车联网中基于Q学习的地理位置路由协议

由于道路拓扑结构的限制以及车辆节点的快速变化，车联网路由协议正面临着很多挑战，例如道路的低连通性、较大的延时以及高开销等。为解决此类问题，提出了一种基于Q学习的地理位置路由协议。该协议将地理区域划分成大小一致的正方形，称为网格。在给定目的地的情况下，根据历史交通流信息计算出车辆从当前网格向不同方向的邻居网格移动的Q值，每辆车存储Q值表，通过查询Q值表选择最优下一跳网格。在选定的下一跳网格中，选择距离目的地最近的车辆，当最优下一跳网格中没有邻居车辆时，选择次优下一跳网格中的车辆。仿真结果表明，与其他基于地理位置的路由协议相比，所提协议能够提高分组投递率，降低传输延时并减少通信跳数。

Geographic Location Routing Protocol Based on Q-learning in Internet of Vehicles

Due to the limitation of road topology and the rapid change of vehicle nodes,the routing protocol of Internet of Vehicles is facing many challenges,such as low connectivity,large delay and high overhead.In order to solve these problems,the authors propose a geographic location routing protocol based on Q-learning.The protocol divides the geographical area into squares with the same size,which is called grid.In a given destination,the Q-value of the vehicle moving from the current grid to the neighbor grid in different directions is calculated according to the historical traffic flow information.Each vehicle stores the Q-table,and then selects the optimal next hop grid by querying the Q-value table.In the selected next hop grid,the neighbor vehicle closest to the destination will be selected.When there are no neighbor vehicles in the optimal next hop grid,the vehicles in the suboptimal next hop grid will be selected.Simulation results show that the proposed protocol can improve the packet delivery ratio,reduce the transmission delay and the number of communication hops when compared with other location-based routing protocols.