An Approach to Detecting Malicious Information Attacks for Platoon Safety

Abstract

Malicious attacks reduce the benefits of cooperative adaptive cruise control (CACC) such as safety, driving convenience, traffic flow, and fuel efficiency, by destabilizing the stability. To reinforce the resiliency of a CACC based platoon of connected and automated vehicles (CAVs), this work investigates a detection method for malicious information attacks in the platoon. In this work, we propose an attack detection method, called LMID (long short-term memory (LSTM) based malicious information detection). We consider two attack models: correlated attacks and non-correlated attacks. In our attack scenarios, one of the platoon members attacks the platoon using the attack models. Using PLEXE, a well-known platoon simulator, we develop a simulation framework to implement attack scenarios and evaluate the proposed detection method. LMID is trained depending on the length of input data and analyzed under various scenarios regarding platoon trajectories, attack types, and an emergency brake case. We have shown that without fast detection of such attacks, crashes may happen within a platoon. The simulation results demonstrate that LMID detects the malicious information attacks with higher than 96% accuracy and the attacks are detected very quickly. The performance evaluation indicates the superiority of the proposed detection method under various circumstances.

Recuperação do Serviço de Pelotão de Veículos Autônomos e Conectados Baseado em Modelos Virtuais Contra Ataques na Rede e Sensores

“…Este framework emprega os conceitos de Teoria dos Jogos e de equilíbrio de Nash na reconfigurac ¸ão do sistema de controle do veículo para mitigar o ataque de rede. Em [Ko and Son 2021], um método de detecc ¸ão de ataque denominado LMID (Long short-term memory (LSTM) based Malicious Information Detection baseia-se nas informac ¸ões maliciosas transmitidas na rede provenientes dos veículos membros do próprio pelotão. Para a detecc ¸ão, eles definiram ataques correlacionados e não-correlacionados e treinaram um modelo de rede neural profunda utilizado em cada veículo no teste.…”

PoF: Proof-of-Following for Vehicle Platoons

“…However, the complex integration of multi-modal physical sensing, computation, and communication creates a particularly challenging environment to safeguard. The safety of the platoon relies on the veracity of the V2V messages exchanged between platoon members, as falsified messages about acceleration, location, and velocity can lead to life-threatening accidents, damage to high-value cargo, and monetary loss [22], [49]. The key security questions for a platooning application are: (a) who is authorized to participate in the platoon and how is the identity of the platoon members verified?…”