Mutual Information and Coded BER Analysis of PAPR Reduced OFDM System with Active Constellation Extension

Active constellation extension (ACE) has been proposed for a peak-to-average power (PAPR) reduction of orthogonal frequency- division multiplexing (OFDM) signals, which projects the clipping noise generated by iterative clipping and filtering (CAF) onto the outside region of the constellation such that its minimum Euclidean distance (MED) is not reduced. Due to this arrangement, some amount of average power increase is introduced, but ACE can achieve better trade-off between uncoded bit-error rate (BER) and power amplifier (PA) efficiency compared to simple CAF. However, the coded BER of ACE has not been analyzed so far, where the degradation of signal-to-noise power ratio (SNR) due to the average power increase may have dominant effect on the performance. Moreover, in order to perform effective error correction, the statistical property of the clipping noise imposed by ACE should be developed. This paper empirically models the clipping noise distribution after ACE such that soft decoding can be employed at its receiver, and investigates its performance in terms of mutual information and coded BER employing near optimal channel coding. It is revealed that the simple CAF may outperform the ACE in terms of both PA efficiency and BER performance when both are protected by practical channel coding.