Channel estimation for PRP-OFDM in slowly time-varying channel: first-order or second-order statistics?

The statistics-based channel estimators can estimate the channel state information (CSI) in the static channel but can only obtain the averaged CSI (ACSI) in the time-varying (TV) channel. This letter investigates both first-order statistics (FOS) and second-order statistics (SOS)-based channel estimators for the pseudo random postfix (PRP) orthogonal frequency-division multiplexing system in the slowly TV channel. It is shown that the FOS-based approach outperforms the SOS-based approaches in the ACSI estimation. Using estimated ACSIs for the channel equalization, simulation results indicate that the SOS-based approaches converge to the FOS-based approach in the high signal-to-noise ratio range.

I. Introduction

Orthogonal frequency-division multiplexing (OFDM) modulator usually employs cyclic prefix (CP) or zero padding (ZP) to mitigate the interblock interference (IBI) [1]. These are called CP-OFDM and ZP-OFDM systems, respectively (see [2]). Recent literatures (e.g., [3]) suggested to replace the zero sequence in ZP-OFDM with a number of known symbols and named the proposed system pseudo random postfix (PRP) OFDM. The major advantage of using PRP technique is enabling the low-complexity first-order statistics (FOS)-based estimation of (quasi-) static channels. In the multiple-input multiple-output (MIMO)-OFDM system, employing distinct PRPs at transmit antennas makes the MIMO channel estimation much easier [4].