Adversarial Attacks Against Deep Reinforcement Learning Framework in Internet of Vehicles

TL;DR

This paper investigates the impact of Sybil-based adversarial attacks on deep reinforcement learning in Internet of Vehicles, showing significant performance degradation in service delay and resource management under attack scenarios.

Contribution

It provides an experimental analysis of Sybil attacks on DRL-based service placement in IoV using real vehicle data, highlighting vulnerabilities and effects.

Findings

Sybil attacks significantly increase service delay.

Resource congestion worsens under attack scenarios.

Performance degradation correlates with the proportion of attacked vehicles.

Abstract

Machine learning (ML) has made incredible impacts and transformations in a wide range of vehicular applications. As the use of ML in Internet of Vehicles (IoV) continues to advance, adversarial threats and their impact have become an important subject of research worth exploring. In this paper, we focus on Sybil-based adversarial threats against a deep reinforcement learning (DRL)-assisted IoV framework and more specifically, DRL-based dynamic service placement in IoV. We carry out an experimental study with real vehicle trajectories to analyze the impact on service delay and resource congestion under different attack scenarios for the DRL-based dynamic service placement application. We further investigate the impact of the proportion of Sybil-attacked vehicles in the network. The results demonstrate that the performance is significantly affected by Sybil-based data poisoning attacks when compared to adversary-free healthy network scenario.