Fast convergent blind equalization for wireless communications

The least-mean-square Constant Modulus Algorithm (CMA) is the most widely studied blind equalization algorithm in the literature and the most widely used in the telecommunication applications. It does not require the transmission of training sequence which saves considerable bandwidth. It also offers the advantage of decoupling the ISI equalization and carrier recovery problems from each other, since the CMA equalizer adaptation does not require carrier phase recovery. However, the slow convergence rate and lacking the ability to track a varying channel prevent it from being used in real time applications of mobile radio communications.; The traditional CMA objective function is optimized on a sample by sample base. In this research, we develop a block CMA objective function, which is found to have two global minima over a very large region of the entire vector space. Based on the block CMA objective function, two rapidly convergent constant modulus algorithms, namely the Block Shanno Constant Modulus Algorithm (BSCMA) and the Block Stochastic Shanno Constant Modulus Algorithm (BSSCMA) are developed. And the convergence theorems are provided for both algorithms. We also develop the Block Conjugate Gradient CMA algorithm which is a slight variation of the BSSCMA.; In our simulation, a multipath fading channel model is used to generate a {dollar}pi/4{dollar}-DQPSK signal. Various algorithms including the LMSCMA, the Recursive CMA, BSCMA, BSSCMA and the Block Conjugate Gradient Algorithm are implemented to equalize the simulated wireless signal. The simulation result shows that the BSCMA and BSSCMA have fast convergence rates and are more robust to the variation of a wireless channel comparing to the conventional CMA algorithms.