基于多态Q学习的无线网络MAC层CSMA协议优化策略

A Multi-state Q-learning Based CSMA MAC Protocol Optimization Strategy for Wireless Networks

由于无线信道具有共享特性,节点之间竞争不可避免。传统P-坚持载波感知多址(CSMA)的传输概率对吞吐量有很大影响[1]。笔者设计了一种多态增强学习(RL)方法,在多态Q学习模型中提出了三种学习类型,定义传输概率Q为节点学习策略,节点没有关于网络的先验信息,只利用历史感知信息包括碰撞次数及成功传输率,来学习最佳策略[2-3]。并通过综合仿真,对不同状态定义下的Q学习模型性能进行了比较。

Due to the shared nature of wireless channels,the competition between nodes is inevitable.The traditional P-persistent carrier-aware multiple access(CSMA)transmission probability has a great influence on throughput.A multi-state reinforcement learning(RL)method is designed for this purpose.Three state representations are proposed in this multi-state Q-learning model,agents have no priori information about the network and learn their optimal strategy using the historical sensory information including the number of collisions or successful transmissions.The performance of the proposed Q-learning agents with different state definitions in comparison with each other is examined via comprehensive simulation results.