An Adaptive DFT-Based Channel Estimation for OFDM

Abstract

An improved discrete Fourier transform (DFT)-based channel estimation for orthogonal frequency division multiplexing (OFDM) systems is proposed. Conventional DFT-based channel estimation improves the performance by suppressing time domain noise. However, it requires the accurate estimation on channel length for high reliability and suffers from the performance degradation otherwise. In contrast, proposed channel estimation can improve the performance by deciding significant channel taps without knowing any statistics. Significant channel taps are detected based on a predetermined threshold. We calculate the threshold to minimize the mean-squared error (MSE) of estimation, and confirm it via computer simulation. Simulation results demonstrate that proposed algorithm can improve MSE about 6.5dB comparing to conventional DFT-based estimation, and the MSE floor does not occur in any channel environments.