Planning Stealthy Backdoor Attacks in MDPs with Observation-Based Triggers
For security researchers studying vulnerabilities in stochastic control systems, this work introduces a novel attack paradigm that could inform defense mechanisms.
This paper proposes a method for planning stealthy backdoor attacks in MDPs where an attacker jointly optimizes a backdoor policy and an observation-based trigger using only a blackbox simulator. Experiments show the method achieves effective attacks while maintaining stealthiness.
This paper investigates backdoor attack planning in stochastic control systems modeled as Markov Decision Processes (MDPs). A backdoor attack involves an adversary deploying a policy that performs well in the original MDP to pass testing, but behaves maliciously at runtime when combined with a trigger that perturbs system dynamics. We consider a sophisticated attacker capable of jointly optimizing the backdoor policy and its trigger using only a blackbox simulator. During execution, the attacker has access only to partial observations of the system state and is restricted to introduce small perturbations to the system's transition dynamics. We formulate the attack planning problem as a constrained Markov game with an augmented state space and two players: Player 0 learns a backdoor policy that maximizes attack rewards when the trigger is active. However, when the trigger is inactive, the backdoor policy behaves near-optimally in the original MDP; Player 1 designs a finite-memory, observation-based trigger to activate the attack. We propose a switching gradient-based optimization algorithm to jointly solve for the backdoor policy and trigger. Experiments on a case study demonstrate the effectiveness of our method in achieving stealthy and successful backdoor attacks, and how the attack performance varies under different parameters related to the stealthiness of the backdoor attack.