Time-Driven and Privacy-Preserving Navigation Model for Vehicle-to-Vehicle Communication Systems

• Published in: IEEE transactions on vehicular technology Vol. 72; no. 7; pp. 8459 - 8470
• Main Authors: Zhu, Congcong, Cheng, Zishuo, Ye, Dayong, Hussain, Farookh Khadeer, Zhu, Tianqing, Zhou, Wanlei
• Format: Journal Article
• Language: English
• Published: New York IEEE 01.07.2023; The Institute of Electrical and Electronics Engineers, Inc. (IEEE)
• Subjects: Communications systems; Decentralized navigation; Deep learning; deep reinforcement learning; Driving conditions; Hidden Markov models; Internet of Vehicles; multi-vehicle system; Multiagent systems; Navigation; Navigation systems; Privacy; Real time; Real-time systems; Reinforcement learning; System effectiveness; Traffic congestion; Traffic models; Vehicle dynamics
• ISSN: 0018-9545; 1939-9359
• DOI: 10.1109/TVT.2023.3248613

Effective time-driven navigation is an operative way to alleviate traffic congestion, which is also a challenging problem in the Internet of Vehicles context. Most existing centralized navigation systems often cannot react promptly to real-time local traffic situations, while most existing distributed navigation systems are vulnerable to privacy attacks. To overcome these drawbacks, we propose a learning model that provides a provable guarantee of vehicles' privacy while still enabling efficient navigation under real-time traffic conditions. The proposed model adopts a novel multi-agent system with customized differentially private mechanisms. To verify the effectiveness and stability of our approach, we implement the proposed method on CARLA, which is an autonomous driving simulator. In four experimental tasks with varying parameters, we demonstrate fully that our proposed method outperforms other benchmarks.