Using context-aware sub sorting of received signal strength fingerprints for indoor localisation

Abstract

Mobile indoor localisation has numerous uses for logistics, health, sport and social networking applications. Current wireless localisation systems experience reliability difficulties while operating within indoor environments due to interference caused by the presence of metallic infrastructure. Current position localisation use wireless channel propagation characteristics, such as RF receive signal strength to localise a user's position, which is subject to interference. To overcome this, we developed a Fingerprint Context Aware Partitioning tracking model for tracking people within a building. The Fingerprint Context Aware Partitioning tracking model used received RF signal strength fingerprinting, combined with localised context aware information about the user's immediate indoor environment surroundings. We also present an inexpensive and robust wireless localisation network that can track the location of users in an indoor environment, using the Zigbee/802.15.4 wireless communications protocol. The wireless localisation network used reference nodes placed at known positions in a building. The reference nodes are used by mobile nodes, carried by users to localise their position. We found that the Fingerprint Context Aware Partitioning model had improved performance than using only multilateration, in locations that were not in range of multiple reference nodes. Further work includes investigating how multiple mobile nodes can be used by Fingerprint Context Aware Partition model to improve position accuracy.