About:
Hanan Yousef-Daher received her B.Sc. in Informatics Engineering. Software Department from Al-Baath University-Syria in August 2014.
Thesis Abstract:
Recently, humanity has witnessed an elevating number of catastrophes, either natural or man-made ones. In order to minimize the evacuation time and save people’s lives, the proficiency of evacuation plans and precautions must be taken into account. Subsequently, evacuation drills must be executed to rehearse how a structure would be emptied in the circumstances of a fire or different types of crises. However, drills are a possible approach but hardly recreating the truly panic conditions in the domain of evacuation studies. This advocates the need to use simulations in order to help design better and more secure evacuation plans.
Simulation modeling becomes the main method applied to investigate evacuation systems since it ensures a viable alternative to drills. However, it is hard to use simulation with the evacuation domain, which is considered as a complex system according to its features such as dynamicity, uncertainty, openness, and heterogeneous. Agent-Based Modeling and Simulation (ABMS) approach is probably the most suitable and realistic solution since it gives us the ability to model each individual with its own unique characteristics, thus recreating the real-world interactions among human beings. Unfortunately, ABMS still has some shortcomings such as requiring to write many lines of code and consuming significant development effort. Thus, their use might become less convenient for time and cost issues.
To address these issues, we propose to benefit from the Model-Driven Development (MDD) approach. Employing this approach in this context deals with the issue of the complexity of the evacuation system. Moreover, it enables the automatic code generation from the system model. In this respect, we designed a Domain-Specific Modeling Language (DSML) related to the evacuation’s domain concepts. We also built a graphical editor to enable building system models easily. Then using Model to Code transformations, we generated the code automatically from the model. This code is applied later to one of the Multi-Agent Systems (MAS) platform.
In order to evaluate our work, we applied our approach to several case studies. First, we modeled the evacuation of the City Center shopping mall in Isfahan, Iran. Second, we modeled the evacuation of the Kermovo shopping mall in Kermovo, Russia. Third, we modeled the evacuation of the Sarpole-e Zahab school in Kermanshah, Iran. The evaluation results showed that using the proposed approach leads to time savings, increases the speed of development, enhances the level of abstraction, and enables high flexibility and productivity in modeling emergency evacuation plans.