significantly increasing output tracking errors by injecting carefully designed perturbations into online data.

Contribution

It develops a novel approach using implicit differentiation to generate effective poisoning attacks on DPC, highlighting vulnerabilities in data-driven control methods.

Findings

Attacks cause an order of magnitude higher tracking error than random noise.

The method effectively exploits the solution map of trajectory optimization.

Numerical experiments validate the attack's impact.

Abstract

Data-driven predictive control (DPC) is a feedback control method for systems with unknown dynamics. It repeatedly optimizes a system's future trajectories based on past input-output data. We develop a numerical method that computes poisoning attacks that inject additive perturbations to the online output data to change the trajectories optimized by DPC. This method is based on implicitly differentiating the solution map of the trajectory optimization in DPC. We demonstrate that the resulting attacks can cause an output tracking error one order of magnitude higher than random perturbations in numerical experiments.