Trust and Reputation Management for Data Trading in Vehicular Edge Computing: A DRL-Based Approach
Elham Mohammadzadeh Mianji, Gabriel‐Miro Muntean, Irina Tal · 2024
The internet of vehicles (IoV) has become a promising technology for enhancing road safety and driving experience by sharing data collected by autonomous vehicles (AVs). In this context, data trading has emerged as a promising approach, where AVs can become sellers and sell their data to infrastructures to increase their benefits following participation in data trading. However, this data may contain sensitive information, making it crucial to ensure the security and trustworthiness of the data and AVs. To address these challenges, we propose a novel reputation management approach for data trading that leverages deep reinforcement learning (DRL) in vehicular edge computing (VEC). Our approach, named dynamic selection of trusted sellers using deep deterministic policy gradient (DSTSDDPG), dynamically optimizes a credibility score threshold. This allows for the adaptive selection of the most trusted AVs and transactions for data trading. The efficient selection of most trusted candidates among sellers, enhances the credibility of the network and minimizes the adverse effects of malicious AVs during data trading. Our proposed approach is robust, scalable, and can be used to promote trustworthiness and secure data trading in VEC. The simulation outcomes highlight the convergence and efficiency of our proposed algorithm, demonstrating its capability to identify trusted transactions from available options, thereby ensuring trustworthiness in data trading.