Towards aRtificiAl Intelligence-based NEtwoRk optimization(TRAINER)

TRAINER is a 6 months Experiment, under the Scientific Excellence category, within the Fourth WiSHFUL Open Call for Experiments. The WiSHFUL project is a Research and Innovation Action under the European Horizon 2020 Programme addressing the work programme topic Future Internet Research and Experimentation.

The WiSHFUL project offers several software platforms that comprise data plane and control plane functionality for advanced and intelligent radio and network control. The WiSHFUL project offers unified radio and network control interfaces for off-the-shelf as well as advanced SDR equipment that allow customizing wireless solutions for specific networking and traffic contexts. The WiSHFUL project offers open and free of charge access to a number of advanced wireless testbeds, such as TWIST (TU Berlin), w-iLab.t (imec), IRIS (TCD), Orbit (Rutgers University), a FIBRE Island at UFRJ, and NITOS (University of Thessaly). These testbeds offer different wireless technologies.

In that context, the main objective of TRAINER was to extend the capabilities of the WISHFUL testbeds through the introduction of knowledge-based capabilities provided by the RapidMiner tool. RapidMiner Studio is a powerful all-in-one tool that features hundreds of pre-defined data preparation and machine learning algorithms to support data science projects. This contributes to expanding the capabilities of the existing WiSHFUL Intelligence framework that offers an experimentation environment for early implementation and validation of end-to-end 5G solutions that improve resource utilization through advanced reconfigurability of radio and network settings. The combination of RapidMiner with this framework expands the range of data mining algorithms with the inclusion of classification, prediction and clustering algorithms. To this end, a classification and categorization of data that can be obtained from the different WiSHFUL testbeds and the methodology to use the designed extension in RapidMiner were outlined.

Furthermore, to illustrate the use of the proposed WiSHFUL extension, a specific experiment using IRIS testbed at Trinity College Dublin was performed. The experiment focuses on the Channel Selection functionality for Cognitive Radio Networks (CRN), so that an access point decides the most appropriate channel to use within a band that is shared among multiple transmitters. This selection is based on a supervised classification provided by the WiSHFUL Intelligence framework extension that allows estimating the number of interfering sources existing in a given frequency channel.