基于无网格压缩感知的雷达通信一体化系统目标参数估计方法研究

本文基于一种基于正交频分复用（Orthogonal Frequency Division Multiplexing，OFDM）的雷达通信一体化系统，利用目标时延与多普勒在时频域优良的稀疏性，提供了多种基于无网格的二维时频联合估计方法以解决传统稀疏恢复方法字典失配导致的性能较差的问题，有效提高了动目标参数估计性能，并提供了多测量向量（Multiple measurement vector，MMV）模型，可有效解决低信噪比下目标参数估计性能差的问题。同时，针对基于矢量二维原子范数计算量较大的问题，本文利用半正定规划（SDP）将传统方法的高维Toeplitz矩阵解耦为两个低维Toeplitz矩阵，可将计算复杂度降低几个数量级，同时保留原子范数在超分辨性能的优势，可适应于OFDM多子载波与多符号数波形体制。仿真结果证明该方法保持了原子范数类算法的估计性能优势，并显著减少了计算量。 

Research on Target Parameter Estimation Method of Radar Communication Integrated System Based on Grid-less Compression Sensing

? ?In this paper， an integrated radar and communication system based on Orthogonal Frequency Division Multiplexing （OFDM） is considered. And based on the excellent sparsity of target delay and Doppler in time and frequency domain， a variety of grid-less two-dimensional delay-Doppler estimation methods are provided solve the problem of poor performance caused by mismatch of estimation dictionary of traditional sparse recovery methods， effectively improving the performance of moving target parameter estimation. For the problems of poor target estimation accuracy and low recovery success rate under low SNR， this paper provides a multiple measurement vector （MMV） model to effectively solve the problem of poor target parameter estimation performance under the above problems. Aiming at the problem that the two-dimensional atomic norm based on the traditional vector calculation method will generate a huge amount of calculation. this paper uses semi-definite programming （SDP） to decouple the high-dimensional Toeplitz matrix of the traditional method into two low-dimensional Toeplitz matrices， which can reduce the computational complexity by several orders of magnitude， while retaining the advantages of atomic norm in super resolution performance. This method can be applied to OFDM multi subcarrier and multi symbol waveform systems. Numerical results show that the algorithm maintains the estimation performance advantage of atomic norm class algorithms， and simulation reduces the computational complexity.