SimConnector: An Approach to Testing Disaster-Alerting Systems Using Agent Based Simulation Models

Abstract

The design, development and testing of intelligent disaster detection and alerting systems pose a set of non-trivial problems. Not only are such systems difficult to design as they need to accurately predict real-world outcomes using a distributed sensing of various parameters, they also need to generate an optimal number of timely alerts when the actual disaster strikes. In this paper, we propose the SimConnector Emulator, a novel approach for the testing of real-world systems using agent-based simulations as a means of validation. The basic idea is to use agent-based simulations to generate event data to allow the testing of responses of the software system to real-time events. As proof of concept, we have developed a Forest Fire Disaster Detection and Alerting System, which uses Intelligent Decision Support based on an internationally recognized Fire rating index, namely the Fire Weather Index (FWI). Results of extensive testing demonstrate the effectiveness of the SimConnector approach for the development and testing of real-time applications, in general and disaster detection/alerting systems, in particular.