Research On Low Complexity Signal Detection Algorithms For MIMO Systems

In wireless communication systems, the use of multiple antennas atboth ends of the wireless link enables to transmit multiple data streams con-currently within the same frequency band. This key method is known asMultiple-Input Multiple-Output (MIMO) technology. MIMO technology cansignificantly improve the channel capability, spectral efficiency and link re-liability of wireless communication without requiring additional channelbandwidth and transmit power. MIMO detection is known as one of keytechnologies to meet the demands of higher data rates and improved quality ofservice of wireless communication systems, which detection performancelargely affects the overall performance of MIMO systems.This thesis focus on the low complexity signal detection algorithms of MIMOsystems. Base on massively related research works of MIMO detection technique inboth domestic and overseas, the Soft-Input Soft-Output (SISO) Minimum MeanSquare Error (MMSE) combined with Parallel Interference Cancellation (PIC) itera-tive detection algorithm and K-best sphere decoding algorithm are detailedly ana-lyzed. We demonstrate their drawbacks and propose three improved algorithms. Thethree improved schemes are as follows:(1) Gray-mapping Look-up Table (GLT) algorithm. This method simplifies thecalculations of posterior Log Likelihood Ratio (LLR) of code bits for SISO MMSE-PIC iterative detection algorithm. The GLT algorithm efficiently calculates out theposterior LLR by substituting the real part and imaginary part of detecting symbolinto a pre-given formula respectively. It is not necessary to carry out the minimiza-tion searches and the calculation of Euclidean distances, which is required in the tra-ditional method of directly calculating the posterior LLRs, so the computationalcomplexity is reduced.(2) Bit-sort (BS) K-best algorithm. This method aims at simplifying the sortingprocess of path candidates in K-best algorithm. It quickly finds out the shortestpaths by successively searching the bits of accumulated weight of every path. The BS K-best algorithm significantly reduces detection complexity and has no performancedegradation compared to that of traditional K-best algorithm.(3) Dynamic bit-sort (DBS) K-best algorithm. This algorithm is a further simpli-fication of BS K-best algorithm, which further reduces about50%searching andcounting complexity by finding dynamic shortest paths.The simulation results and complexity analysis indicate that all these three im-proved algorithms can greatly reduce the computational complexity of traditionalmethods without any performance loss in the same time. It will be great beneficial tothe practical implementation of MIMO detection technique.