Sparsity-aware channel estimation with contaminated pilot sequence

In this paper the up-link channel estimation problem in a multi-cell network with a Single-Input-Multiple-Output (SIMO) fading correlated channel and finite-length, nonorthogonal and contaminated pilots is considered. Our contribution is a novel channel estimation technique, referred to as the Sparse Channel Estimator (SCE), which exploits the Karhunen-Loéve Transform (KLT) sparse-representation of the channel. We show that, by simply utilizing an interference-contaminated channel sample-covariance matrix, the proposed approach outperforms Least Squares (LS)-based and Minimum Mean Squared Error (MMSE) techniques relying on the same type of assumptions.