Research On Key Technology Of Sparse Channel Estimation

With the rise and development of high-speed communication such as Underwater AcousticCommunication, increasing wireless communication channels appear to be sparse, as a result, itbecomes the hot and difficult point to research the sparse channels for the stable and reliablewireless communication. And it has not achieved the expected goals when traditional methodswhich perform superiorly and are also maturely used are adopted to estimate these sparsechannels. The reason is that these methods ignore the influence of the sparse characteristics ofthe channels. To start with the estimation of sparsity and the detection of most significanttaps(MST), the thesis solves the mentioned problem to some extent, and it also provides somethought for the forward research. The main work and research achievements include thefollowing sections:First, as the sparsity is so important for estimating the sparse channel that the thesis realizesthe estimation of sparsity. Beginning with the number of actual paths which determine the valueof the sparsity, it analyses the relationship among the residual error of Matching Pursuit(MP)algorithm, the iterations of the algorithm and the number of actual paths. Then, it is came up toestimate the sparsity of channels that the method which compares the differences between twoadjacent residual-error-power or the cross correlation coefficients between the residual-error andthe signal. And when the estimated conditions is blind, it is conducted by the subspace algorithm.Second, with the guidance of MP algorithm, the thesis raises the method of sparse channelestimation based on nonzero taps detection. The solutions contain training-sequence aided andnon-training-sequence aided. It might transfer to estimate a {0,1}Lsequence with On-off-keyingalgorithm and MAP algorithm. And when there is no training sequence aided, it is based onexploiting the connection between the autocorrelation of received signal and the position ofnonzero taps. After that, the Sparse Least Squares algorithm is adopted to estimate the nonzerotaps which are detected just now, therefore, the sparse channel is finally estimated.Then, as for the problem that estimats the sparse channel without nonzero taps detecting,the thesis adopts the method of transform domain and the restraint of sparse characteristics basedon LMS adaptive filtering algorithm. With the help of frequency domain least mean squaresalgorithm, it can eliminate effects of the sparse charateristics of the channel. And it all dependson the restraint such as bound norm to restrain the sparse characteristics of the channel when themethod of restraint of sparse characteristics is used.At last, the thesis summarizes the main research achievements by epitomizing the wholeresearch contents, and it also points some problems there are still to be resolved, and prospects the future work.