A Study On Equalization Technique For Nonlinear Channel

In the wireless communication systems, high power amplifier is an important nonlinear component, which would break down the system's transfer characteristic, bring in intersymbol interference. So the nonlinear components would degrade the data transfer speed and the removability of communication system, which should be solved by the receiving terminal's equalizer.In this paper, we introduce the adaptive equalization theory and Volterra filter at the first, then give a brand-new & systemic research about the nonlinear channel equalization. This paper mostly explores new structures and researches new algorithms. Strict formula derivation is another characteristic of this paper.Firstly, we used Hammerstein model and Wiener model in non-linear structure instead of the Volterra series model in order to reduce the computing complexity, and proposed a nonlinear transmission system model constructed by Hammerstein model and Wiener model. Based on the proposed system model, Nonlinear Channel LMS(NCLMS) algorithm, improved algorithm for type 1 NCLMS Newton and improved algorithm for type 2 NCLMS Newton are proposed and derived. The simulation results show that, NCLMS algorithm is slow convergence and instability performance, improved algorithm for type 1 NCLMS Newton overcomes the NCLMS algorithm's shortcomings of slow convergence and improves the stability of the algorithm, improved algorithm for type 2 NCLMS Newton inherits the virtues of improved algorithm for type 1 NCLMS Newton and further improve the convergence speed, which is close to ideal NCLMS Newton algorithm.Secondly, based on the proposed nonlinear channel transmission system, we research other several important algorithms. Three important algorithms which are applicable for time-varying nonlinear channel are proposed and deduced respectively: Nonlinear Channel Recursive Least Squares (NCRLS) algorithm, Nonlinear Channel Kalman (NCKalman) algorithm and Nonlinear Channel Recursive Prediction Error Method (NCRPEM) algorithm, and compared each other. The simulation results show that NCKalman performs best in reducing residual mean square error, NCRPEM is followed by, NCRLS is the worst. In respect of iterative convergence speed, NCRPEM performs best, NCRLS is followed by, NCKalman is the worst. Comprehensive evaluation, NCRPEM performs best, NCKalman is followed by, NCRLS is the worst.Thirdly, in this paper we research the nonlinear channel equalization algorithm based on infinite impulse response (IIR) structure. Based on the proposed nonlinear channel transmission system above, an IIR nonlinear channel transmission system is constructed. Then based on the new system, IIR NCLMS algorithm and IIR NCLMS Newton algorithm are proposed to estimate nonlinear equalization's parameters. The simulation results show that IIR NCLMS Newton performs better than IIR NCLMS in reducing residual mean square error. And the IIR NCLMS Newton's residual mean square error is influenced by SNR(signal to noise ratio). However, it's acceptable within certain limits of SNR.Lastly, in this paper we used Wiener model to construct power amplifier and used Hammerstein model to construct predistortion, and proposed a Nonlinear Filtered LMS(NFLMS) algorithm based on LMS algorithm and predistortion system model. Based on the proposed algorithm, NFLMS Newton algorithm and improved algorithm for NFLMS Newton are proposed and derived. The simulation results show that, improved algorithm for NFLMS Newton has faster convergence speed than the NFLMS algorithm, and brings down the algorithm's residual error more quickly.