Sensor Node Localization Using Machine Learning for Indoor Location Estimation

Abstract

Localization plays an important role in wireless sensor networks. A system that uses received signal strength indicator (RSSI) measurements, Bluetooth Low Energy (BLE), and RSSI fingerprinting is put forth. The location of sensor node, which is not known, can be found without the assistance of GPS, GPRS, or any satellite service that provides the facility to find the missing object in an indoor environment. BLE beacons are positioned all across the indoor environment to create a network of reference points. These beacons send out signals with distinc- tive identities, which mobile devices can detect and gauge signal strength from. A fingerprinting database is created by gathering RSSI values from various beacons and connecting particular RSSI patterns to well-known places inside the museum. Reducing the number of anchor nodes and improving location accuracy are the key objectives in this situation. Applying the gaussian filter to the algorithm makes sure that all the unwanted signals sent by the anchor nodes will be ignored, only RSSI value will be considered for the process, which will contribute to the increase in accuracy of the end result. When compared to other algorithms, the Cooja simulator evaluation of the algorithm reveals that this approach offers superior precision and accurate positioning.