Research On Channel Estimator And Their Performance In Multi-antenna System

In order to satisfy the high-speed transmission of various application requirements, a wireless communication system with a high transform rate is urgently required. The multi antenna wireless system which introduces multi-antenna technology can not only provide high rate transmission advantage, but also contains a huge capacity, so it applies widely in our life soon. However, it also makes the complicated wireless channel environment has become more complex, The interference which is unknown, seriously hindered the accuracy rate of receiving the signal. As we all know, the performance of the communication system is entirely determined by the accuracy of the channel state information(CSI)at the receiver, In most cases, for wireless communication systems, channel information is unknown to the receiver and sender. At present, the approaches which is the most popular and widely used to the channel estimation is to employ pilot signals (which is also called as training sequences), send the training symbols in the sender and then to estimate the channel based on the knowledge of training symbols in receiver, the common estimation methods are Maximum Likelihood(ML) estimator, linear least squares(LS) estimator, minimum mean-square-error(MMSE) estimator.For the multi-input multi-output (MIMO) system with M antennas at base station and a user with K antennas, in this paper, the traditional linear least squares (LS) and minimum mean-square-error (MMSE) estimators are considered. Then propose new scaled LS (SLS) and relaxed MMSE estimators which can offer a substantially improved performance relative to the LS estimator and require less knowledge of the channel than MMSE estimator. Those approaches provide a good tradeoff between the achieved performance and the required channel knowledge. Moreover, the optimal choices of training matrices are investigated for the aforementioned estimators and the channel estimation errors are analyzed. In the case of multiple LS channel estimates, an optimal scheme for their linear combining which greatly reduces the estimation error is developed, the so-called best linear unbiased estimation (BLUE) approach. At last, simulations to confirm these conclusions are shown.For a massive MIMO system, which equipped with a large number of antennas at the base station,this paper propose a simple decoding method, called as "direct decoder" is this thesis.This method so based on fact that the wireless channels from different users to the base station are mutually asymptotic orthogonal when the number of the antennas goes to infinity. Compared to traditional Maximal Ratio Combining(MRC), zero-fringe(ZF) or MMSE decoders, the proposed method dose not need to calculate the inverses of matrices. Thus, the decoding complexity is reduced.Simulations show that the proposed method has comparable performances with the traditional decodes when the number of the antennas goes to large.