Research On Intelligent Tracking And Scheduling Algorithm For Targets With Complex Motion Trajectory In Mobile Sensor Networks

Recently,mobile sensor networks(MSN)has attracted more and more researchers' attention due to its advantages of network flexibility,low energy consumption and network robustness.As a core application in wireless sensor networks(WSN),target tracking is also a research hotspot in MSN.However,the introduction of mobile nodes in MSN makes target tracking and node scheduling more difficult.Besides,tracking target with complex motion trajectories is also a difficulty in WSN.Therefore,a series of studies on tracking target with complex motion trajectories and node scheduling in MSN will be carried out in this thesis.The main work of this thesis is as follows:(1)Aiming at the tracking problem of the targets with complex moving trajectorys,this thesis designs an application scenario of target tracking using fixed nodes in MSN,and proposes an intelligent tracking algorithm based on deep deterministic policy gradient algorithm for targets with complex moving trajectorys in this scenario.In this thesis,the convergence and stability of the algorithm are verified by training results,and the influence of different reward parameters on the training results of the algorithm is discussed.It is verified that the tracking accuracy and tracking energy consumption in this algorithm are contradictory.In addition,the proposed intelligent tracking algorithm is compared with the greedy strategy based interacting multiple model(IMM)algorithm by simulation.The simulation results show that the intelligent tracking algorithm proposed in this thesis has better adaptability,lower tracking error and lower tracking energy consumption than IMM tracking algorithm for CVAT targets and random trajectory targets.(2)Aiming at the scheduling problem of multi-mobile nodes jointly tracking targets with complex moving trajectorys,an application scenario of multi-mobile nodes jointly tracking targets in MSN is designed in this thesis.In this scenario,an intelligent scheduling algorithm based on hierarchical deep deterministic policy gradient algorithm is proposed for targets with complex moving trajectorys.In this thesis,greedy exploration strategy and random exploration strategy are used to train this algorithm,and the training results verify the convergence and stability of the algorithm.At the same time,the training results show that greedy exploration will greatly affect the performance of the exploration stage,accelerate the convergence speed of this algorithm,but has little influence on performance when algorithm converged.In addition,the proposed intelligent scheduling algorithm is compared with the IMM scheduling algorithm based on distance priority and greedy strategy.The simulation results show that the proposed intelligent scheduling algorithm has better adaptability,lower tracking error,lower tracking energy consumption and larger network coverage than IMM scheduling algorithm for CVAT targets and random trajectory targets.