The Research Of OFDM Sparse Channel Estimation Method Based On Compressed Sensing

As a kind of efficient multi-carrier modulation scheme, orthogonal frequency divisionmultiplexing (OFDM) technique can effectively overcome the problem of signal fading andinter symbol interference in wireless channel, realize high-speed data transmission, so it hasbeen widely used in the field of wireless communications. In OFDM communication system,channel estimation is a key technologie, it relates directly to the overall performance of thesystem. Accurate and real-time channel estimation is the necessary condition for the datademodulation at the receiver and it is also the important guarantee to receive data reliably.Channel estimation problem has been a research focus in the OFDM system.In recent years,the OFDM system has gradually use compressed sensing (CS) to implement channelestimation.In this thesis, the channels of the wireless communication system and the principle ofOFDM technology are firstly introduced, Based on the analysis of characteristics of wirelesschannel, On this basis, The paper proposes OFDM channel estimation method based onorthogonal matching pursuit (OMP) algorithm, compression sampling matching pursuit(CoSaMP) algorithm and subspace tracking (SP) algorithm respectively combined with thecompressed sensing theory. The simulation results show that compared with the traditionalLS algorithm, the channel estimation method based on compressed sensing can improve thesystem spectrum and the estimation performance efficiency,because it can make full use ofthe sparse characteristics of wireless channel. Moreover, the paper analysis the affect of thenumber of the pilot, pilot insertion method, and priori channel sparse degree on the eachalgorithm performance through simulation respectively, provides certain theoreticalguidance for practical applications.In view of the shortcomings of CoSaMP algorithm, CoSaMP algorithm has differentcriterias in the process of iteration, it causes the problems of slow convergence speed, thispaper puts forward corresponding improvement method,by increasing consideration andchanging the stop of iteration, effectively reduce the possibility of the problem. Thesimulations show that applied the improved algorithm to estimation OFDM channel, canobtain better estimation performance at the cost of high computation complexity, under thecondition of the same SNR, bit error rate and mean square error (MSE) of improvedalgorithm are significantly lower than the CoSaMP and the LS algorithm, improvedalgorithm achieved its aims. In addition, in view of the shortcomings of SP algorithm needthe sparse degree as a prior condition, the paper proposes an improved SP algorithm.through two improvements: gradually increase the selected atomic number in theprocess of iteration and according to the changes of the signal residual before and after thetwo iterations process determine whether or no terminate the iteration, effectively avoid thealgorithm dependence on the priori channel sparse degree. The simulation results show thatthe improved algorithm can significantly improve the estimation accuracy under thecondition of complexity increased slightly, what is more, the improved algorithm do notneed channel sparse degree as the prior condition, compared with the original algorithm, theimproved algorithm has a larger range of application.