Secure Crowdsensing in 5G Internet of Vehicles: When Deep Reinforcement Learning Meets Blockchain

Abstract

Vehicular crowdsensing has been widely leveraged in the intelligent transportation systems to collect real-time traffic data. However, the data security and privacy preservation issues have not been well investigated for vehicular crowdsensing. This article proposes a blockchain-enabled vehicular crowdsensing system to protect user privacy and data safety in 5G Internet of Vehicles (IoV). In order to maximize the security and minimize the latency of blockchain, a deep reinforcement learning (DRL)-enabled algorithm is proposed to select proper active miners and transactions. Then, a two-sided matching-based algorithm is put forwarded to allocate the nonorthogonal multiple access subchannels to minimize the maximum uploading delay of all users to guarantee the freshness of messages. Extensive experimental results demonstrate the effectiveness of our system. Finally, system analysis demonstrates that our system can protect user privacy, guarantee data security as well as integrity, and defend against common attacks. With the development of information communication techniques, vehicular crowdsensing is widely leveraged in the Intelligent Transportation Systems (ITSs) to collect real-time traffic information (e.g., traffic accidents and jams). Recently, vehicles have begun to be regarded as the consumer electronics product, due to the advancements in the development of autonomous vehicles, and users in vehicular crowdsensing focus more and more on their privacy protection and data security. In order to guarantee the connectivity and communication of users participating in the vehicular crowdsensing, intelligent management of network traffic and offering services to vehicles has become a vital need, especially for the highly dynamic topology and untrusty communication environment in 5G Internet of Vehicles (IoV).1 However, the vehicular crowdsensing-based data collection and transmission in 5G IoVs face the following major challenges: