Modeling and tracking for WSSUS fading channels

Abstract

In general, mobile fading channels are well represented by linear filters with randomly time variant characteristics, which are usually assumed as wide-sensestationary uncorre- lated scattering(WSSUS). Modeling the WSSUS fading channels can provide a thorough understanding of the underlying radio propagation environment and the effective means of performance test and evaluation of wireless communication systems in laboratories. However,the simulation models in the literature,typically the linear interpolation model, autoregressive model(AR), Clarke’s model and Zheng & Xiao’s sum-of-sinusoids(SOS) model, have some shortcomings in generating WSSUS fading channels. In this thesis, we introduce a new family of SOS models which have better correlation properties than the typical simulation models. We also propose a new gradient based variable forgetting factor recursive least squares algorithm (GVFF-RLS), which can be applied to track the time-varying fading channels effectively. In the study of channel modeling, we show that the linear interpolation model is nonstationary whereas AR model has unavoidable numerical problems. Both of Clarke’s model and Zheng & Xiao’s model can approximate the desired statistical properties. But Clarke’s model needs a number of random variables while Zheng & Xiao’s model does not have good statistical properties for long time delay. We propose a new family of SOS models with the path number N = 4LM which need less random variables and their correlation properties are close to the desiredones. Especially,the SOS models with L ≈ M achieve the best agreement with the desired correlation function, which is better than other proposed SOS models with small L and Zheng & Xiao’s model. Furthermore, the correlation variances of the proposed SOS models are smaller than those of Clarke’s model and Zheng & Xiao’s model. In the study of tracking WSSUS fading channels,the RLS algorithm is appliedbe- cause of its good convergence property and small mean square error in stationary environ- ments. The performance of the RLS algorithm for channel estimation in WSSUS fading channels is analyzed. Unlike most of RLS analyses, our analysis takes the correlation of the inverse of correlation matrix into account and hence yield an improved recursive formula for the mean square error(MSE) and steady-state MSE equations. Based on the steady-state MSE equation,the optimal forgetting factor and minimum MSE expressions are derived by using Jakes’ model, which approximate the theoretical value very well. A new VFF exponential windowed RLS (EW-RLS) is proposed based on the optimal for- getting factor expression. Furthermore, the improved MSE analysis of the time-variable error weighting RLS (TW-RLS) algorithm yields a new control mechanism of the GVFF- RLS algorithm to control the forgetting factor dynamically.The resulting GVFF-RLSal- gorithm can reduce the forgetting factor when large model error is detected and increase the forgetting factor when the model error be comes small. Compared with other variable forgetting factor schemes, the proposed gradient scheme is shown to yield fast tracking and small square model error in time-varying fading channels with Gaussian noise.

Date of Award	3 Oct 2006
Original language	English
Awarding Institution	City University of Hong Kong
Supervisor	Shu Hung LEUNG (Supervisor)