Subspace based carrier frequency offset estimations for OFDM systems

Abstract

In this thesis, subspace based carrier frequency offset (CFO) estimation techniques are exploited for Orthogonal Frequency Division Multiplexing (OFDM) systems. The accuracy of a super resolution subspace-based method has motivated this work. We proposed and analyzed the estimation methods for conventional OFDM and OFDM based wireless local area network (WLAN) systems. For the conventional OFDM system, matrix pencil method, which has the accuracy of a super resolution subspace-based method, can estimate the CFO effectively by using only one OFDM symbol with equispacing. For the OFDM based WLAN, the signal model of the preamble section is derived as the signal model of the direction of arrival (DOA) estimation in the uniform linear array signal processing so that to exploit the plethora of DOA estimation algorithms for CFO estimation. Thus, one of the subspace-based DOA estimations, called MUSIC, is applied to estimate the CFO. A first order perturbation analysis for this method is also derived. Since the signal model of the preamble section is identical to an uniform linear array signal model, maximum entropy method also presents a useful method for CFO estimation. Next, extensive computer simulations have been conducted to validate the MUSIC and the maximum entropy methods.