Training-based frequency synchronisation and highly frequency selective channel estimation for OFDM/OQAM systems

This work addresses frequency synchronisation and channel estimation for highly frequency selective orthogonal frequency division multiplexed systems with offset quadrature amplitude modulation. A two-step approach is proposed, wherein a coarse carrier frequency offset (CFO) estimator, which does not require channel knowledge, is designed in the first step by exploiting the correlation between the received and training signals. A time-domain model is derived for the joint estimation of fine CFO (FCFO) and highly frequency selective channel in the second step. In contrast to the frequency domain model in the literature, the derived time-domain model does not require the channel to be frequency flat. Based on the time-domain model, the weighted least squares and minimum mean square error estimators are investigated to jointly estimate the FCFO and frequency selective channel. The Cramer-Rao lower bounds (CRLBs) are derived for the proposed estimators, and are used to evaluate the performance of the joint FCFO and channel estimation in the presence of inter-symbol-interference due to the overlapping of adjacent time-domain orthogonal frequency division multiplexing/offset quadrature amplitude modulation (OFDM/OQAM) symbols. Simulation results demonstrate achievability of the CRLB bounds and performance gain of the proposed estimators over the state-of-the-art existing estimators.