作者投稿和查稿 主编审稿 专家审稿 编委审稿 远程编辑

计算机工程 ›› 2023, Vol. 49 ›› Issue (5): 191-197. doi: 10.19678/j.issn.1000-3428.0065747

• 移动互联与通信技术 • 上一篇    下一篇

基于改进MOEA/D的车联网通信资源分配算法

郑丽萍, 赵玉娟, 费选   

  1. 河南工业大学 人工智能与大数据学院, 郑州 450001
  • 收稿日期:2022-09-14 修回日期:2022-11-18 发布日期:2023-05-10
  • 作者简介:郑丽萍(1976-),女,讲师、硕士,主研方向为多目标优化、车联网资源分配;赵玉娟,副教授、硕士;费选,副教授、博士。
  • 基金资助:
    河南省科技攻关项目“面向遥感影像城市覆盖分类的多尺度多特征深度学习算法研究”(222102210108)。

Communication Resource Allocation Algorithm Based on Improved MOEA/D in Internet of Vehicles

ZHENG Liping, ZHAO Yujuan, FEI Xuan   

  1. School of Artificial Intelligence and Big Data, Henan University of Technology, Zhengzhou 450001, China
  • Received:2022-09-14 Revised:2022-11-18 Published:2023-05-10

摘要: 为获得车联网通信资源分配的最优解,提出一种基于改进MOEA/D的车联网通信资源分配优化算法。将车联网资源请求的阻塞率和资源请求成功的总成本这2个相互冲突的网络通信资源分配要素作为网络通信资源分配的2个优化目标,根据车联网中行驶车辆的特点,对请求资源车辆和提供资源车辆设置约束条件。在此基础上,采用自适应邻域策略平衡进化过程中种群的收敛性和分布性,并将迭代次数引入自适应度,调节交叉算子和变异算子,使种群中较差的个体也具有遗传性,从而保证种群的多样性。同时,随着迭代次数的增加,种群中较差个体遗传性降低,较好个体遗传能力增强,从而保证种群的优化。仿真结果表明,该算法针对最小化阻塞率和最小化成本这2个目标能够获得满意的优化效果,在迭代次数、车辆数和资源请求数变化情况下都存在最优解,在相同迭代次数下,与基于支配的多目标算法SPEA2和NSGA-II相比具有较低的阻塞率和较好的收敛性。

关键词: 车联网, 通信资源分配, 多目标进化算法, MOEA/D算法, 阻塞率, 成本

Abstract: To achieve an optimal solution for communication resource allocation in Internet of Vehicles(IoV),this study proposes a communication resource allocation optimization algorithm based on improved MOEA/D in IoV.The two conflicting network communication resource allocation elements:the blocking rate of resource requests in the vehicle network and the total cost of successful resource requests are considered as the two optimization objectives of network communication resource allocation.According to the characteristics of vehicles running in IoV,the constraint conditions are set for the vehicles requesting and providing resources.Accordingly,the adaptive neighborhood strategy is adopted to balance the convergence and distribution of the population in the evolution process,the iteration number is introduced into the adaptive degree,and the crossover operator and mutation operator are adjusted to ensure poor individuals also have heredity.Simultaneously,as the number of iterations increases,the heritability of poor individuals decreases,and that of good individuals increases,thereby ensuring the optimization of the population.The simulation results show that the algorithm can achieve satisfactory optimization results for both objectives.The optimal solution exists under different iteration times,number of vehicles,and number of resource requests.Under the same iteration times,the algorithm has lower blocking probability and better convergence compared with the dominated multi-objective SPEA2 and NSGA-II algorithms.

Key words: Internet of Vehicles(IoV), communication resource allocation, Multi-Objective Evolutionary Algorithm(MOEA), MOEA/D algorithm, blocking probability, cost

中图分类号: