Autoregressive Stochastic Modeling and Tracking of Doubly Selective Fading Channels

A.D. Abu Al-khair and M.M. Banat (Jordan)


Frequency selective fading, time selective fading, AR modeling, Kalman filter, channel tracking. Our channel modeling and tracking algorithm in this paper is based on the more general statistical approach to describe the channel tap coefficients. The time evolution of each tap coefficient will be modeled as an autoregressive (AR) random process. AR modeling of time varying channel tap coefficients was first proposed in [7].


A popular approach to incorporate time variations into the delay line model is to represent the tap coefficients using deterministic time varying basis expansion fucntions (e.g., using complex exponentials) [3], [4]. This sort of modeling is particularly useful when the multipath is mainly caused by a few strong reflectors and when path delays exhibit variations due to the kinematics of the mobiles [3], [4], [5]. This paper presents a new autoregressive (AR) stochastic modeling and tracking method of doubly selective fading channels. The AR model is used to generate a wide-sense stationary uncorrelated scattering (WSSUS) channel impulse response. A Kalman-based tracking algorithm is then designed to continuously track the channel at the receiver. A more general approach to describe time-varying communication channels is by treating the delay line coefficients as lowpass Gaussian uncorrelated stationary random processes [1], [2], [6]. This approach is suitable for situations where large a number of scatterers exist.

Important Links:

Go Back