Abstract:

To adapt the rapid dynamic change of network topology and sudden network congestion in emergency communication, a routing algorithm MLSA suitable for emergency communication network is proposed. In this method, the ant colony algorithm framework is combined with the reinforcement learning process for system modeling. The performance statistical analysis process is limited to higher priority areas. By detecting the status information of local neighbor nodes, the different routing decision processes are scored, and the policy adjustment is made according to the network feedback, thereby the overall performance of the network is improved and network congestion is relieved. It increases the real-time and stability of data transmission. Experimental results show that in the environment nodes move in high speed, MLSA has obvious performance advantages over the classic on-demand routing protocols AODV and DSR for network congestion caused by emergency communication.

Key words: emergency communication, ant colony algorithm, reinforcement learning, Qos routing, traffic equalization

摘要：

针对应急通信中网络拓扑的快速动态变化及突发的网络拥塞现象，提出了一种适用于应急通信网络的路由算法MLSA，利用蚁群算法框架结合强化学习过程进行系统建模，将网络链路性能统计分析过程限制在较高优先级的区域，通过探测局部邻居节点的状态信息，对不同路由决策过程进行打分，并根据网络反馈做出策略调整，从而改善网络整体性能，有利于缓解网络拥塞，增加了数据传输的实时性、稳定性。实验证明，在较高节点移动性环境下，针对应急通信中产生的网络拥塞现象，MLSA较经典按需路由协议AODV和DSR，具有明显的性能优势。

关键词: 应急通信, 蚁群算法, 强化学习, Qos路由, 流量均衡