Publish/subscribe system based on event calculus to support real-time multi-agent evacuation simulation

Abstract

Large scale disasters often create the need for evacuating affected regions to save lives. Disaster management authorities need evacuation simulation tools to assess the efficiency of various evacuation scenarios and the impact of a variety of environmental and social factors on the evacuation process. Therefore, sound simulation models should include the relevant factors influencing the evacuation process. More specifically, to be reliable in critical situations, evacuation simulations must integrate information on time-varying phenomena that can affect the evacuation process, such as the impact of meteorological conditions, road incidents, or other relevant events. Sensor networks constitute an efficient solution for gathering data on such events and feeding the evacuation simulation. However, the coupling of sensors with multi-agent simulation tools is not straightforward. In this chapter, we present a publish/subscribe system based on Event Calculus to support real-time multi-agent evacuation simulations. The system identifies relevant events from sensor data gathered through a Sensor Event Service that implements the OGC SWE standards and the Event Pattern Markup Language (EML). Then, the publish/subscribe system acts as a middleware between the Sensor data publishers and the multi-agent evacuation simulation through a Sensor Processing Service based on Event Calculus that infers the impact of events on the characteristics of the road network. The result is an evacuation simulation that can be deployed to assess various evacuation scenarios in real-time, during the crisis response.