Analysis and Optimization of Trellis Shaping Concatenated with Bit-Interleaved Coded Modulation

Signal shaping is an important factor to achieve Shannon capacity in wireless communications, and trellis shaping (TS) is an effective shaping method to reduce average power by exploiting the trellis structure at the transmitter. TS uses a convolutional code for shaping, but no analysis has so far been presented in the literature that reveals the dependence of the average power reduction capability on the trellis code structure. In this paper, we propose a soft-in soft-output (SISO) decoding for TS in the framework of bit- interleaved coded modulation (BICM) and analyze the relationships between its performance and the structure of the shaping convolutional codes in terms of average power reduction capability and achievable average mutual information (AMI). It is also shown that the minimum free distance of shaping code has a dominant effect on the resulting bit-error rate (BER) performance of TS.