Channel estimation in OFDM: Interpolation versus ESPRIT

Use of Orthogonal Frequency Division Multiplexing (OFDM) has been popular, especially in wireless channels, due to its ability to perform frequency-domain channel equalization. Estimating channel frequency response is essential to successful data recovery. In this paper, we compare traditional estimation methods using interpolation among pilot measurements to a parametric model (PM)-based channel estimation. This PM-based technique uses the ESPRIT (Estimation of Signal Parameters by Rotational Invariance Techniques) method to estimate time delays of multiple paths in the channel. Then a least-squares estimator is used to estimate the channel frequency response. Though this subspace method is computationally more expensive, experimental results show that it performs much better than traditional interpolation techniques.