Yentl Van Tendeloo

An evaluation of DEVS simulation tools

DEVS is a popular formalism for modeling complex dynamic systems using a discrete-event abstraction. Owing to its popularity, and the simplicity of the simulation kernel, a number of tools have been constructed by academia and industry. However, each of these tools has distinct design goals and a specific programming language implementation. Consequently, each supports a specific set of formalisms, combined with a specific set of features. Performance differs significantly between different tools. We provide an overview of the current state of eight different DEVS simulation tools: ADEVS, CD++, DEVS-Suite, MS4 Me, PowerDEVS, PythonPDEVS, VLE, and X-S-Y. We compare supported formalisms, compliance, features, and performance. This paper aims to help modelers in deciding which tools to use to solve their specific problems. It further aims to help tool builders, by showing the aspects of their tools that could be extended in future tool versions.

A PDEVS simulator supporting multiple synchronization protocols: implementation and performance analysis:

With the ever-increasing complexity of simulation models, parallel simulation becomes necessary to perform simulation within reasonable time bounds. The built-in parallelism of Parallel DEVS is often insufficient to tackle this problem on its own. Several synchronization protocols have been proposed, each with their distinct advantages and disadvantages. Due to the significantly different implementation of these protocols, most Parallel DEVS simulation tools are limited to only one such protocol. In this paper, we present a Parallel DEVS simulator, grafted on C++11 and based on PythonPDEVS, supporting both conservative and optimistic synchronization protocols. The simulator not only supports both protocols but also has the capability to switch between them at runtime. The simulator can combine each synchronization protocols with either a threaded or sequential implementation of the PDEVS protocol. We evaluate the performance gain obtained by choosing the most appropriate synchronization protocol. A comparison is made to adevs in terms of CPU time and memory usage, to show that our modularity does not hinder performance. We compare the speedup obtained by synchronization with that of the inherent parallelism of PDEVS in isolation and combination, and contrast the results with the theoretical limits. We further allow for an external component to gather simulation statistics, on which runtime switching between the different synchronization protocols can be based. The effects of allocation on our synchronization protocols are also studied.

Concrete syntax: a multi-paradigm modelling approach

Domain-Specific Modelling Languages (DSLs) allow domain experts to create models using abstractions they are most familiar with. A DSLs syntax is specified in two parts: the abstract syntax defines the languages concepts and their allowed combinations, and the concrete syntax defines how those concepts are presented to the user (typically using a graphical or textual notation). However important concrete syntax is for the usability of the language, current modelling tools offer limited possibilities for defining the mapping between abstract and concrete syntax. Often, the language designer is restricted to defining a single icon representation of each concept, which is then rendered to the user in a (fixed) graphical interface. This paper presents a framework that explicitly models the bi-directional mapping between the abstract and concrete syntax, thereby making these restrictions easy to overcome. It is more flexible and allows, amongst others, for a model to be represented in multiple front-ends, using multiple representation formats, and multiple mappings. Our approach is evaluated with an implementation in our prototype tool, the Modelverse, and by applying it on an example language.

Domain-Specific Modelling for Human–Computer Interaction

Model-driven engineering (MDE) is an important enabler in the development of complex, reactive, often real-time, and software-intensive systems, as it shifts the level of specification from computing concepts (the “how”) to conceptual models or abstractions in the problem domain (the “what”). Domain-specific modelling (DSM) in particular allows to specify these models in a domain-specific modelling language (DSML), using concepts and notations of a specific domain. It allows the use of a custom visual syntax which is closer to the problem domain and therefore more intuitive. Models created in DSMLs are used, among others, for simulation, (formal) analysis, documentation, and code synthesis for different platforms. The goal is to enable domain experts, such as a power plant engineer, to develop, to understand, and to verify models more easily, without having to use concepts outside of their own domain. The first step in the DSM approach when modelling in a new domain is, after a domain analysis, creating an appropriate DSML. In this chapter, we give an introduction to DSML engineering and show how it can be used to develop a human–computer interaction interface. A DSML is fully defined by its syntax and semantics. The syntax consists of (i) the abstract syntax, defining the DSML constructs and their allowed combinations, captured in a metamodel, and (ii) the concrete syntax, specifying the visual representation of the different constructs. The semantics defines the meaning of models created in the domain. In this chapter, we show how two types of semantics (operational and translational) can be modelled using model transformations. Operational semantics gives meaning to the modelling language by continuously updating the model’s state, effectively building a simulator. Translational semantics defines mappings of models in one language onto models in a language with known semantics. This enables the automatic construction of behavioural models, as well as models for verification. The former can be used for automated code synthesis (leading to a running application), whereas the latter leads to model checking. We choose to specify properties for model checking using the ProMoBox approach, which allows the modelling of properties in a syntax similar to the original DSML. A major advantage of this approach is that the modeller specifies both requirements (in the form of properties) and design models in a familiar notation. The properties modelled in this domain-specific syntax are verified by mapping them to lower-level languages, such as Promela, and results are mapped back to the domain-specific level. To illustrate the approach, we create a DSML for modelling the human–computer interaction interface of a nuclear power plant.

Increasing the performance of a Discrete Event System Specification simulator by means of computational resource usage “activity” models

Domain-specific simulators often have an edge on general-purpose simulators in terms of performance. Their intricate knowledge of the domain allows them to aggressively optimize and take shortcuts. In contrast, simulators for more general formalisms, such as Discrete Event System Specification (DEVS), need to support a wider set of models. Their inability to use domain information prevents DEVS simulators from achieving as high performance as their domain-specific variants. To solve this problem, we introduce a way to enhance the simulation performance of DEVS models through the use of computational resource usage models, often termed “activity” models. These models augment general-purpose DEVS models with domain-specific information, which can be used by the simulator. We apply this information in the context of data structure optimization, load balancing, and model allocation. Activity-awareness is a non-invasive extension to the DEVS formalism, meaning that activity-augmented models remain perfectly valid for use in activity-unaware simulators. Similarly, models without activity can still be simulated by an activity-aware simulator. Our approach is validated by making PythonPDEVS, a Parallel DEVS simulator, activity-aware and evaluating the performance impact on a set of benchmark models.

Logisim to DEVS Translation

We propose a transformation from digital logic circuits modelled in the Logisim modelling language (and tool) to behaviourally equivalent models in the Discrete-EVent System specification (DEVS) formalism. This is achieved by mapping each Logisim component to a corresponding atomic DEVS model and by preserving the component coupling. The challenge in this work is the faithful preservation of all details of the Logisim semantics. The transformation is described and the translation of an example data path is given.