Research And Improvement Of Blind Equalization Algorithm For Multi - Amplitude Modulation Signals

Communications technology is increasingly develop ing towards high reliability and high capacity. Blind equalization is an important technology to improve the reliability of the communication system. It mainly solves the Inter-Symbol Interference(ISI) caused by the multipath. And in order to obtain a high channel capacity, high order modulation signal is becoming an inevitable trend. Among the high order signals, Quadrature Amplitude Modulation(QAM) and Amplitude Phase Shift Keying(APSK) have been widely used. Thus it’s significant to research the blind equalization algorithms of these signals. Both Q AM signal and APSK signal have multiple amplitude values, so that the commonly used blind equalization algorithms which use constant amplitude value to construct cost function have slow convergence rate and large residual ISI. Therefore, the research in this thesis is in order to solve these problems, and the main contents are as follows:(1)In this thesis, the dual- mode equalization algorithms suitable for QAM signal and APSK signal are modified. Dual- mode algorithm consists of two algorithms that each o f them has its specific advantage. For QAM signal, the first mode algorithm applies the cost function which combines the cost functions of Multiple-Modulus Algorithms(MMA) and Direct Decision Least Mean Square(DDLMS), and DDLMS is used as the second mode algorithm. For APSK signal, the first mode algorithm applies the cost function which combines the cost functions of Constant Modulus Algorithms(CMA) and Decision Adjusted modulus algorithm(DAMA), and DAMA is used as the second mode algorithm. Besides, the effects of the equalization tap-length and decision threshold on the performance of the blind equalization are analyzed. Based on the results of the analysis, the variable decision threshold dual- mode equalization algorithm is proposed. Compared with the algorithm used constant decision threshold, the variable decision threshold algorithm can obtain both fast convergence rate and stable performance.(2) The blind equalization algorithms based on Per-Survivor Processing(PSP) is researched and modified in this thesis. The modified algorithm has stable convergence for high order QAM and APSK signal. Furthermore, the effects of the equalization tap-length and step size on the p erformance of the blind equalization are analyzed. Based on the results of the analysis, the variable step size is used to further improve the algorithm performance. The equalization algorithm use s variable step size has the same residual ISI with the algorithm uses constant step size, but its convergence rate becomes faster.