Solution Workflows for Model-Based Analysis of Complex Systems

The development and analysis of increasingly complex systems require the intensive use of models and of sophisticated approaches to systems modeling. This paper focuses on workflows supporting the solution of complex, composed, formal models used to study and/or develop real-world systems. The workflows we deal with orchestrate multiple distributed tools and applications in order to provide the user with a powerful, composed solution environment. The aim is to automate and reproduce analysis and simulation tasks starting from a high level, graph-based description of the model to be solved. This paper thus introduces solution workflows and presents the Solution Process Definition Language (SPDL) for the specification of solution workflows processes. One of the key elements of SPDL is its formal semantics, which allow for unambiguous specification of its constructs and validation of the workflows. A workflow pattern analysis of SPDL is also provided. SPDL and its execution environment, the OsMoSys framework, are then applied to a homeland security scenario. The OsMoSys framework and the SPDL language provide a practical contribution to the applicability of model engineering techniques by enabling the semiautomatic solution of complex models.

I. Introduction

Developing increasingly complex systems requires proper methodologies and tools in order to meet requirements and reduce system development costs and time. Model-based development methods and supporting technologies may cope with this challenge but they often require the building and analysis of very complex models. These models are possibly developed according to compositional and multiformalism approaches and consist of several submodels (expressed by means of different modeling languages). The work presented in this paper is motivated by the problem of supporting and (partially) automating the analysis of complex formal models of real systems, thus enabling the practical usage of model engineering techniques. The automation of the process that must be enforced to solve a complex model is a particular kind of workflow, here called solution workflow. Solution workflows offer several advantages in solving and analyzing complex models: they greatly improve the efficiency of the solution processes and support multiple experiments, allow for reuse of process definitions, promote the development of flexible and dynamic models, enable the reuse and the sharing of existing models, and the use of external (available) computational resources. The aim of this paper is to introduce solution workflows and to present the Solution Process Definition Language (SPDL). This paper also addresses the semiautomatic generation of SPDL solution workflows starting from a graph-based description of the model to be solved. Important features of SPDL and of its solution framework OsMoSys, which justify the introduction of an ad-hoc workflow language are: (a) the possibility of using the available solution and analysis tools without requiring any modifications; (b) the semiautomatic generation and execution of SPDL specifications from a high-level description of the composed model; (c) the formal semantics on which SPDL is based, allowing a rigorous definition of its constructs and the formal verification of the correctness of the flow; and (d) the expressive power of SPDL, in terms of the workflow patterns that it is able to implement.