OFDM Channel Estimation with Optimal Threshold-Based Selection of CIR Samples

In this paper, we address the problem of OFDM data-aided channel estimation based on the selection of the most significant samples (MSS) of the channel impulse response (CIR), i.e. those samples which contain most of the useful energy. We provide a novel and complete analytical characterization for MSS selection based on threshold crossing, which yields a closed form for the estimate mean-square error (MSE), that we use to derive analytically the optimum threshold in the minimum MSE sense. The optimum threshold value is matched to the specific channel power profile, but this information is hardly available to the receiver. For these reason, we also propose a suboptimal method for threshold setting that does not require any knowledge of channel statistics. We show that the performance of this sub-optimal method is very close to the optimum case, as well as to Wiener channel estimation in the case of sparse multipath channels. Our proposed method outperforms previous approaches based on heuristically set thresholds.