MIMO-OFDM系统中改进的QRD-M检测算法

摘要

In signal detection of the MIMO-OFDM system,traditional QRD-M algorithm approaches the performance of ML detection with lower complexity and has good application prospect.However the M value of this algorithm must be large enough,and its computation complexity is higher as well.Aiming at these shortcomings,based on QRD-M algorithm,first the modified Gram-Schmidt orthogonalisation algorithm is applied on QR decomposition,which makes the receiving end be able to detect the layer with larger SNR first so as to reduce the possibility of missing the ML solution.Secondly,in the process of tree search the algorithm is combined with the DFE algorithm and a new parameters T is introduced.In the improved algorithm,the front T levels are detected by M branch search algorithm while the remained other levels are detected by DFE algorithm.This method reduces the complexity of traditional algorithm and meanwhile increases the detection flexibility at the receiving end.Simulation results show that the improved algorithm achieves the performance of approaching the maximum likelihood detection more with lower complexity.%在MIMO-OFDM系统的信号检测中传统的QRD-M算法以较低的复杂度逼近了ML检测的性能，具有很好的应用前景。但是该算法M值必须足够大且计算复杂度较高。针对此缺点，在QRD-M算法的基础上，首先对QR分解采用改进的修正Gram-Schmidt正交化算法，使得接收端能够最先检测信噪比较大的层，从而减少错过ML解的可能性；其次在树搜索过程中与DFE（Deci-sion Feedback Equalization）算法相结合，引入新的参数T。改进算法的前T层用M分支搜索算法检测，剩余的其他层用DFE算法检测。这种方法降低了传统算法复杂度，同时增加了接收端检测的灵活性。仿真结果显示，改进的算法以更低的复杂度获得更接近最大似然检测的性能。