Title: "Evolving ML-based algorithms for wireless communications: the case of waveform detection"
In modulation identification issues, like in any other classification problem, the classification task's performance is significantly impacted by the features characteristics. Both feature weighting and denoising boost the performance of machine learning-based identification algorithms, particularly the class of instance-based learning algorithms such as the Minimum Distance (MD) classifier, in which the distance measure is susceptible to the magnitude of features. The aim is to improve the performance of a blind digital modulation detection approach in the context of multiple-antenna systems. For that purpose, a feature denoising approach is introduced. Also, a metaheuristic optimization algorithm is applied to optimize feature weights for an MD classifier.
Sofiane Kharbech - received an Engineering degree in networking and telecommunications from the National Institute of Applied Science and Technology, Tunis, Tunisia, in 2009, the M.Sc. degree in electronic systems and communication networks from the Tunisia Polytechnic School, Carthage, Tunisia, in 2012, and the Ph.D. degree in electrical engineering from the University Polytechnic Hauts-de-France, Valenciennes, France, in 2015.
From 2012 to 2015, he was a Research Engineer with the Laboratory IEMN/DOAE (CNRS UMR 8520, France). Since 2015, he has been an Assistant Professor with the Higher Institute for Technological Studies of Gabes, Gabes, Tunisia, and a Senior Researcher with the Laboratory Sys'Com (ENIT, Tunisia). In 2019/2020, he was a postdoc fellow at the University of Lille within the European project Emulradio4rail. Currently, he is a fixed-term Assistant Professor at IMT Lille Douai. His main research interests wireless communications, machine learning, optimization, and cryptography.
Contact(s) & information(s) pratique(s)
Sébastien HOUCKE - Dpt MEE (IMT Atlantique, site de Brest)