To cope with the complexity of wireless networks, academic and industry researchers in the eWINE project proposes to utilise machine learning techniques to better understand, diagnose, optimize and remedy wireless networks. However, machine learning and wireless networking are two research domains that can, from time to time, be difficult to combine with each other. This paper gives (i) an overview of application domains in which machine learning has successfully been applied, (ii) discusses the benefits and limitations of using different machine learning techniques for improving network performance and (iii) gives a detailed tutorial-style example on how to apply machine learning for detecting devices types and protocol types based on wireless traces. The paper mainly targets researchers and developers with network protocol experience that do not yet have a deep understanding of machine learning techniques and that wish to learn more about the possibilities and possible applications of these techniques for network optimization.
Authors: Merima Kulin, Carolina Fortuna, Eli De Poorter, Dirk Deschrijver, Ingrid Moerman, “Data-Driven Design of Intelligent Wireless Networks: An Overview and Tutorial”, Sensors 2016, 16(6), 790, doi: 10.3390/s16060790 ” (see http://www.mdpi.com/1424-8220/16/6/790).