【文献阅读03】Deep Reinforcement Learning Based Resource Allocation for V2V CommunicationsDeep Reinforcement Learning Based Resource Allocation for V2V Communications(点击可见原文)
Deep Reinforcement Learning Based Resource Allocation for V2V Communications(点击可见原文)
p.s.此文19年发表,到20年8月被引199次
论文要解决的问题
单播和广播场景下,考虑V2V通信的资源分配,使用分布式方案,在无 global information 的前提下为 V2V链路 or 车辆 找到最优的子带和功率等级,该算法能满足V2V链路的延迟约束并最小化对 V2I 的干扰。
使用深度强化学习解决,已开源并有哥们写了double DQN的版本。
通信场景 // for unicast communication
【文献阅读03】Deep Reinforcement Learning Based Resource Allocation for V2V CommunicationsDeep Reinforcement Learning Based Resource Allocation for V2V Communications(点击可见原文)
存在 M 个 V2I 链路,K 对 V2V 链路。为提高频谱效率,假设V2V共享V2I的上行链路频谱,这是因为基站处的干扰更易控制且上行链路使用较少。
m号V2I 的 SINR 为:
【文献阅读03】Deep Reinforcement Learning Based Resource Allocation for V2V CommunicationsDeep Reinforcement Learning Based Resource Allocation for V2V Communications(点击可见原文)
其中
【文献阅读03】Deep Reinforcement Learning Based Resource Allocation for V2V CommunicationsDeep Reinforcement Learning Based Resource Allocation for V2V Communications(点击可见原文)
和
【文献阅读03】Deep Reinforcement Learning Based Resource Allocation for V2V CommunicationsDeep Reinforcement Learning Based Resource Allocation for V2V Communications(点击可见原文)
分别表示 m号CUE和 k号VUE 的发射功率,h是与 m号CUE相关的信道的增益,
【文献阅读03】Deep Reinforcement Learning Based Resource Allocation for V2V CommunicationsDeep Reinforcement Learning Based Resource Allocation for V2V Communications(点击可见原文)
是 k号VUE的干扰增益,
【文献阅读03】Deep Reinforcement Learning Based Resource Allocation for V2V CommunicationsDeep Reinforcement Learning Based Resource Allocation for V2V Communications(点击可见原文)
为频谱分配的 indicator
m号V2I 的容量为:
【文献阅读03】Deep Reinforcement Learning Based Resource Allocation for V2V CommunicationsDeep Reinforcement Learning Based Resource Allocation for V2V Communications(点击可见原文)
k号VUE的SINR为:
【文献阅读03】Deep Reinforcement Learning Based Resource Allocation for V2V CommunicationsDeep Reinforcement Learning Based Resource Allocation for V2V Communications(点击可见原文)
其中,
【文献阅读03】Deep Reinforcement Learning Based Resource Allocation for V2V CommunicationsDeep Reinforcement Learning Based Resource Allocation for V2V Communications(点击可见原文)
,
【文献阅读03】Deep Reinforcement Learning Based Resource Allocation for V2V CommunicationsDeep Reinforcement Learning Based Resource Allocation for V2V Communications(点击可见原文)
分别表示 与V2I共享频谱引入的干扰功率、与其他V2V共享频谱引入的干扰功率,
【文献阅读03】Deep Reinforcement Learning Based Resource Allocation for V2V CommunicationsDeep Reinforcement Learning Based Resource Allocation for V2V Communications(点击可见原文)
表示k号VUE的功率增益,
【文献阅读03】Deep Reinforcement Learning Based Resource Allocation for V2V CommunicationsDeep Reinforcement Learning Based Resource Allocation for V2V Communications(点击可见原文)
是m号CUE的干扰功率增益,
【文献阅读03】Deep Reinforcement Learning Based Resource Allocation for V2V CommunicationsDeep Reinforcement Learning Based Resource Allocation for V2V Communications(点击可见原文)
是 k' 号VUE的干扰功率增益
k号VUE的容量为:
【文献阅读03】Deep Reinforcement Learning Based Resource Allocation for V2V CommunicationsDeep Reinforcement Learning Based Resource Allocation for V2V Communications(点击可见原文)
【文献阅读03】Deep Reinforcement Learning Based Resource Allocation for V2V CommunicationsDeep Reinforcement Learning Based Resource Allocation for V2V Communications(点击可见原文)
V2I和V2V的容量作为衡量此分配方案导致的对V2I和V2V的干扰的指标
时延作为penalty
累计奖励:
【文献阅读03】Deep Reinforcement Learning Based Resource Allocation for V2V CommunicationsDeep Reinforcement Learning Based Resource Allocation for V2V Communications(点击可见原文)
【文献阅读03】Deep Reinforcement Learning Based Resource Allocation for V2V CommunicationsDeep Reinforcement Learning Based Resource Allocation for V2V Communications(点击可见原文)
【文献阅读03】Deep Reinforcement Learning Based Resource Allocation for V2V CommunicationsDeep Reinforcement Learning Based Resource Allocation for V2V Communications(点击可见原文)
多播场景V2V资源分配的DRL解法
系统模型
【文献阅读03】Deep Reinforcement Learning Based Resource Allocation for V2V CommunicationsDeep Reinforcement Learning Based Resource Allocation for V2V Communications(点击可见原文)
CUE用户集(CUE主体为车,连接到BS,其实就是V2I用户集)为
【文献阅读03】Deep Reinforcement Learning Based Resource Allocation for V2V CommunicationsDeep Reinforcement Learning Based Resource Allocation for V2V Communications(点击可见原文)
,V2V用户集为
【文献阅读03】Deep Reinforcement Learning Based Resource Allocation for V2V CommunicationsDeep Reinforcement Learning Based Resource Allocation for V2V Communications(点击可见原文)
【文献阅读03】Deep Reinforcement Learning Based Resource Allocation for V2V CommunicationsDeep Reinforcement Learning Based Resource Allocation for V2V Communications(点击可见原文)
m号CUE用户的容量为:
【文献阅读03】Deep Reinforcement Learning Based Resource Allocation for V2V CommunicationsDeep Reinforcement Learning Based Resource Allocation for V2V Communications(点击可见原文)
对于 k 号车辆的第 j 个接收者,其SINR为
【文献阅读03】Deep Reinforcement Learning Based Resource Allocation for V2V CommunicationsDeep Reinforcement Learning Based Resource Allocation for V2V Communications(点击可见原文)
其中
【文献阅读03】Deep Reinforcement Learning Based Resource Allocation for V2V CommunicationsDeep Reinforcement Learning Based Resource Allocation for V2V Communications(点击可见原文)
、
【文献阅读03】Deep Reinforcement Learning Based Resource Allocation for V2V CommunicationsDeep Reinforcement Learning Based Resource Allocation for V2V Communications(点击可见原文)
k 号车辆的第 j 个接收者的信道容量为:
【文献阅读03】Deep Reinforcement Learning Based Resource Allocation for V2V CommunicationsDeep Reinforcement Learning Based Resource Allocation for V2V Communications(点击可见原文)
【文献阅读03】Deep Reinforcement Learning Based Resource Allocation for V2V CommunicationsDeep Reinforcement Learning Based Resource Allocation for V2V Communications(点击可见原文)
【文献阅读03】Deep Reinforcement Learning Based Resource Allocation for V2V CommunicationsDeep Reinforcement Learning Based Resource Allocation for V2V Communications(点击可见原文)
【文献阅读03】Deep Reinforcement Learning Based Resource Allocation for V2V CommunicationsDeep Reinforcement Learning Based Resource Allocation for V2V Communications(点击可见原文)
左图:车多导致V2V多,导致对V2I造成的干扰增大,导致V2I速率下降
右图:纵坐标为满足延迟约束的车辆的占比
【文献阅读03】Deep Reinforcement Learning Based Resource Allocation for V2V CommunicationsDeep Reinforcement Learning Based Resource Allocation for V2V Communications(点击可见原文)
【文献阅读03】Deep Reinforcement Learning Based Resource Allocation for V2V CommunicationsDeep Reinforcement Learning Based Resource Allocation for V2V Communications(点击可见原文)
【文献阅读03】Deep Reinforcement Learning Based Resource Allocation for V2V CommunicationsDeep Reinforcement Learning Based Resource Allocation for V2V Communications(点击可见原文)