5. Conclusions and future work
We presented a perennial simulation framework that targets crisis management simulation. This framework incorporated concepts of dynamic data-driven systems, symbiotic simulation as well as human-in-the-loop techniques. We detailed the framework and its components, and described three experiments that used the framework and served as case studies.
The first case study described a building evacuation scenario, and demonstrated that symbiotic simulation can improve the egress times of building occupants if real time guidance or information is provided. The second case study dealt with incident response on crowded roadways, and demonstrated the real time decision making capability of perennial simulation.
The third case study showed a novel approach of coupling two simulators, where one simulator was used for providing guidance, while the other acted as a surrogate model for the real world. Taken together the case studies provide a strong motivation for setting up perennial simulations: they are long-lasting, provide benefits during non-crisis situations, and are pivotal in establishing information superiority during a crisis an event in which time is a scarce resource, and any informed decision or control is welcome.
As future work, we aim to establish an upper bound on the runtime of “What-If?” simulations and validate case study 3 on Singapore’s road network. We aim to speed up the performance of “What-If?” simulations by using parallel computing techniques like multi-core architectures or GPUs.
