Tracking of dispersive DS-CDMA channels: an AR-embedded modified interacting multiple-model approach

In this paper, we consider the channel tracking problem for a direct-sequence code division multiple-access (DS-CDMA) communication under Gaussian noise. This problem is important in the design of a receiver over multipath time-varying fading channels. The main contribution of this paper is two-fold. The first one is a multiple-model framework for modeling the channel, composed of a set of known first- and second-order auto-regressive models (AR(i), i = 1,2) with the transitions between models obeying a Markov chain. Each AR model assumes a velocity sub-range of mobile users, and the set covers the range of possible dynamics of the time-varying fading channel. The second contribution is the adaptation of the known interacting multiple-model (EMM) estimation algorithm for channel tracking based on the above modeling framework. This tracking solution in combination with Kalman-based symbol detector leads to a feasible receiver architecture for attaining satisfactory performance