Low complexity estimation of multiple frequency offsets using optimized training signals

We consider training-aided low complexity estimation of multiple carrier frequency offsets (CFOs) in OFDM-based wireless systems. Typically multiple CFOs incur a loss of orthogonality between the transmitted training sequences, either causing a severe error floor in the performance of an estimator or requiring estimation algorithms of higher complexity. In this paper we introduce design criteria for training signals that allow estimators for multiple CFOs to be decoupled nearly free of such multiple access interference. We propose three different approaches to training signal design fulfilling the presented criteria. Consequently, even estimators developed for the single CFO case may be applied. Our simulations show that the proposed approach provides excellent estimator performance and at the same time allows low complexity estimators to be used.