Improved commissioning in indoor wireless networks through sensor fusion using clustering
Stijn Crul, Geoffrey Ottoy, Liesbet Van der Perre, Lieven De Strycker · 2018
Commissioning of wireless networks can be a burdensome and costly task when the number of nodes increases significantly. This paper presents a method to reduce the required effort and cost by automating parts of the commissioning process, based on the location of the wireless devices. We have investigated how we can use sensor data to create a “signature” for a room by conducting a measurement campaign using a sensor fusion platform. We observed similar data being captured from each sensor within the same room, thus the sensor data can be used as location information for the commissioning process. An unsupervised machine learning technique was chosen to process the collected data and cluster all the data from each room together. After validating the algorithm in simulation, a Proof-of-Concept (PoC) was built to validate the system empirically. The PoC demonstrates how settings are pushed towards devices automatically based on their location and how the required effort can be reduced by automation.