Research On Multicarrier System Channel Estimation Based On Compressed Sensing

Channel estimation is one of the key technologies in wireless communication system. Due to limits of Nyquist sampling theorem, it is hard to obtain a higher channel estimation accuracy by conventional channel estimation algorithms. Moreover, traditional methods do not consider sparseness of the channel, leading to higher computational complexity. Considering sparse characteristics of the channel, scholars apply compressed sensing theory to channel estimation,which improves the channel estimation accuracy effectively and reduces computational complexity.This paper revolves around OFDM. Firstly, it analyses the theory of OFDM channel technology based on compressed sensing estimation. Then, CDMA system and OFDM technology are combined to realize MC-CDMA system, channel estimation of that can be analyzed using compressed sensing. At last, the MIMO technology is introduced on the basis of OFDM. Using correlation between respective antennas in MIMO-OFDM system, channel estimation of this system can be done by distributed compressed sensing.To begin with,this thesis studies applications of classical OMP algorithms in OFDM channel estimation and compares OMP algorithms with traditional channel estimation algorithms. After that,it conducts a research on effects of number of pilot on the accuracy of channel estimate. Through simulation, it comes to a conclusion that the number of pilot has great influence on MSE of channel estimation and BER of system when it is less than six times of number of multipath.Then, this paper introduces compressed sensing theory into MC-CDMA system to do channel estimation. Classical OMP algorithm needs sparseness of the channel as a channel estimation prior conditions. However, sparseness of channel is generally unknown in fact. SAMP algorithm this thesis puts can effectively estimate the channel impulse response without sparseness of the channel,fulfilling requirements better. Related simulation proves that SAMP algorithm gets high performance at the cost of certain time complexity.Finally, in MIMO-OFDM communication system, this thesis selects distributed compressed sensing as channel estimation theory. General compressed sensing algorithms do not take correlations between antennas into account. However, distributed compressed sensing theory this paper presents can obtain higher performance of channel estimation. Simulation shows that performance of DCS algorithm is better than that of CoSAMP algorithm based on CS by 5dB.