Probabilistic indoor human movement modeling to aid first responders

Abstract

The arrival of new devices and techniques has brought tracking out of the investigational stage and into the wider world. Using Wi-Fi signals is an attractive and reasonably affordable option to deal with the currently unsolved problem of widespread tracking in an indoor environment. Here we present a system which aims at overcoming weaknesses in existing real time location systems (RTLS) by using the human approach of making educated guesses about future location. The hypothesis of this proposal is that knowledge of a person’s historical movement habits allows for future location predictions to be made in the short, medium and long term. The primary research question that is foremost is whether the tracking capabilities of existing real time locating systems can be improved automatically by knowledge of previous movement especially in the short term in the case of emergency first responders by the application of a combination of artificial intelligence approaches. We conclude that HABITS improves on the standard Ekahau RTLS in term of accuracy (overcoming black spots), latency (giving position fixes when Ekahau cannot), cost (less APs are required than are recommended by Ekahau) and prediction (short term predictions are available from HABITS). These are features that no other indoor tracking system currently provides and could prove crucial in future emergency first responder incidents.