Kernel-based machine learning using radio-fingerprints for localization in WSNs
IEEE Transactions on Aerospace and Electronic Systems
This paper introduces an original method for sensors localization in wireless sensor networks that uses radio-location fingerprinting and machine learning. It consists of defining a set of fingerprints, relating some Received Signal Strength Indicator (RSSI) measurements to the locations where they are collected. Fingerprints are then used to define a model, whose inputs and outputs are the RSSIs and the sensors locations respectively. Several kernel-based machine learning techniques are investigated to define this model.