计算机科学与探索 ›› 2024, Vol. 18 ›› Issue (7): 1923-1934.DOI: 10.3778/j.issn.1673-9418.2307081

• 网络·安全 • 上一篇    

车联网区块链分布式车对车计算卸载方法研究

孟珍,任冠宇,万剑雄,李雷孝   

  1. 内蒙古工业大学 数据科学与应用学院,呼和浩特 010080
  • 出版日期:2024-07-01 发布日期:2024-06-28

Research on Distributed V2V Computation Offloading Method for Internet of Vehicles Blockchain

MENG Zhen, REN Guanyu, WAN Jianxiong, LI Leixiao   

  1. College of Data Science and Application, Inner Mongolia University of Technology, Hohhot 010080, China
  • Online:2024-07-01 Published:2024-06-28

摘要: 车辆边缘计算是通过车辆间传输计算任务来完成计算卸载,可以提高车辆的计算能力。目前的卸载任务分配策略没有同时考虑数据安全性、卸载任务优先级、计算资源释放和激励车辆共享计算资源,难以适应动态车辆环境。针对以上背景,研究分布式车对车(V2V)计算卸载问题,建立马尔科夫决策过程,设计了基于深度强化学习的计算卸载最优任务分配策略,利用动态定价来激励车辆共享计算资源,考虑卸载任务优先级和计算资源释放机制。将卸载方案嵌入到区块链中,通过建立车联网区块链的身份认证机制,对车辆信息、任务信息、交易信息等敏感信息进行加密处理,满足保障数据安全性的需求。仿真实验结果验证了所提出方法的性能,与其他算法相比,该方法在节省10.28%训练时间的情况下,将系统平均效益至少提高了9.58%。

关键词: 车辆边缘计算, 计算卸载, 深度强化学习, 区块链

Abstract: Vehicle edge computing enhances the computational capabilities of vehicles by offloading computations. The existing task offloading strategies do not simultaneously consider data security, priority of offloading tasks, computation resource release, and incentivizing vehicle to share computation resources, making it difficult to adapt to dynamic vehicle environments. To address these problems, this paper investigates the distributed vehicle-to-vehicle (V2V) computation offloading problem, establishes a Markov decision process, and designs a deep reinforcement learning-based optimal task allocation strategy for computation offloading. This paper utilizes dynamic pricing to incentivize vehicle to share computation resources while considering task priorities and computation resource release mechanisms. Additionally, this paper embeds the offloading scheme into the blockchain, employs the vehicles blockchain with identity authentication mechanisms to encrypt sensitive information such as vehicle information, task information, and transaction information, thus ensuring data security. Simulation experimental results validate the performance of the proposed method. Compared with other algorithms, the average efficiency of the system is improved by at least 9.58% while the training time is saved by 10.28%.

Key words: vehicular edge computing, computation offloading, deep reinforcement learning, blockchain