Self-organized Evacuation Based on LifeBelt

In this paper, we have investigated the feasibility of a self-organized evacuation process when compared with a centralized control. The evacuation strategy is based on ’predicted exit time’ (a relation of ’estimated time to reach to an exit’, ’exit capacity’ and ’exit population’) for each of the exit in a multi-exit environment, selecting the minimum value exit. The self-organized strategy is based on information propagation in a peer-to-peer fashion, initiated by a special agent in each of the exit area. The propagation range (’zone of influence’) is dependent on intensity and direction of peers interaction. Based on the propagated dataset, each agent can make an autonomous decision, conceptually a converse of centralized strategy where each agent is directed by a server. The evacuation process in supported by a wearable device, i.e. LifeBelt. Through large scale simulations using cellular automata technique and a challenging airport terminal model, we have proved that an efficient evacuation based on principles of self-organization is a real possibility, even in an infrastructureless environment.