Trellis-Assisted Constellation Subset Selection for PAPR Reduction of OFDM Signals

In this paper, we propose a new peak-to-average power ratio (PAPR) reduction technique, which is referred to as trellis-assisted constellation subset selection (TACSS, or TAX for short), for orthogonal frequency-division multiplexing (OFDM) signals. We control the signal transition by judiciously selecting the appropriate constellation subset for PAPR reduction with the controlling bits that are subject to the trellis-based constraint imposed by a bank of memories. A novel constellation design suitable for the proposed system is also introduced. A major property offered by the proposed approach is that the controlling bits and the information bits are perfectly separable at the receiver, which leads to a simple implementation of its soft-in-soft-out (SISO) decoder for coded OFDM systems. Through the performance analysis of PAPR reduction capability, power amplifier (PA) efficiency, average mutual information (AMI), and bit error rate (BER), the achievable tradeoff relationship between the transmitter and receiver performances of the proposed system is addressed. Furthermore, performance comparisons of the proposed scheme with other major PAPR reduction schemes such as trellis shaping, discrete Fourier transform (DFT) precoding, and clipping and filtering demonstrate its practical advantage in a high-spectral-efficiency regime.