Title:Cloud Enabled Emergency Navigation Using Faster-than-real-time Simulation

Abstract:State-of-the-art emergency navigation approaches, which aim to evacuate civilians during a disaster, commonly make decisions in a real-time fashion with respect to one pre-defined algorithm and living sensory data. Hence, fatalities caused by the inappropriate guidance of an approach can only be unveiled until the end of an evacuation process and is impossible to be remedied. Previous research implies that the performance of routing algorithms for evacuation proposes are sensitive to initial distribution of evacuees, occupancy rate, disaster type as well as disaster location. In other words, a well performed algorithm in one scenario may achieve bad results in another scenario. This problem is especially serious in heuristic-based routing algorithms where results are affected by the configuration of certain parameters. Therefore, this paper proposes a simulation-based routing algorithm to realise near-optimal evacuations by making use of the high computational power of cloud servers. Rather than guiding evacuees with a routing algorithm directly, a robust Cognitive Packet Network based algorithm is first evaluated via a cloud-based simulator in a faster-than-real-time manner. Towards all the perished evacuees in the simulation, a variant of Dijkstra's algorithm is then employed to solve optimal paths for them and all the evacuees will finally follow the desired paths to exits. Furthermore, the "tsunami of data" phenomenon caused by simultaneous information exchanges among massive mobile devices and cloud servers is avoided as the proposed algorithm can calculate desired paths only based on the initial situation.