2020/06/29 by Rick Fritschek, Fritschek, Rick, Rafael F. Schaefer +3
Computer Science · #Direction-of-Arrival Estimation Techniques #FOS: Computer and information sciences #Information Theory (cs.IT) #Speech and Audio Processing #Wireless Signal Modulation Classification
paper · pdf · doi:10.48550/arxiv.2006.16015
openalex publication_date 2020/06/29 · openalex created_date 2022/07/26 · openalex updated_date 2026/07/28
Deep learning based physical layer design, i.e., using dense neural networks\nas encoders and decoders, has received considerable interest recently. However,\nwhile such an approach is naturally training data-driven, actions of the\nwireless channel are mimicked using standard channel models, which only\npartially reflect the physical ground truth. Very recently, neural network\nbased mutual information (MI) estimators have been proposed that directly\nextract channel actions from the input-output measurements and feed these\noutputs into the channel encoder. This is a promising direction as such a new\ndesign paradigm is fully adaptive and training data-based. This paper\nimplements further recent improvements of such MI estimators, analyzes\ntheoretically their suitability for the channel coding problem, and compares\ntheir performance. To this end, a new MI estimator using a \``reverse\nJensen'' approach is proposed.\n