【摘要】 在节点高速移动或节点密度较大的移动无线自组织网络中，传统AODV算法在路由请求使用洪泛广播RREQ,选择路由跳数最少的链路，并没有考虑到网络拓扑的频繁变化导致的链路中断，在节点数量较多时其洪泛所导致的广播风暴将对网络性能产生影响。针对上述情形，提出了一种基于GPS信息和Q学习相结合的AODV改进算法，GQ-AODV算法同时考虑了节点位置和节点速度，通过节点位置计算偏差角度和前程值，节点与下一跳节点的相对速度来确定链路稳定度，采取下一跳节点与其邻居节点的平均相对速度、Q学习训练的下一跳节点与其邻居节点的历史平均相对速度，来避免下一跳选取陷入局部最优。NS3仿真表明，GQ-AODV算法能够选择较好的下一跳，降低了路由开销、时延和抖动，提高了分组投递率和吞吐量，在节点数目较多的场景下更具优势。更多还原

【Abstract】 In mobile ad hoc networks with high-speed mobile nodes or high node density, the traditional AODV algorithm uses flooding broadcast RREQ in routing requests, and selects the link with the least hops. It does not consider the link interruption caused by frequent changes in network topology. When the number of nodes is large, the broadcast storm caused by flooding will have an impact on network performance. In view of the above situation, an improved AODV algorithm based on GPS information and Q-learning is proposed. The GQ-AODV algorithm considers the node position and node speed at the same time. The link stability is determined by calculating the deviation angle and the forward value through the node position, and the relative speed between the node and the next hop node Q-learning is used to train the historical average relative speed of the next hop node and its neighbors to avoid the next hop selection falling into local optimum. NS3 simulation results show that GQ-AODV algorithm can select better next hop, reduce routing overhead, delay and jitter, improve packet delivery rate and throughput, and has more advantages in the scenario with more nodes.更多还原