Geo-referenced multi-agent architecture for surveillance

Abstract

Surveillance systems are evolving and analog cameras give way to IP cameras, wireless sensors are used in home surveillance, remote mobile phones access to video surveillance is current practice, but the concept itself, of “surveillance systems”, has not evolved accordingly. Control rooms are still the heart of operations, where several security guards spread their attention over an array of monitors, trying to find events. In some cases, a set of sensors may be used to help detect intrusion or fire, but these are seldom integrated with video surveillance. Under a research project (DVA) we developed a new approach to surveillance systems based in geographic position of events, sensors and security agents. This new architecture, based in geographic positions, is fully distributed (empowering sensors with processing skills), bridging them together in a society of software agents that, supported by inference capabilities, work for a common awareness and, given that they are location aware, can direct the appropriate security officers for each kind of emergency.