Introduction to the special issue on data-driven and large-scale distributed simulations

Welcome to this special issue on data-driven and large-scale distributed simulations! Modeling and simulation is required in many areas of science and engineering, for example, for predicting the behavior of new systems being designed or for analyzing natural phenomena. These simulation systems often require huge computing resources, and the datasets required by the simulation may also be geographically distributed. Furthermore, the development of such complex simulation applications usually requires collaborative efforts from researchers with different domain knowledge and expertise, possibly at different locations.

With the recent development in sensor technology, widespread availability of mobile devices, and the convenience of accessing the Internet, data have been generated at unprecedented levels. The last decade has witnessed an explosion of interest and innovation in the field of data-driven and large-scale distributed simulations. To address challenges in data-driven and large-scale distributed simulations, this special issue includes eight papers on the following topics: frameworks and toolkits for large-scale parallel and distributed simulations, system and performance issues related to large-scale data-driven and distributed simulations, and applications.

An important issue to achieve scalable simulations is to identify opportunities for parallelization and distribution in the underlying simulation models so that the simulation can exploit modern high-performance computing and cloud computing environments. To this end, Yao et al identified major computational characteristics that cause large computational requirements in analytic simulations and proposed a framework to explore parallelism at three different levels. Developing a large-scale distributed simulation based on IEEE high level architecture (HLA) standard often requires considerable effort because of the complexity of the standard, lack of proper documentation, and ready-to-use examples. To ease the development of HLA-based distributed simulation, Falcone et al described a general-purpose, domain-independent software framework in their paper. To enable efficient and scalable simulation, Romdhanne and Nikaein proposed a general-purpose coordinator–master–worker model in their paper.

Power and energy consumption have become a major concern for many high-performance and mobile computing applications and these concerns have not yet been sufficiently addressed for parallel and distributed simulations. However, energy consumption is an important concern when the simulation operates within battery-operated mobile devices as in some data-driven applications. In their paper, Biswas and Fujimoto discussed the importance of these factors and proposed metrics to measure power and energy overhead in distributed simulations. Container-based network emulation offers both high fidelity and scalability. In their paper, Yan and Jin developed a lightweight container-based virtual time system for the Linux Kernel in order to improve temporal fidelity and scalability of emulation. Load balancing is an important performance consideration in order to achieve speedups and scalable simulations. Lin et al studied load-balancing issue in parallel discrete-event stochastic simulation of the chemical reactions and diffusion in their paper.

This special issue closes with two papers on applications. In their papers, Shen et al compared internal mechanisms of existing algorithms to simulate disease transmission, while Millar et al reviewed the distributed virtual environment literature with regard to consistency and fairness.

The theme of this special issue is large-scale and data-driven distributed simulations. The collection of papers in this special issue captures current research in the field and provides useful insights to the readers. Finally, we would like to express our appreciation to all the authors and the reviewers for their contributions to this special issue.