Machine Learning Framework for Sensing and Modeling Interference in IoT Frequency Bands

Spectrum scarcity has surfaced as a prominent concern in wireless radio communications with the emergence of new technologies over past few years. As result, there is growing need for better understanding spectrum occupancy newly emerging access supporting Internet Things. In this paper, we present framework to capture and model traffic behavior short-time IoT applications shared bands determine existing interference. The proposed capturing method utilizes software defined monitor short bursts transmissions by time series data which converted power spectral density extract observed occupancy. Furthermore, propose use an unsupervised machine learning technique enhance conventionally implemented energy detection methods. Our experimental results show that temporal frequency can be well-captured using combination two models, namely, semi-Markov chains Poisson-distribution arrival rate. We conduct extensive measurement campaign different urban environments incorporate spatial effect on spectrum.