PERMS: An efficient rescue route planning system in disasters

The occurrence of natural and man-made disasters usually leads to significant social and economic disruption, as well as high numbers of casualties. Such occurrences are difficult to predict due to the huge number of parameters with mutual interdependencies which need to be investigated to provide reliable predictive capabilities. In this work, we present high-Performance Emergent Rescue Management e-System (PERMS), an efficient rescue route planning scheme operating within a high-performance emergent rescue management system for vehicles based on the mobile cloud computing paradigm. More specifically, an emergency rescue planning problem (ERRP) is investigated as a multiple travelling salesman problem (MTSP), as well as a novel phased heuristic rescue route planning scheme. This consists of an obstacle constraints and task of equal division-based K-means++ clustering algorithm (OT-K-means++), which is more suitable for clustering victims in disaster environments, and a glow-worm swarm optimisation algorithm based on chaotic initialisation (GSOCI), which provides the appropriate rescue route for each vehicle. A prototype is developed to evaluate the performance of this proposed approach, which demonstrates a better performance compared to other well-known and widely used algorithms. As demonstrated by the validation process, our approach enhances the accuracy and convergence speed for solving the emergency rescue planning problem. Furthermore, it shortens the length of the rescue route, as well as rescue time, and it leads to reasonable and balanced allocation of emergency rescue tasks, whilst achieving an overall efficient rescue process. More specifically, by considering scenarios with 200 victims, compared with K-means and K-means++, OT-K-means++ reduces the time cost of clustering by 9.52% and 17.39% respectively, and reduces the number of iterations by 11.11% and 15.78% respectively. Compared with ACO or GA, GSOCI reduces the length of rescue route by 9.81% and 16.36% respectively, and reduces the time of rescue by 4.35% and 15.38% respectively.