Time-Varying Channel Identification and Prediction in OFDM Systems using 2-D Frequency Estimation

We consider the long range channel estimation and prediction problem in adaptive OFDM systems. A deterministic frequency- and time-selective channel model, which contains the Doppler shift, delay and amplitude parameters for each path, is applied to describe the channel. It is shown that the parameter identification problem is essentially a 2-D frequency estimation problem. We propose a new joint 2-D frequency estimation algorithm to estimate the parameters in the deterministic channel model. The proposed algorithm possesses several advantages in identifiability, performance, and complexity. The estimated parameters are then used to predict channel variations. The simulation results demonstrate that the proposed algorithm offers better performance than existing channel prediction algorithm based on 1-D frequency estimation