Emergency Navigation in Confined Spaces Using Dynamic Grouping

Abstract

The performance of Emergency Management Systems (EMS) in confined spaces is highly dependent on the decision algorithm employed for the safe navigation of the evacuees to the available exits. In the algorithm proposed in this paper, we have considered evacuees under two groups, based on their age and physical condition, and we tailor two routing metrics, one for each group, in finding suitable paths for the evacuees. A dynamic grouping mechanism that can adjust an evacuee's group, and therefore routing metric, according to its on-going health condition is employed during the evacuation. To implement the routing metrics, we have used the Cognitive Packet Network (CPN) with random neural networks (RNN) and reinforcement learning. The CPN is an adaptive routing protocol that is loop-free at all times and easily handles multiple quality of service (QoS) metrics. Simulation results show that allowing the navigation system to be sensitive to the on-going health conditions and mobility of the evacuees, using our proposed dynamic grouping, can achieve higher survival rates.