First International Workshop on Multi-level Agent-based Simulation of Smart Cities (MASW13)

Conference’s Details

Important Dates

Submission deadline: December 8, 2012

Notification: February 8, 2013

Final date for camera-ready copy: March 15, 2013

Workshop: June 25-28, 2013


The modeling of smart cities is of great theoretical and practical interest. In the past two-decade research from a broad range of fields such as computer graphics, physics, robotics, energy, social science, safety science and training systems has created simulations involving collections of elements immersed in smart cities (individuals or devices). Two major kinds of simulations may typically be distinguished depending on whether they seek to achieve: a high-level of realism of behavior (safety simulation, social sciences, energy…) or high-quality visualization (movie productions, computer games, virtual reality). Within the first category, simulation results are generally consistent with observations of real components and individuals and can therefore serve as a basis for theoretical studies for the evaluation and prediction of the system behavior. In the second area, behavior models are not the priority and do not match quantitatively the real worlds. For example, pedestrians are fully animated 3D characters and application users may have a high degree of interaction with the simulation. Recent research and applications tend to unify these two areas, especially in the domain of training systems where both aspects are necessary for an effective training and evaluation.

Many works have been devoted to the study of collective behaviors and their inherent emergent properties such as spontaneous organizations of pedestrians into lines, oscillations at gates, etc. Among all the existing approaches in simulation, those offering the highest level of realism in behavior are microscopic approaches because they explicitly attempt to model the features that take part in the expression of specific behaviors of individuals. Agent-Based Simulations (ABS) are one of the approaches to support micro-simulation. ABS principle relies upon a set of autonomous agents, which encapsulate the behaviors of individual entities (pedestrians, vehicles, devices…) Agent-based modeling allows complex behaviors of various interacting entities to emerge from a set of simpler behaviors. Phenomena such as flocks of birds, schools of fish and crowds are good examples of how systems with simple goals can exhibit emergent behaviors as the result of the interactions between the individuals. Moreover, in contrast to other micro-simulation techniques, ABS allows to catch the variety of behaviors composing a real system easily. ABS has proven to be well suited for the simulation of situations where there is a large number of heterogeneous individuals who may behave somewhat differently.

However, as soon as we consider a micro-simulation of several agents and their relationships, the complexity of the system and associated computational costs increase. We are therefore faced a dilemma common in the field of simulation: to manage a compromise between performance and accuracy. Multi-level models are an interesting direction of exploration to obtain accurate and efficient models for the simulation of the individuals in complex systems such as smart cities.

MASW13 will be held in Halifax, Nova Scotia, Canada (25-28 June 2013) in conjunction with the 4th International Conference on Ambient Systems, Networks, and Technologies (ANT 2013).