Sparse Channel Recovery Algorithm In MIMO-OFDM System

MIMO-OFDM technology has the advantages of high speed, high efficiency and high reliability. It becomes the core technology in the wireless communication system. Channel estimation is an indispensable technology in MIMO-OFDM system. It provides the channel state information that MIMO system pre-coding, space-time decoding and coherent demodulation at the receiver needed.Traditional channel estimation algorithms based on the assistance of extensive pilots, which reduce the spectral efficiency and throughput of the system. Experiments show that many wireless channels are sparse. Since compressed sensing theory proposed in recent years, sparse signal can be reconstructed with high probability using only a few sample points. More and more experimental evidences show that Sparse channel estimation based on the CS theory, reduces the using of pilots, improves the utilization efficiency of channel spectrum, and achieves higher channel estimation accuracy compared to traditional channel estimation.This paper studies the recovery algorithms of MIMO-OFDM system sparse channel estimation, mainly analyzes the application of greedy reconstruction algorithm in channel estimate field. Specific research contents and innovations are as follows: 1. In this paper, the conventional dense MIMO-OFDM system channel and the sparseMIMO-OFDM system channel are modeled. The classical traditional channelestimation algorithms LS and MMSE are simulated, compared and analysed. Theapplication of CS theory in channel estimation is introduced in detail. The classic OMP,CoSaMP, ROMP algorithm are simulated under MIMO-OFDM system sparse channelmodel, while the OMP algorithm and LS algorithm are simulation under MIMO-OFDMsystem normal channel model. The two experimental results are analyzed and compared,which highlights the advantages of sparse channel estimation to the conventionalchannel estimation. 2. We propose an improved algorithm CoSaMP. The improved algorithm drew on thethoughts of Subspace Pursuit algorithm, added a least square estimation process by theend of the iteration process, given the estimation results a secondary screening, whichmake the final result more accurate. Moreover it optimized the number of atoms incandidate set and the stop conditions of iterations. Simulation results show, in OFDMsystem, the improved algorithm gains better channel estimation performance fasterestimation speed than CoSaMP algorithm. 3. We introduced the MMP(Multipath Matching Pursuit) algorithm and its simplifiedalgorithm. MMP algorithm applies the idea of combinatorial to sparse channelestimation, and it also combines the greedy algorithm. So MMP algorithm can trackmultipath simultaneously, resist noise interference effectively. The MMP algorithm canalso theoretically ensure the accuracy of the estimation results, overcome the defects ofgreedy algorithms which are sensitive to noise and not having high estimation stability.Simulation results show that in the MIMO-OFDM system, the mean square errorperformance of the MMP algorithm and its simplified algorithm is better than severaltypical greedy algorithms.