Research On The Semidefinite Relaxation Method Of Higher-Order QAM In MIMO Systems

Multiple-input multiple-output (MIMO) is one of key techniques for the next-generation mobile communication system, because it can effectively improve the capacity and spectral efficiency of mobile communication system. The quality of a communication system largely depends on the accuracy of the detection signals. As a result, the way to design a high-performance detection algorithm with low complexity for the next-generation mobile communication widely receives great attention.Maximum likelihood (ML) detection algorithm can achieve the best detection performance, however, it has high computational complexity due to the fact that it is a non convex optimization problem. Semidefinite relaxation (SDR) is an attempt to approximate it using a concex program that can be efficiently solved in polynomial time. It will not only reduce the complexity of the detection algorithm, but also get a relatively good detection performance. In this thesis, SDR methods in high-order QAM (Quadrature Amplitude Modulation) MIMO detection are deeply studied.First, several traditional V-BLAST (Vertical-Bell Labs Layered Space-Time)-system-based MIMO detection algorithms are analyzed. Through the analysis of the advantages and disadvantages of traditional algorithms, we find it was necessary to study the SDR method in other means.Second, the state-old-art of SDR method in high-order QAM MIMO detection is introduced. A new high-performance SDR detection algorithm (SDF-SDR, Successive Decision Feedback SDR) with low complexity is presented. The new detection algorithm can achieve sub-optimum performance without high computational complexity even in high-order modulations. This algorithm can make a good tradeoff between complexity and performance.Third, semidefinite relaxation method in the high-order QAM blind detection are analyzed. According to the understanding of the existing blind SDR methods, two equivalent blind SDR methods (LF-PISDR, Linear Fractional Polynomial Inspired SDR,LF-VASDR, Linear Fractional Virtually Antipodal SDR) and a high-performance blind SDR method (LF-TBCSDR, Linear Fractional Tightened Bound Constrained SDR) are derived. The performances are simulated and evaluated.Finally, in the course of the previous derivation, it's found that the high-order QAM blind detection problem can be solved without the pilot symbol. Therefore, a new blind SDR method (LF-EVASDR, Linear Fractional Extend SDR) with high bandwidth-efficiency was presented. The method can achieve high performance without the pilot symbol.