Gabriel A. Wainer

CD++: a toolkit to develop DEVS models

The features of a toolkit for modeling and simulation based on the DEVS formalism are presented. The tool is built as a set of independent software pieces running on different platforms. Not only are the main characteristics of the environment presented, a focus on its use is also considered by inclusion of application examples for a variety of problems. Many models can be defined in an automated fashion, simplifying the construction of new models and easing their verification. The use of this formal approach has allowed the development of safe and cost‐effective simulations, significantly reducing development time. Copyright

Application of the Cell-DEVS Paradigm for Cell Spaces Modelling and Simulation

We present the results obtained when using the Cell-DEVS paradigm for cell spaces modelling and simulation. This formalism allows one to model and simulate cell spaces, including delay functions, to specify their timing behavior. Cell spaces can be defined in an automated fashion, simplifying the construction of new models, and easing the verification of the structural models. The approach was implemented in a development tool, showing that development times can im prove by several orders of magnitude. The main results of development experiences are pre sented, showing the usefulness of the approach.

Timed cell-DEVS: modeling and simulation of cell spaces

DEVS and Cellular Automata formalisms are applied to define a modeling paradigm for cellular models. Different delay functions to specify the timing behavior of each cell, allowing the modeler to represent the timing complex behavior in a simple fashion. Implementation models for the formalism are presented according with the modeler and developer points of view. As a result, efficient and cost-effective development of cellular models simulators could be achieved.

Applying Cell-DEVS Methodology for Modeling the Environment

Recent research efforts have focused on the analysis of environmental systems using cellular models. Although most of the existing solutions are based on the cellular automata formalism, this technique has some problems that constrain its power, usability and feasibility for studying large complex systems. Instead, combining cellular automata with discrete event systems specifications (DEVS) showed excellent results in terms of quality and performance. Despite these encouraging results, the environmental science/engineering community still prefers more traditional approaches, as DEVS-based techniques require a fundamental change of the modeling and simulation paradigm, while entailing expertise in advanced programming, distributed computing, etc. Cell-DEVS and the CD++ toolkit were created to address these problems: they simplify the construction of complex cellular models by allowing simple and intuitive model specification. The discrete event nature of the formalism provides better precision and performance, and models can run in different simulation environments (single user, real-time, distributed/parallel) without special expertise required. Environmental applications can be easily constructed, making it possible for users with basic training in the techniques and software tools to face the study of complex problems. We present the definition of different models of environmental applications, including the pollution on a basin, fire spreading, watershed formation and viability of a population, focusing on how to define such applications using Cell-DEVS methodology, using an approach that facilitates this paradigm shift.

Models of complex physical systems using Cell-DEVS

We present the definition of diverse models of physical systems using the Cell-DEVS paradigm. Cell-DEVS is an extension of the DEVS formalism that allows the definition of cellular models. We have developed a tool implementing these theoretical concepts, making easy the definition of cell spaces with explicit timing delays. Diversity of problems can be attacked in a simple fashion, reducing the development times of complex models. A wide variety of models have been developed using this approach, and here we include examples of a fire spreading model with different conditions, formation of a watershed and robots in a manufacturing plant. These examples allow us to show the potential application of the formalism and related tools to attack different problems.

N-dimensional Cell-DEVS Models

This article presents an extension to the timed binary Cell-DEVS paradigm. The goal is to allow the modeling of n-dimensional generic cell spaces, including transport or inertial delays for each cell. The automatic definition of cell spaces is achieved, simplifying the construction of new models. The model definition is independent of the simulation mechanism, easing the verification of the structural models. It was shown that the Cell-DEVS models can be integrated in a DEVS hierarchy, improving the definition and description of complex systems. This approach allows improvements in the execution times and precision for the cell spaces simulations due to the use of a continuous time base.

Implementing parallel Cell-DEVS

Cell-DEVS is a formalism intended to model complex physical systems as cell spaces. Cell-DEVS allows to describe cellular models using timing delay constructions, allowing a simple definition of complex timing. The original specifications were extended to permit parallel specification of these models, and an associated simulation mechanism allows their execution. We present implementation issues related to the definition of parallel simulators for Cell-DEVS.

Distributed simulation of DEVS and Cell-DEVS models in CD++ using Web-Services

Abstract Discrete event system specification (DEVS) is a modeling and simulation formalism that has been widely used to study the dynamics of discrete event systems. Cell-DEVS is a DEVS-based formalism that defines spatial models as a cell space assembled of a group of DEVS models connected together. CD++ is a modeling and simulation toolkit capable of executing DEVS and Cell-DEVS models that has proven to be useful for executing complex models. We present the design and implementation of a distributed simulation engine, known as D-CD++, which exposes CD++ simulation utilities as machine-consumable services. In addition, we present the design and implementation of the Web-Service components which enable D-CD++ to expose the simulation functionalities to remote users. Enabling CD++ with Web-Services technology provides a solid framework for interoperating different DEVS implementations in order to achieve a standard DEVS Modeling Language and simulation protocols. This paves the road towards DEVS standardization, while providing a mashup approach, which can lead to higher degree of reuse and reduced time to set up and run experiments, and making sharing among remote users more effective. To prove this fact, we integrate it within larger services (such as a 3D visualization engine), showing the mechanism to incorporate to other environments (including geographical information systems, web-based applications and other modeling and simulation tools) through using standard Web-Service tools. Performance of D-CD++, major bottlenecks and communication overheads are analyzed.

Parallel Environment for DEVS and Cell-DEVS Models

Discrete Event System Specification (DEVS) is a sound formalism to describe generic dynamic systems in a hierarchical and modular way. Cell-DEVS is a DEVS-based formalism intended to model compleX physical systems as cell spaces. This work presents new techniques for eXecuting DEVS and Cell-DEVS models in parallel and distributed environments based on the WARPED kernel, an implementation of the Time Warp protocol. The optimistic simulator PCD++, built as a new simulation engine for CD++, is a toolkit that implements the DEVS and Cell-DEVS formalisms. We redesign algorithms in CD++ to carry out optimistic simulations using a non-hierarchical approach that reduces the communication overhead. The message-passing organization is analyzed using a high-level abstraction referred to as wall clock time slice. We propose a two-level user-controlled state-saving mechanism to achieve efficient and fleXible state saving at runtime. Various optimization strategies are applied to PCD++ and their effects are analyzed quantitatively, including a risk-free message type-based state-saving strategy to reduce the number of states saved during the simulation significantly, and a one log file per node strategy to break the bottleneck caused by file I/O operations. It is shown that PCD++ markedly outperforms other alternatives and considerable speedups can be achieved in parallel and distributed simulations.

Specification of Discrete Event Models for Fire Spreading

The fire-spreading phenomenon is highly complex, and existing mathematical models of fire are so complex themselves that any possibility of analytical solution is precluded. Instead, there has been some success when studying fire spread by means of simulation. However, precise and reliable mathematical models are still under development. They require extensive computing resources, being adequate to run in batch mode but making it difficult to meet real-time deadlines. As fire scientists need to learn about the problem domain through experimentation, simulation software needs to be easily modified. The authors used different discrete event modeling techniques to deal with these problems. They have qualitatively compared the Discrete Event System Specification (DEVS) and Cell-DEVS simulation results against controlled laboratory experiments, which allowed them to validate both simulation models of fire spread. They were able to show how these techniques can improve the definition of fire models.