多目标联合优化的车联网动态资源分配算法  

Dynamic resource allocation algorithm for multi-objective joint optimization in Internet of vehicle

作　　者：宋晓勤[1,2] 张文静 雷磊[1] 宋铁成[2] 赵丽屏 SONG Xiaoqin;ZHANG Wenjing;LEI Lei;SONG Tiecheng;ZHAO Liping(School of Electronic and Information Engineering,Nanjing University of Aeronautics and Astronautics,Nanjing 211106,China;National Mobile Communications Research Laboratory,Southeast University,Nanjing 210096,China;The 63rd Research Institute,National University of Defense Technology,Nanjing 210007,China)

机构地区：[1]南京航空航天大学电子信息工程学院,南京211106 [2]东南大学移动通信全国重点实验室,南京210096 [3]国防科技大学第六十三研究所,南京210007

出　　处：《东南大学学报(自然科学版)》2025年第1期266-274,共9页Journal of Southeast University:Natural Science Edition

基　　金：国家自然科学基金资助项目(62371232);东南大学移动通信全国重点实验室开放基金资助项目(2024D13)。

摘　　要：为了解决车联网(IoV)信道高动态不确定性及多用户干扰所导致的通信传输性能下降问题,提出了一种基于多智能体增强型双深度Q网络(EDDQN)的多目标联合优化资源分配算法。首先,考虑车辆运动和信道时变特性,建立多用户干扰下频谱共享和功率控制联合优化的资源分配决策模型,在满足时延和可靠性等约束下,最小化网络时延和能耗加权和(成本);然后,将模型转换为马尔可夫决策过程(MDP),利用双深度Q网络(DDQN),并引入优先经验回放和多步学习,通过集中式训练和分布式执行,优化车间(V2V)链路的频谱共享和功率分配策略。结果表明,所提算法具有良好的收敛性,在不同负载下相较对比算法成本减少8%以上,负载传输成功率提升19%以上,有效提高了通信传输性能。To address the problem of communication transmission performance degradation resulting from high dynamic uncertainty of channels and multi-user interference in the Internet of vehicles(IoV),a multi-ob-jective joint optimization resource allocation algorithm based on multi-agent enhanced double deep Q network(EDDQN)is proposed.First,considering the vehicle motion and the time-varying channel characteristics,a resource allocation decision model for the joint optimization of spectrum sharing and power control under multi-user interference is established to minimize the cost,which is the weighted sum of network delay and en-ergy consumption,under the constraints of delay and reliability.Then,the model is transformed into a Mar-kov decision process(MDP).The priority experience replay and the multi-step learning are introduced into the double deep Q-network(DDQN)with centralized training and distributed execution to optimize the sub-band and power allocation strategies for vehicle to vehicle(V2V)links.The results demonstrate that the pro-posed algorithm has good convergence.Compared with the benchmark algorithms,the proposed algorithm re-duces the cost by over 8%and increases the transmission success rate by over 19%under different loads,effec-tively improving the communication transmission performance.

关 键 词：车联网 多用户干扰 多目标联合优化 深度强化学习 

分 类 号：TN92[电子电信—通信与信息系统]