The Research Of Channel Estimation And Signal Detection Algorithm In MIMO-OFDM System

In the MIMO-OFDM system, wireless communication environment is unusually harsh and time-varying greatly, and with the rapid movement of a mobile station, the relative displacement between the transmitter and mobile station will make the receiving capability of the receiving end further deterioration, then the error will be generated in the channel, thereby the reliability of the transmission signal will be affected, so it is necessary to estimate the channel and to compensate the error parameters of a receiving signal at the receiver. In addition, the performance of receiver signal detection algorithm and its complexity is directly related to the quality of the communication system. However, the signal detection algorithm with excellent performance of detection is often accompanied by high algorithm complexity. And the signal detection algorithm with low complexity is often accompanied by the inefficient signal detection performance. So the research of signal detection algorithm which is better performance of signal detection and moderate algorithm complexity is the key to good performance of the MIMO-OFDM system to realize.For fast time-varying system, an improved channel estimation algorithm based on compressed sensing is put forward in this thesis. The algorithm combine channel estimation algorithm based on compressed sensing and channel estimation algorithm based on base extended model effectively, make channel estimation in the fast time-varying channel environment better. At first, in this thesis, according to the theory of compressed sensing, analyzes the channel model of fast time-varying systems, which is typical sparse. Then make the compressed sensing theory application to this kind of channel, which can ideally obtain the position information of pilot, however, what the most important of algorithm based on the base extended model is to obtain pilot position information. Finally, obtained pilot position matrix is applied to the base extended model algorithm, and the channel is estimated under the condition of without increasing the complexity of algorithm accurately.For the application of dimension reduction technique in signal detection algorithm, and the combination of dimension reduction technology advantages and sphere decoding detection characteristics, this thesis presents the SD signal detection algorithm based on the lattice reduction. The dimension reduction technique is applied to SD algorithm in this algorithm, and the signal detection and recovery are achieved in the lower algorithm complex. The key step in the SD algorithm is to determine the initial ball radius, the radius is smaller, the searching process is shorter, and the cost time is less, so the complexity of the detection algorithm is lower. However, the initial searching radius RSD of ball is associatedwith an important parameter rik, if the parameter is appropriate, RSD will be reduced heavily, then the amount of calculation in the SD search process will be reduced greatly, thus the signal will be detected more easily. And the signal detection based on lattice reduction can transformed the Q, R and T into QLLL , RLLL and TLLL by the lattice reduction algorithm, whose values are smaller condition number than before transformed values. So you can use the elements rik(LLL) in the QLLL matrix instead of rik of the traditional SD algorithm, and then determine the initial search radius of the ball. Finally detect and recovery the signal in accordance with the general steps of the SD algorithm.