Research On Measurement Matrix Construction For CS-MIMO Radar

As a new radar system,Multiple-Input Multiple-Output(MIMO)radar can make full use of space diversity and waveform diversity to obtain more degrees of freedom so as to improve target detection ability and parameter estimation performance.The sparsity of the target distribution in MIMO radar detection scene provides the conditions for the application of the Compressed Sensing(CS)theory,and the parameter estimation in MIMO radar based on CS has become one of the hot topics in radar signal processing field.This paper introduces CS theory into MIMO radar system architecture,studies on the relationship between waveform design,array configuration and CS measurement and puts forward the measurement matrix design idea based on waveform design and array configuration.Besides,the measurement matrix is optimized through waveform and array optimization design to improve parameter estimation performance in CS-MIMO radar.The main work of this paper is as follows:The CS-MIMO radar signal model is studied,including basic theory of CS and the parameter estimation principle in CS-MIMO radar.Then the influence factors of the reconstruction performance are analyzed.Besides,the methods of measurement matrix construction and optimization are discussed,and the idea of equivalent construction of measurement matrix based on waveform design and array configuration is put forward.The chaotic sequence is used to design MIMO radar transmitted signal and the chaotic pseudo random signal is considered as the measurement operator,which turn the optimization problem of measurement matrix in CS-MIMO radar into waveform optimization problem equivalently and reduce the complexity of system model effectively.On that basis,in order to optimize the signal spectral shape,suppress the interference noise and ensure the orthogonality and randomness of the desired waveform,a cognitive MIMO radar waveform optimization algorithm is proposed.Simulation results show that the equivalent model and optimization algorithm are effective and can significantly improve the accuracy of parameter estimation.The CS-MIMO radar signal model based on sparse random linear array is investigated,in which the compressed measurement is realized by the randomness of array element positions.Furthermore,it is guaranteed that the equivalent sensing matrix could satisfy the CS non-uniform recovery property.In order to reduce the coherence of the equivalent sensing matrix and enhance the sparsity upper limit to improve the reconstruction performance,the measurement matrix is optimized by optimizing thearray configuration based on Simulated Annealing(SA)algorithm.Simulation results show that the algorithm can improve the DOA reconstruction probability and estimation accuracy.One-dimensional(1-D)sparse random linear array is extended to two-dimensional(2-D)L-shaped array in CS-MIMO radar.It is proved that the Kronecker product of the transmitting and receiving array steering vectors can play the role of equivalent measurement matrix and satisfy the non-uniform recovery property.Also,in order to reduce the coherence of equivalent sensing matrix,a L-shaped array optimization method based on Particle Swarm Optimization(PSO)algorithm is studied,which obtains the optimization of the measurement matrix.Simulation results show that the optimization algorithm can effectively improve the performance of 2-D parameter estimation in CS-MIMO radar.