On The Receiver Of Layered MIMO System

The multi-input multi-output (MIMO) system, which is firstly presented by Bell laboratory, is one of the landmark technologies in communication. Aiming at increasing the system transmission rate, the layered MIMO can break the transmission bottleneck of the current system by using advanced error-correcting code and signal processing technology, and becomes the mainstream of MIMO technology. This dissertation studies the receiver of the layered MIMO system and focuses on its several key problems, i.e. the simulation of wireless fading channel, the computation of extrinsic information transfer (EXIT) function, low complexity iterative processing algorithm and construction method of the LDPC code. The author's major contributions are as follows:1.To overcome the fault that the Jakes' fading channel simulator can't produce multiple uncorrelated channels, a sum-of-sinusoids method based on dividing each major-path into 4 sub-paths is presented to simulate wireless fading channel. This method sets the parameters as follows:the parameters of major-path are produced randomly and statistical uncorrelated with each other, and the 4 sub-paths associated with a major-path possess fixed parameter configuration. The simulation method with this configuration can produce uncorrelated channels that satisfy the Clarke reference model with few numbers of sinusoid.2.Two methods for calculating the EXIT function, i.e. the histogram method and the direct average method, are analyzed in detail, where the application scope of the direct average method is also given. To solve the problem that the precision is affected by the bin width of histogram, a data-based optimal histogram is adopted to improve the available method. The optimal histogram can adaptively adjust bin width according to the log-likelihood ratio, and then results in a precise EXIT function. Moreover, this improved method simplifies the computation procedure since it incorporates the histogram operation and the EXIT function calculation without manual selection of parameters.3.To reduce the computation complexity of the list sequential (LISS) algorithm, a selective extended LISS (SE-LISS) algorithm is presented. The SE-LISS algorithm selects some points with high reliability from the constellation using a-priori information to extend the detection tree. Compared with the existed LISS algorithm, its complexity is dramatically reduced since the extended points are greatly reduced. In addition, by fully exploiting the a-priori information, the high reliable points are used for extending the detection tree, so that the computation complexity of the detection method is obviously reduced with little performance loss.4. The EXIT chart is used as the main tool to analyze the extrinsic information transfer characteristics of MIMO detector and error corrected decoder, both of which are key components of the iterative receiver of layered MIMO system. Based on the results of EXIT chart analysis, the hybrid detection and internal iterative decoding of turbo code are optimized. The optimized method can reduce the computation of the whole iterative receiver system by 1/3 at the cost of BER loss 0.1-0.3 dB.5.Two fast, balanced search tree based implementations of the progressive edge-growth (PEG) algorithm are presented. Compared with the indicator based method, the balanced tree based methods trace the dynamic change of the check points' layer structure during the tree extension of the Tanner graph. In this way, the tree extension operation of the Tanner graph is needed only when the first edge linked to the root node is set up, so the computation complexity is greatly reduced. The second method, which adopts an array of balanced tree, reduces the complexity further by decouple the sorting according to the distance from the one according to the degree. Both of these two balanced tree based implementations of PEG algorithm can fast construct long code-length random LDPC.