Spectral Efficiency of OFDM Systems With Random Residual CFO

Orthogonal frequency division multiplexing (OFDM) over wireless channels is sensitive to carrier frequency offset (CFO), which destroys orthogonality amongst sub-carriers, giving rise to inter-carrier interference (ICI). Different techniques are available for estimating and compensating for the CFO at the receiver. However, in practice, a residual CFO remains at the receiver after CFO estimation, where the estimation accuracy depends primarily on the fractions of time and power used by the estimator. In this paper, we propose to measure the efficiency of OFDM systems with CFO estimation errors in terms of the spectral efficiency, which accounts for both, the degradation in signal-to-interference plus noise ratio (SINR) due to the residual CFO, and the penalty of the extra power and spectral resources allocated to achieve the desired CFO estimation accuracy. New accurate expressions are derived for the spectral efficiency of wireless OFDM systems in the presence of residual CFO and frequency-selective multipath fading channel. These are used to compare between two common CFO estimation methods in wireless OFDM systems, namely, the cyclic prefix based and the training symbols based CFO estimation techniques for fixed and variable pilot power. These results are further extended to include OFDM systems with transmit diversity techniques. In addition, the impact of imperfect channel estimation on the overall spectral efficiency is also included. Numerical results reveal that the cyclic prefix based CFO technique is more efficient than the training symbols based CFO technique when perfect channel state information (CSI) is known blindly at the receiver. Furthermore, fixed pilot power results in a spectral efficiency ceiling as SNR increases, whereas spectral efficiency increases with SNR without bound in the equal pilot and signal powers case.