Research Of Doubly Selective Channel Estimation Based On Compressed Sensing In OFDM Systems

The performance of the wireless communication system is greatly constrained by thewireless channel. As the wireless signal transmission suffers from all kinds of complexterrain, the receiver signal has different degrees of distortion in terms of amplitude, phaseand frequency. Although Orthogonal Frequency Division Multiplex (OFDM) technologycan overcome the frequency selective fading, it is sensitive to the frequency shift.Therefore, channel estimation is very important. Usually, the traditional channelestimation methods on OFDM systems assumes that the channel has so much multi-paththat lots of pilot symbols can be available to obtain accurate channel state information,however, this estimation greatly reduces rate of frequency spectrum resource utilization.In order to resolve this problem, the thesis researches on channel estimation algorithms inOFDM systems based on the compressed sensing theory.According to the three key technologies in compressed sensing, this thesisinvestigated the sparse channel estimation in OFDM system on three levels: the sparserepresentation of channel coefficients, the designing of the pilot symbols and the selectionof reconstruction algorithm. Moreover, aiming at some problems of the existing sparsechannel estimation algorithms, the thesis gives the corresponding resolutions.As the existing sparse channel estimation algorithms only consider the frequencyselective fading channel, the thesis also considers the time selective fading channel. In theactual system, the delay and Doppler frequency shift usually cannot be integer sampled,and it will cause the problem of energy leakage which greatly reduces the channel sparsity.Aimed to this problem, a redundant dictionary of high discrete accuracy is used to replacethe discrete Fourier orthogonal basis to improve the channel coefficients sparsity atdictionary domain. Meanwhile, the measurements required for the reconstructionalgorithm, the pilot number, is reduced. The simulation results show that no matter it issingle antenna or multiple antenna channels, although the redundant dictionary algorithmincreases the computational complexity, it improves the channel estimation accuracy andreduces the requirement for the pilot number.As the existing multiple antenna sparse channel estimation algorithms simply dividethe multiple antenna channels into multiple single channels, by analyzing the joint sparsecharacteristics of multiple antenna channels, the thesis uses distributed compressed sensing theory to realize joint channel estimation. The simulation results show that thejoint sparsity channel estimation algorithm uses the cross-correlation and correlationcharacteristics between channels to make the joint sparsity support set more accurate, andits performance is better than the channel estimation algorithm based on single aparsity.