A novel RMS delay-spread estimation technique for wireless OFDM systems

This paper presents a novel RMS delay-spread estimation technique for wireless OFDM systems in exponentially decaying multipath power-delay channels. The technique utilizes a correlation function evaluated over the cyclic-prefix of the OFDM signal. The RMS delay-spread is estimated by means of a minimum mean-squared-error (MMSE) fitting between the observed correlation function and its theoretical expectation. Estimation of the symbol timing and frequency offset parameters is also inherent in the technique. Simulation results demonstrate the accuracy of the technique for the cases of (i) with perfect synchronization, and (ii) with synchronization error. In particular, it is shown that an RMS estimation error as low as 1-2 samples can be achieved for RMS delay-spread, for high channel SNR values (>15 dB). Pilot-symbol-assisted OFDM channel estimation techniques, which requires the RMS delay-spread of the wireless channel for optimal frequency domain channel response interpolation, can be benefited by the proposed technique.