2020/04/10 by Fayçal Ait Aoudia, Jakob Hoydis, Aoudia, Fayçal Ait +1 · 1 citation
Computer Science · Engineering · #Advanced Wireless Communication Techniques #FOS: Computer and information sciences #FOS: Electrical engineering #Information Theory (cs.IT) #Machine Learning (stat.ML) #PAPR reduction in OFDM #Signal Processing (eess.SP) #Wireless Signal Modulation Classification #electronic engineering #information engineering
paper · pdf · doi:10.48550/arxiv.2004.05062
openalex publication_date 2020/04/10 · openalex created_date 2022/07/26 · openalex updated_date 2026/07/28
We introduce a trainable coded modulation scheme that enables joint\noptimization of the bit-wise mutual information (BMI) through probabilistic\nshaping, geometric shaping, bit labeling, and demapping for a specific channel\nmodel and for a wide range of signal-to-noise ratios (SNRs). Compared to\nprobabilistic amplitude shaping (PAS), the proposed approach is not restricted\nto symmetric probability distributions, can be optimized for any channel model,\nand works with any code rate k/m, m being the number of bits per channel\nuse and k an integer within the range from 1 to m-1. The proposed scheme\nenables learning of a continuum of constellation geometries and probability\ndistributions determined by the SNR. Additionally, the PAS architecture with\nMaxwell-Boltzmann (MB) as shaping distribution was extended with a neural\nnetwork (NN) that controls the MB shaping of a quadrature amplitude modulation\n(QAM) constellation according to the SNR, enabling learning of a continuum of\nMB distributions for QAM. Simulations were performed to benchmark the\nperformance of the proposed joint probabilistic and geometric shaping scheme on\nadditive white Gaussian noise (AWGN) and mismatched Rayleigh block fading (RBF)\nchannels.\n