基于Q学习的Ad Hoc网络路由协议的改进与研究

Improvement and Research of Ad hoc Network Routing Protocol Based on Q Learning

采用Q学习可以有效保证Ad Hoc网络的可靠性,该文将启发Q学习算法应用在备用多径按需路由协议的路由寻找中,融合传统Q学习算法根据网络状态适当的调整路由周期寿命,提高路由对动态变化的业务分布的自适应性。建立NS3网络模型对无线网络的时延、吞吐量及开销等主要性能指标进行理论分析,研究结果表明:优化后的路由协议更加适用于状态多变的移动Ad Hoc网络。

Q-learning can effectively guarantee the reliability of Ad Hoc network. In this paper,heuristic Q-learning algo-rithm is applied to the routing search of standby multipath on-demand routing protocol,which integrates traditional Q-learning algo-rithm to adjust the routing cycle life according to the network status,and improves the adaptability of routing to dynamic traffic dis-tribution. The NS3 network model is established to analyze the delay,throughput and overhead of wireless network. The results showthat the optimized routing protocol is more suitable for mobile Ad Hoc networks with changeable states.