Research On Parameter Estimation For MIMO Radar Based On Compressed Sensing

Compared with the traditional radar,Multiple Input Multiple Output(MIMO)radar can overcome the performance loss suffering from Radar Cross Section(RCS),and obtains more observation channels and degrees of freedom by using diversity technology.It not only improves the target detection performance and parameter estimation accuracy,but also has the advantages of anti-jamming and anti-stealth.But with the advantages of high resolution and multiple channels,MIMO radar system also brings a large amount of data to be processed.Applying traditional parameter estimation methods to MIMO radar will result in a larger data scale,which is not conducive to real-time signal processing.In fact,in most radar scenarios,target density is very low.The sparseness of target and the complexity of massive data operation form an obvious information asymmetry,which provides a wedge for the application of Compressed Sensing(CS)theory.CS technology breaks through the traditional Nyquist sampling theorem,greatly reduces the amount of data and hardware cost of the system,therefore,it has promising prospect in applications.The accuracy of existing CS estimation algorithms for MIMO radar system is limited.The base mismatch(the real target deviates from the preset grid point)is one of the main reasons that lead to an unsatisfactory performance.Focusing on this issue,the paper studies the sparse signal reconstruction problem and adaptive grid correction problem in the case of grid mismatch to achieve good estimation performance.The main contributions of this paper are described as follows:1.This paper not only introduces the CS theoretical framework and classical reconstruction algorithms,but also studies the principle and classification of MIMO radar,derives the mathematical models of two conventional collocated MIMO radars.On this basis,we extend the traditional estimation methods to the DOA estimation for monostatic MIMO radar.Furthermore,we analyze the processes of MIMO-CAPON,MIMO-MUSIC and MIMO-ESPRIT in detail.2.Based on CS theory,this paper studies the DOA estimation for monostatic MIMO radar,solves the problem of large computation and limited estimation accuracy in current algorithms.Considering the sparse characteristics of detection targets,we firstly establish a sparse echo model,secondly construct a sparse observation dictionary with first-order approximation,finally propose a grid adaptive DOA estimation algorithm(GADE).The algorithm is a two-layer iterative algorithm based on sparse Bayesian learning,which includes sparse recovery,grid deviation learning and parameter updating in the inner layer,and sparse observation dictionary reconstruction in the outer layer.Through internal and external iteration,the proposed algorithm achieves grid adaptation,overcomes the estimation error,and has better estimation accuracy than the state-of-the-art methods.3.Based on CS theory,this paper studies the joint angle-range estimation for FDA-MIMO radar.Aiming to explore the information of range and angle,we model the FDA-MIMO radar system and deduce the joint transmit-receive space frequency.Moreover,we construct a two-dimensional angle-range domain grid,then derive the formula of the grid deviation vector,finally propose a grid adaptive joint angle-range estimation algorithm(GAJARE).The algorithm corrects grid error based on sparse Bayesian learning,acquires the flexibility of initial angle and range grid granularity.In addition,the algorithm can directly achieve accurate target localization without the matching process of angle and range.Simulation results show that the algorithm achieves higher estimation accuracy and better anti-interference ability than the traditional algorithms.