Research On Blind Equalization Algorithm Based On Quadrature Amplitude Modulation

In modern high-speed underwater acoustic communication system, the inter-symbol interference caused by multipath underwater acoustic channel is the main obstacles in high-speed underwater acoustic communication, blind equalization technique does not need training sequence is very suitable for the very limited bandwidth of underwater acoustic channel. Quadrature amplitude modulation (QAM) can get a higher channel band efficiency, has been widely used in high-speed underwater acoustic communication. In this paper, quadrature amplitude modulation system blind equalization algorithm in high-speed underwater acoustic communication is analyzed and the main work is as follows:(1)By studying the principle of MAMA, improved a class algorithm (CMAMA) based on the constellation features. CMAMA algorithm uses a different cost function for the different QAM signal, matches the error model and QAM signal constellation, gets a smaller inter-symbol interference. At the same time, The article also study the phase rotation of32QAM CMAMA algorithm problems and the instability of64QAM CMAMA algorithm instability, and given the improved algorithm.(2)When the channel has serious distortion, CMAMA algorithm not reliable convergence. CMAMA algorithm as the tracking algorithm and combination with multi-mode algorithm, use of mean square error switching, get square QAM hybrid algorithm and cross QAM hybrid algorithm, to make the algorithm convergence in the initial stage can be stabilized, to improve the convergence performance of CMAMA algorithm.(3)For the shortcomings of WMMA algorithm convergence is slow and the high complexity, proposed an variable mode algorithm based on MMA for the16QAM signal, through point judgment adaptively change the algorithm to reduce the computation and speed up the rate of convergence, and through simulation studies different point of decision regions and different switching criteria on the performance of the algorithm.(4)Using related capacity of wavelet transform and excellent small sample learning ability of support vector machine to improve the convergence rate of the WMMA algorithm, gets two fast convergence of the WMMA algorithm.