New Methods For MIMO Radar Antenna Placement Based On Target Detection

This thesis studies the optimal antenna placement for target detection in distributed multiple-input multiple-output(MIMO)radar.Considering the problem of placing a limited number of radar antennas in a given spacial area to maximize the detection performance,an optimization problem is formulated with MIMO radar detection performance as a metric.The problem is a high-dimensional nonlinear integer programming problem.With the increase of the number of antennas and the feasicale spacial area,the problem size increases and the computational complexity of trandition methods may increase dramatically.In this work,we discretize the feasicable spacial area into several grid points.Then the antennas and grid points can be viewed as the vertices in a bipartite graph,and the impact of the placement on the detection performance is characterized by the edges with weights.Thus,the antenna placement problem is transformed into an equivalent matching problem in a bipartite graph.We propose to use the stable matching(SM)method and the maximum weighted matching(MWM)method to solve the antenna placement problem,the solution of which can be obtained within the polynomial complexity.The MIMO radar received signal model including the selection variables of the transmit antenna positions is built and the optimal detector under the Newman-Pearson(NP)criterion is derived.The resulting performance is investigated in the indicator of the output signal-to-noise radio(SNR)of the detector,and the influence of the antenna placement on the detection performance is illustrated.The detector output SNR is derived,which is used as an objective function to formulate the optimal antenna placement problem.This optimization problem is further studied and transformed into an equivalent bipartite graph matching problem.A bipartite graph model is developed and the antenna placement based on SM and MWM matching methods are proposed,respectively.The SM and MWM based methods can achieve optimal performance approximately and exactly,both with affordably low computational complexity.For placing antennas with the SM based methods,the Gale-Shapley(GS)algorithm is employed.Note that,using the SM matching method,assigning the priority to the grid points(positions)or the antennas influences on the detection performance.When the priority is given to the positions side,a GS algorithm based on position preference(PGS)is proposed.When the antennas side has priority,a GS algorithm based on antenna preference(AGS)is proposed.Through theoretical and simulation analysis,it is proved that the SM results obtained by PGS and AGS methods are the same.They can reach the performance that is very close to the optimal antenna placement,while the calculation complexity is much lower than that of the traditional methods.Relatively,the complexity of AGS algorithm is smaller than that of the PGS algorithm.For placing antennas with the MWM based methods,the Kuhn-Munkres(KM)algorithm is adopted.By adding virtual antennas to match the number of antennas and the number of spatial grid points,a virtual-antenna KM(VKM)algorithm is proposed.To reduce the complexity,through analysis,an antenna placement method based on the extended KM(EKM)algorithm is proposed.It is shown that both algorithms can provide optimal solution to the antenna placement problem exactly,with much lower computational complexity compared with that of the traditional methods.Comparaly,the complexity of the EKM is lower than that of the VKM.Compared with the AGS algorithm which cannot always solve the problem optimally,the EKM method can insure optimal solution at the cost slightly higher complexity.The influence of the number of antennas on detection performance is also discussed,when the antennas are placed according to the EKM and AGS algorithms.It is shown that more antennas can lead to better performance.However,when the antenna number reaches to a threshold,further adding more antennas leads to little performance gain.The threshold of antenna number is analyzed,which can provide a reference to the system designers for assigning the antenna resources.