Probabilistic Shaped Joint Source-Channel Polar-Coded Modulation

Guo-Jun Liao,Lin Zhou,Ran An,Chen Chen,Xian-Zhong Xie
DOI: https://doi.org/10.1109/lsp.2024.3407478
2024-06-15
IEEE Signal Processing Letters
Abstract:This letter proposes a high-order probabilistic shaped (PS) system for joint source-channel coding (JSCC) using polar codes, which are employed for both source compression and channel coding. The modulation is based on a selection rule derived from the proposed bit classification distribution matching (BCDM) algorithm. An external binary random sequence is used to select and redundantly insert the codeword sequence, which makes the modulation symbol probability approach the Gaussian distribution, instead of being uniformly distributed as in the conventional constellation mapping. At the receiver, a joint belief propagation (JBP) decoder, which consists of a source and a channel BP decoder, is applied. Simulation results demonstrate that the proposed PS joint source-channel polar-coded modulation (PS-JSCPCM) system yields a significant enhancement to the bit error rate (BER) performance compared with the traditional unshaped joint source-channel code-modulation (JSCCM) system.
engineering, electrical & electronic
What problem does this paper attempt to address?