| Compressive sensing(CS)is a new theory that can recover the sparse signal with high probability by a small number of observations in signal processing.Due to the sparse nature of the wireless channels,it can effectively reduce the number of pilots and improve the performance of channel estimation by applying compressive sensing technology to channel estimation for multiple-input and multiple-output orthogonal frequency division multiplexing(MIMO-OFDM)systems.The optimal pilot pattern of traditional linear channel estimation methods is uniform.However,the uniform pilot placement is not appropriate in CS-based channel estimation methods.Therefore,the design of pilot locations and pilot symbols in CSbased channel estimation needs special study.In this dissertation,we focus on the channel estimation and pilot optimization based on CS in MIMO-OFDM systems.The main contributions are as follows:1)By modeling the channel estimation problem in MIMO-OFDM systems as the sparse signal reconstruction problem in CS theory,an improved compressive sampling matching pursuit(CoSaMP)algorithm which selects the atoms by threshold is proposed.The simulation results show that the improved CoSaMP algorithm can achieve better channel estimation in terms of both the mean square error(MSE)and the bit error rate(BER)than the traditional CoSaMP algorithm.2)The pilot design based on CS channel estimation in MIMO-OFDM systems is studied.By minimizing the coherence of measurement matrix,a stochastic search algorithm for joint optimization of pilot locations and pilot power is proposed.The simulation results show that the proposed pilot design algorithm has better channel estimation performance than existing pilot design schemes.3)In order to estimate channels of massive MIMO-OFDM systems,the superimposed pilot design is adopted to reduce the pilot overhead.By employing the structured compressive sensing theory,the block sparsity adaptive matching pursuit(BSAMP)algorithm without the prior information of the channel sparsity is proposed.The simulation results show that the channel estimation methods based on structured CS substantially outperform the traditional CS methods which use individual sparse channel estimation.4)The pilot design based on structured CS in massive MIMO-OFDM systems is studied.By minimizing the total interblock coherence and the total subblock coherence,the genetic algorithm of jointly optimizing the pilot locations and pilot symbols is proposed.The simulation results show that the proposed algorithm has a good performance improvement over random pilot. |
