{"about":{"site":"https://codewithpapers.app","non_affiliation":"Code with Papers and Syntology are not affiliated with, endorsed by, or sponsored by Papers with Code, Meta, or the pwc-archive mirror.","licence":"CC BY-SA 4.0","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","attribution":"https://codewithpapers.app/attribution","modified":"archive material modified by Syntology; see the attribution page"},"url":"/paper/fast-deep-learning-for-automatic-modulation","title":"Fast Deep Learning for Automatic Modulation Classification","arxiv_id":"1901.05850","date":"2019-01-16","proceeding":null,"authors":["Sharan Ramjee","Shengtai Ju","Diyu Yang","Xiaoyu Liu","Aly El Gamal","Yonina C. Eldar"],"abstract":"In this work, we investigate the feasibility and effectiveness of employing\ndeep learning algorithms for automatic recognition of the modulation type of\nreceived wireless communication signals from subsampled data. Recent work\nconsidered a GNU radio-based data set that mimics the imperfections in a real\nwireless channel and uses 10 different modulation types. A Convolutional Neural\nNetwork (CNN) architecture was then developed and shown to achieve performance\nthat exceeds that of expert-based approaches. Here, we continue this line of\nwork and investigate deep neural network architectures that deliver high\nclassification accuracy. We identify three architectures - namely, a\nConvolutional Long Short-term Deep Neural Network (CLDNN), a Long Short-Term\nMemory neural network (LSTM), and a deep Residual Network (ResNet) - that lead\nto typical classification accuracy values around 90% at high SNR. We then study\nalgorithms to reduce the training time by minimizing the size of the training\ndata set, while incurring a minimal loss in classification accuracy. To this\nend, we demonstrate the performance of Principal Component Analysis in\nsignificantly reducing the training time, while maintaining good performance at\nlow SNR. We also investigate subsampling techniques that further reduce the\ntraining time, and pave the way for online classification at high SNR. Finally,\nwe identify representative SNR values for training each of the candidate\narchitectures, and consequently, realize drastic reductions of the training\ntime, with negligible loss in classification accuracy.","url_abs":"http://arxiv.org/abs/1901.05850v1","url_pdf":"http://arxiv.org/pdf/1901.05850v1.pdf","source":{"archive":"pwc-archive (Hugging Face), CC BY-SA 4.0","snapshot":"2025-07-28","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","row_kind":"abstracts"},"code_links":[{"paper_slug":"fast-deep-learning-for-automatic-modulation","repo_url":"https://github.com/dl4amc/source","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":1,"framework":"tf","reach":null},{"paper_slug":"fast-deep-learning-for-automatic-modulation","repo_url":"https://github.com/dharaspatel/CNN_Signal_Classification","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"none","reach":null}],"tasks":[{"task_slug":"classification-1","task_name":"Classification"},{"task_slug":"deep-learning","task_name":"Deep Learning"},{"task_slug":"classification","task_name":"General Classification"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":null,"mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}