Research On The Channel Estimation Algorithm In MIMO-OFDM System

This thesis focuses on the communication technology beyond3G. The research was carried out for the channel estimation technique in the multi-input multi-output (MIMO) and orthogonal frequency division multiplex (OFDM) environment. Due to the particularity of the channels in the MIMO-OFDM wireless communication systems, the channel estimation technique is the key technique that can directly affect the system performance. Whether or not the channels can be correctly or approximately estimated can directly influence the correctness that the decoder can demodulate the transmission signal and the robustness and effectiveness of the whole system. Thus, the research on the channel estimation algorithms in the MIMO-OFDM system has a very important academic and industry value. A lot of research on this area has already been conducted in China and abroad.In the existing channel estimation algorithms for MIMO-OFDM systems, the blind channel estimation algorithm has a very high computational complexity and poor performance, while the estimation algorithms based on pilot signals have a small computational complexity and high accuracy. Thus, in this thesis, the research is only carried out for the pilot signal assisted estimation algorithms. In the research for the existing algorithms,, the characteristics of the existing pilot signal assisted estimation algorithms are analyzed, including the design of the pilot signal insertion, the estimation algorithm for the pilot channel, and the interpolation algorithm for the data channels. Since in the current pilot signal assisted estimation algorithm, the insertions of the pilot signals take some channels which can be used to transmit the data, it reduces the information transmission rate and waste some system resource. Thus, in this thesis, it will optimize and revise these algorithms in order to boost its performance and overcome its shortcomings. To achieve this purpose, this thesis carries out the following creative research.1) In the current pilot signal structure of the OFDM channel estimation algorithms, the pilot symbols are repeatedly inserted in the same location of each OFDM symbol (the comb structure), or are periodically inserted in the every sub-carrier of a OFDM symbol (block pilot structure), to achieve one round of channel estimation. In this thesis, it proposes a new cross pilot structure, i.e. in each OFDM frame, the pilot signals are inserted in different locations with the same distance. Compared with the classical block pilot structure and the comb pilot structure, this pilot structure can better trace the channel features.2) It proposes a new channel estimation algorithm that corresponds to the cross pilot signals, i.e., in the estimation of a data channel, it selects the neighboring pilot channels to get the weighted average of these channel gains. This method not only considers the feature of the channel estimation in the pilot channels, but also utilizes the correlation between neighboring sub-channels. Compared with the classical algorithms of least squares (LS) and least minimum mean square error (LMMSE) algorithm, this algorithm has a better accuracy under the condition of the same number of pilot signals.3) According to the channel features of the MIMO-OFDM wireless communication system, multiple transmission antennas and receive antennas consist of multiple parallel antenna transmission channel sets. This thesis effectively considers the increase of the channel capacity and channel interference which is brought by the multiple channels. Based on this, it proposes a new scheme of new zero-padding orthogonal pilot structure to reduce the interference of antenna channels, enable the receiver end to correctly distinguish the pilot data from different transmission antennas, so that the LS and LMMMSE algorithms in the single antenna environment can be revised to directly apply to and implement the MIMO-OFDM wireless communication systems. This method not only can effectively enhance the performance of the channel estimation of the whole system without the increase of the system bandwidth, but also is easy to implement.