基于进化势博弈的多无人机传感器网络K-覆盖

An evolutionary potential game theoretic approach for the K-COVER problem in multi-UAV sensor networks

针对多无人机传感器网络K-覆盖问题,提出了基于势博弈与Log-linear学习的分布式最优传感器配置方法.介绍了传感器最优配置的国内外研究现状,特别对基于进化势博弈理论的方法进行了详细的阐述.利用博弈理论对无人机分布式传感器配置问题进行建模,通过个体局部收益函数与全局性能函数的设计将问题构造成网络势博弈.提出了一种基于时变Log-linear学习的分布式求解方法,并利用非齐次马尔可夫链理论对收敛性与最优性进行了证明.仿真对比实验证明了所提算法的可行性、有效性和优越性.

This paper focuses on the K-COVER problem in unmanned aerial vehicle （UAV） sensor networks and proposes a game theoretic approach based on the network potential game and Log-linear learning. After detailing the research progress at home and abroad, we firstly model the K-COVER problem in UAV networks as network potential game via appropriately designing of individual and global payoff functions. Secondly, a Log-linear learning based algorithm is proposed, whose convergence and optimality are proven using inhomogeneous Markov chain theory. Simulation results demonstrate the feasihlity, effectiveness, and superiority.