Adaptive Algorithms for Wireless Channel Estimation<

Accurate channel state information is important in communication systems. This is especially challenging in a wireless environment where the channel exhibits strong frequency and time selectivity. There has been considerable prior work on this problem. We present two contributions to an unified approach to channel estimation problems.

In the first contribution, we study generalized LMS algorithms with different types of error and data non-linearity. We show that we can analyze the performance issues such as steady state error, stability and convergence rates by using a weighted energy relation method. Our technique solves all known generalizations of the LMS algorithm with error or data non-linearity, as well as provides more accurate estimates of the performance.

In the second contribution, we propose joint channel identification and data detection algorithms for OFDM transmission. Our algorithm exploits multiple structures of both the data and channel. Data structures include pilots, cyclic prefix, channel code and finite alphabet. Channel structures include tap sparsity, delay spread and Doppler spread. We show that a Kalman filter framework can be used to provide a unified solution.