基于机器学习链路权重优化的无人机网络路由算法

针对分布式“马赛克战”场景下侦察-判断-决策-行动(observe orient decide act,OODA)环通信加速的要求，提出了基于机器学习链路权重优化的无人机集群网络路由算法。针对OODA环当前阶段业务量通信需求，对业务量通信的完成时间进行建模，以最小化业务量通信完成时间为优化目标，通过利用机器学习梯度下降方法实现无人机集群网络分布式路由链路权重的优化，从而满足OODA环的通信加速要求，使己方可以先敌行动，获得战场主动权。仿真表明，相比于现有无人机网络的路由算法，提出的算法能显著降低OODA环的通信时间，提高数据报文传输成功率。

Research on UAV networks routing algorithm based on machine learning-enabled link weight optimization

To accelerate communication procedure of OODA loop in the scenario of distributed mosaic warfare, we propose a novel unmanned aerial vehicle (UAV) routing algorithm based on machine learning-enabled link weight optimization in this paper. Our algorithm first constructs the communication completion time model for the procedure of OODA loop. Then, to minimize the communication completion time of OODA loop, the algorithm leverages the gradient descent method of machine learning to achieve link weight optimization for UAV cluster network distributed routing. Computer simulation results show that, compared with the current UAV routing algorithm, our algorithm can observably shorten the communication time of OODA loop, and increase the rate of packet success transmission.