Machine Learning Based Indoor Localisation Using Environmental Data in PhyNetLab Warehouse

The ongoing Industry 4.0 revolution integrates more and more Cyber Physical Systems (CPS) into the industrial processes. Use of PhyNode as a smart warehousing container in the field of logistics is considered here as an example. Application of PhyNode will simplify the overall warehouse structure by removing central decision making compartment. However, new challenges arise with localisation within a warehouse which is the focus of this work. Some environmental data are available from active sensors on each PhyNode; including light intensity, temperature, accelerations and passive metrics like Received Signal Strength Indicator (RSSI) provided from the communication system. According to the resource limitations, localisation has to be done with limited energy and very small memory footprint. This work proposes different machine learning algorithms addressing indoor localisation within a warehouse considering these limitations.