Signal Detection Based On Sphere Decoding In Spatial Modulation System

The traditional multiple input multiple output(MIMO) systems offer a significant increase in spectral efficiency in comparison to single input single output(SISO) systems, however, this comes at the expense of an increase in severe inter channel interference(ICI) and higher decoding complexity at the receiver. Moreover, some of the MIMO technologies are not suitable for asymmetric antenna system. As the wireless communication is on the horns of a dilemma and falls into the circumstances of discomfiture, a novel scheme for MIMO system comes into being, which is named spatial modulation(SM) and has gradually been in a hot. In SM system, the antenna index is brought to transmit additional information to increase the spectral efficiency, and only one antenna is activated to transmit the modulated information at each time instance to avoid the ICI.In order to exploit low-complexity detection algorithms with near-optimal performance, this paper primarily studies on the signal detection in SM system, and mainly focus on the sphere decoding(SD) detector. Firstly, through detailedly theoretical derivation and analysis on the detection process and updating radii of the traditional SD detector, the fundamental cause of high computational complexity lies in previous SD algorithms is found out. It is the large number of redundancy included in the updating radii that cause the computational complexity difficult to decrease. Then, a novel conclusion is drawn by theoretical deduction and simulation analyses, which is the redundancy in updating radii obey the noncentral chi-squared distribution. Finally, a few detection algorithms with low-complexity are proposed in SM and GSM systems, and the main work of this paper is summarized as follows:(1)An SD detector based on the QR factorization and the expectations of chi-squared distribution is proposed in SM system;(2)Another SD detection based on the QR factorization and the maximum of chi-squared distribution is presented in SM system;(3)An improved SD algorithm based on the expectations of chi-squared distribution with no need for QR factorization is introduced in GSM system;(4)An SD detector based on the maximum of chi-squared distribution without QR factorization is extended in GSM system.Lastly, a mass of monte-carlo simulations for the improving SDs proposed above are carried out in SM and GSM systems, respectively. Through theoretical analysis and simulation results, it is shown that the proposed solutions provide a substantial reduction in computational complexity with respect to the conversional SDs and Maximum Likelihood(ML) detection while retaining the same bit error rate(BER) as the ML-optimum detector.