Compressed sensing for clipping noise cancellation in DCO-OFDM systems based on observation interference mitigation

In this paper, we propose a modified clipping noise cancellation scheme using compressed sensing (CS) technique with observation interference mitigation for direct current biased optical (DCO) orthogonal frequency division multiplexing (OFDM) systems. The interference components in potential observations are theoretically analyzed and approximately estimated by exploiting the statistical model of the clipped DCO-OFDM signal. Then, the modified CS scheme eliminates the estimated interference from the potential observations, which can minimize the contamination influence of channel noise on CS recovery algorithm. In addition, a strategy jointly considering the compressed ratio and the decision noise is presented to generate the measurement matrix, which can sample the reliable data tones as final observations. With this scheme, the clipping noise in time domain can be effectively corrected and the bit error rate (BER) performance of the system is significantly improved. Simulation results demonstrate that the proposed scheme can perform well even under severe clipping conditions.