A Framework for the Estimation of Time-Variant Channels in OFDM

OFDM combines the advantages of high achievable rates and relatively easy implementation. However, for proper recovery of the input, the OFDM receiver needs accurate channel information. In this talk, we propose an algorithm for channel (and data) recovery in OFDM transmission over time-variant environments.

The algorithm makes use of the rich structure of the underlying communication problem– a structure induced by the data and channel constraints. These constraints include pilots, the cyclic prefix, space-time code, and the finite alphabet constraints on the data. The constraints also include sparsity, finite delay spread, and the statistical ((frequency and time correlation) and spatial correlation of the channel. The algorithm boils down to a forward-backward (FB) Kalman filter. We also suggest a suboptimal modification (essentially, a forward-only Kalman) that is able to track the channel and recover the data with no latency

We finally introduce two recent extensions of our framework. We present an implementation of the algorithm in the frequency-domain that helps reduce computational complexity. We also demonstrate how the algorithm can be made robust against uncertainties that result from imperfect channel estimates and from Inter-carrier interference.