Study On Blind(Semi-blind) Equalization Algorithms Of QAM Signal

In wireless communication scenarios, transmission effects such as multipath propagation and limited bandwidth cause inter-symbol interference(ISI) at the receiver output. What’s more, the received signals are often corrupted by cochannel interference(CCI) in the multiple-input multiple-output(MIMO) systems. Channel equalizationis a good technique to compensate these distorting effects,which can effectively improve the service quality of communication system.Conventional equalizers generally need transmitting a predetermined training sequence periodically to adjust its taps. However, a training sequence occupies much spectrum resource, which may seriously affect the transmission efficiency of a communication system. Semi-blind equalization algorithm is a compromise of the algorithm based on training sequence and blind equalization algorithm. Semi-blind equalizer(SBE) requaires a small number of training symbols, which not only preserves the accuracy and simplicity of training sequence based algorithms, but also improves the transmission efficiency in a certain extent. Without any training sequence, blind equalization algorithm(BEA) compensates the system distortion on the basis of the apriori information of transmitted signal or the channel. The BEA efficiently improves the transmission efficiency. What’s more important, it is the sole method to compensate the system distortion. Today, in the condition of shortage of spectrum resources, blind(semi-blind) equalization algorithm can make best use of spectrum resources. What’s more, MIMO systems can significantly improve transmission efficiency by only utilizing the same spectrum resources as SISO systems. However, CCI is bringed about by MIMO systems and it increases the difficulty of compensation of channel distortion. On the other hand, quadrature amplitude modulation(QAM) schemes have become popular in numerous wireless network standards for better use of spectrum resources, because QAM signals have high availability of spectrum, good performance of transmission, and relative simplicity in modulation and demodulation. Owing to the aforementioned advantages, MIMO communication systems employing high throughput QAM signals have found widespread applications. Consequently, there is a pressing demand to design an efficient equalizer for MIMO systems.The main work of this paper include the following aspects:1 、 The misadjustments of property restoral adaptive blind equalization approach is analysed by using the constellation information of QAM signals. We introduces a two stage blind equalization algorithm for high order QAM systems. In the first stage, MMA based on least square method(LSM) is used for rough equalization. In the second stage, an adaptively selected decision region method is used for Decision-Directed Algorithm(DDA) when the error level is reasonably low, and then the Improved Soft Decision-Directed Algorithm(ISDDA) based on LSM is used for fine equalization. The former confirms the convergence performance and the latter improves the equalization accuracy and avoids the misadjustment of Consrtant Modulus algorithm(CMA)(and multimodulus algorithm, MMA) in the steady state. What’s more, the proposed modified Least Square Method(LSM), is used to optimize the cost function. The correlation matrix(and its corresponding inversion matrix) is constant with the iteration which reduce the computation load noticeably comparing with general Newton-type methods. What’s more, we prove the approximately quadratic convergence of the proposed LSM based ISDDA in the second stage.2. A space-time semi-blind equalizer(ST-SBE) is proposed for dispersive multiple-input multiple-output(MIMO) communication systems that employ high throughput quadrature amplitude modulation(QAM) signals. A novel cost function(CF) that integrates multimodulus algorithm(MMA) with soft decision-directed scheme is established, which avoids DDA, accurately matchs constellations of the transmitted signal and improves equalization precision. An modified Newton method(MNM) for opertimizing the established cost function is proposed. The proposed MNM not only preserves the quadratic order of convergence as Newton methods, but also is always stable unlike Newton methods since it adopts the positive definite modified Hessian matrix. The proposed MNM has a much lower computational load than Newton methods because its modified Hessian matrix and the corresponding inversion are invariant with the iteration number. In addition, the proposed method needs a small number of training data and samples to ensure correct convergence and to be close to the solution of an optimal Minimum Mean Square Error(MMSE) equalizer, and possesses a good anti-noise ability and less channel maximum distortion(MCD).3. A blind equalization algorithm is proposed for MIMO communication systems that employ QAM signals. Firstly, a good initial value for the equalizer is derived by the time diversity of equalizer, which ensures the local minmums are avioded and solves the delay ambiguity. Then, an Improved MMA(IMMA) is proposed to find the optimal equalizers corresponding to every input signal serially. This mehod efficiently accelerates the convergance speed, avoids the misadjustments of conventional equalzation methods and improves the equalization performance. What’s more, the channel impulse response(CIR) is perfectly estimated when separate the recovered signals from the received signals. Moreover, the novel MNM is employed to fast search the optimal equalizer. Theoretical analysis are provided to illustrate that the proposed algorithm significantly reduces the computational load compared with general Newton method for its fixed Hessian matrix and the quadratic rate of convergence of the MNM is proved.