Wireless lnteroperability Spectrum Explorer using Machine Learning
April 2019 - June 2020
Integrated Design of Embedded Systems

One can observe a growing interest in adopting Machine Learning techniques in various fields of research, among which the analysis of radio signais.

The creation of synergies between Machine Learning and the SOR philosophy opens up new and very promising axis of research that the WISE-ML project would like to explore. ln particular, the WISE-ML project aims at evaluating the relevance and the effectiveness of the most promising ML implementations in the field of spectrum analysis.

As a first step the emphasis shall be placed on the study of classical methodologies for RF spectrum analysis. The purpose of this task shall be to lay the groundwork for a qualitative assessment of ML solutions. The study of ML techniques applicable to the SOR domain shall be performed in the second part of the project. An ML spectrum analysis application shall be developed/evaluated during this phase. Even though ML applications are starting to gain traction in the world of SORs, particularly in the developments dedicated to RF spectrurn analysis, there are very few attempts to implement these solutions in an SOR hardware platform. The third and last step of the WISE-ML project shall thus consist in the implementation of one or more techniques issued from previous work. The number and the types of algorithms implemented shall depend on the qualitative results obtained. ldeally, at least one ML implementation shall be implemented in order to provide a demonstrator that shall be presented as a demo at the end of the project.
The hardware and the programming language that shall be used to implement the demonstrator are not yet defined and will also depend on the findings of the first to work packages.