WIP: Impulsive Noise Source Recognition with OFDM-WiFi Signals Based on Channel State Information Using Machine Learning

This study presents contributions on the detection of impulsive noise sources using OFDM-Wifi signals and Machine Learning models. The influence of impulsive noise sources on WiFi signals is used to obtain the features for supervised Machine learning models. A measurement campaign has been carried out at two indoor locations, using the Atheros CSI tool to obtain the channel state information. Feature extraction for impulsive noise detection has been performed by processing the amplitude of the channel state information of each subcarrier. These features have fed two supervised Machine Learning models, a Random Forest algorithm, and a higher level algorithm such as a Deep Neural Network. The results obtained indicate that Wifi-OFDM signals can be used for impulsive noise source recognition. The main contributions of this work focus on the extraction of suitable features for the identification of impulsive noise sources through machine learning models. The accuracy greater than 0.9 in source identification validates the proposed model, which serves as a precedent for future studies in this area.