Research On Distributed Radar Network Location And Power Allocation

A distributed MIMO radar network has become a trend in the development of domestic and foreign radar.Its main advantage is the information sharing within the network,so that each radar can obtain and recognize more information,and then self-correct parameters and enhance the performance.Through the feedback of information received by the receiving station,a distributed MIMO radar network can adjust its transmitting parameters adaptively according to the feedback information.In practical applications,the resources of a radar network system are usually limited to not support full-power work style at all stations.In order to achieve the optimal localization,how to utilize the prior information derived from the receiving station is the main research purpose during this dissertation.For this purpose,the dissertation proposes to allocate the transmit resources reasonably in different systems and backgrounds to improve the target localization performance of the distributed radar network systems.According to the numbers of radars and the targets,the main content of this dissertation is divided into several points: Self-Sending and Self-Receiving radar localization for the only target,bistatic MIMO localization for single target,several targets localization from a bistatic MIMO radar system,and a bistatic MIMO radar system localization for many targets which distribute in a wide area.On the basis of fully understanding and referring to the related domestic and foreign algorithms,this dissertation makes a carefull research on the power allocation algorithm of a distributed radar network system,and the main contributions are as follows:1.A power allocation algorithm for target localization in a MIMO radar is proposed,in order to achieve the optimal allocation under the case of the limited power in a system.Firstly,Cramer Rao lower bound corresponding to the target-localization minimum mean square error is given,and regarded as the(objective)cost function of optimization.Secondly,it is proved that the objective function is nonconvex from the basic definition of convex function.Thirdly,chapter two improve the existing convex relaxation algorithm.Lastly.the parameters of the convex relaxation are corrected by self feedback,so that the final results can be approximated optimally.2.Power allocation can raise the utilization of power resources in a distributed radar system.This section analyze two characteristics of the Cramer Rao bound(Cramer-Rao low bound,simply referred to as CRLB)to target localization MSE(Mean square error,referred to as MSE)for distributed radar network system.On the foundation of the classical power allocation methods,we propose an efficient power allocation algorithm and apply it in cognitive distributed radar systems.We call this algorithm as alternate global search algorithm(Alternating Global Search Algorithm,referred to as AGSA).In this dissertation,the desired objective is to directly estimate the target-localization CRLB which is nonconvex and nonlinear.The convergence of the algorithm is theoretically analyzed by using the invariance principle of LaSalle.We analyze the computational complexity of the related algorithms.Through this algorithm,the well-known Pareto optimal solution set related to power allocation is obtained,which can indirectly minimize the total power under ensuring the needed localization accuracy.3.To solve the the problems of power allocation in MIMO radar for locating multiple targets,we do the following research:(1)two traditional optimization models are integrated into a more complete model of power allocation by using their complementary advantages;(2)combining with alternating search,Sequential quadratic programming method and extremum-search,we design a programming problem for Bi-objective optimization method;(3)for the power allocation of target group,this dissertation gives the corresponding adaptive algorithm.4.With the application background of localization for a huge number of discrete targets,a search method based on function values is designed by combining human behavior algorithm and genetic algorithm.For a power allocation problem of simultaneous localization of many targets,even if the dimension is reduced by the minimal variable set,the function is still very complex,and the traditional optimization algorithm usually be traped into the local minimum point.In case of solving this problem,this dissertation designs the algorithm.The algorithm mainly includes:(1)In genetic algorithm,the several optimal function values are selected as the starting point of the next step for genetic transmission;(2)In this algorithm,just a single peak point cannot stop the searching,it will imitate human behavior and continue to search for the next peak point.