Catálogo de publicaciones - libros

Dependable service-oriented computing is challenging because it faces an open, heterogeneous, and dynamic computing environment. In a service-oriented computing system, services developed by different vendors on different platforms and in different programming languages perform computations collaboratively through open standard protocols. This paper presents an innovative dynamic reconfiguration technology that can be embedded into a service-oriented application to make the application reconfigurable. Traditional reconfiguration algorithm assumes each component can independently switch without collaboration. The proposed reconfiguration agents are embedded in different services, and they communicate via a collaborative reconfiguration protocol to achieve a consistent reconfiguration decision. In addition, the reconfiguration protocol itself is fault-tolerant.

In applying Component-Based Software Engineering (CBSE) techniques to the domain of Distributed Real-time and Embedded (DRE) Systems, there are five critical challenges: 1) discovery of relevant components and resources, 2) specification and modeling of components, 3) exploration and elimination of design assembly options, 4) automated generation of heterogeneous component bridges, and 5) validation of context-related embedded systems. To address these challenges, this paper introduces four core techniques to facilitate high-confidence DRE system construction from components: 1) A component and resource discovery technique promotes component searching based on rich and precise descriptions of components and context; 2) A timed colored Petri Net-based modeling toolkit enables design and analysis on DRE systems, as well as reduces unnecessary later work by eliminating infeasible design options; 3) A formal specification language describes all specifications consistently and automatically generates component bridges for seamless system integration; and 4) A grammar-based formalism specifies context behaviors and validates integrated systems using sufficient context-related test cases. The success of these ongoing techniques may not only accelerate the software development pace and reduce unnecessary development cost, but also facilitate high-confidence DRE system construction using different formalisms over the entire software life-cycle.

DECOS (Dependable Components and Systems) is an EU-funded integrated research project (IP) with the goal to develop a framework and an associated design methodology for the component-based design of dependable embedded systems. The core of DECOS is based on the Time-Triggered Architecture (TTA), a distributed architecture for high-dependability real-time applications. In the first part of this paper the design flow of DECOS from the Platform Independent Model (PIM) to the Platform Specific Model (PSM) is discussed and the DECOS execution environment is introduced. In the second part the fault-tolerance mechanisms of DECOS are explained. After a deliberation of the fault hypothesis, the support for the implementation of triple-modular redundancy (TMR) is presented.

The advanced mechatronic systems of the next generation are expected to behave more intelligently than today’s systems by building communities of autonomous agents which exploit local and global networking to enhance their functionality. Such mechatronic systems will therefore include dynamic structural adaptation at the network level and complex real-time coordination protocols to adjust their behavior to the changing system goals leading to cooperative self-adaptation in a safe and coordinated manner. In this paper the Mechatronic UML approach and its concepts for compositional modeling and verification of crucial safety properties for cooperative self-adaptive mechatronic systems are outlined. Based on former results for the compositional verification of the real-time coordination and safe rule-based dynamic structural adaptation, we present in this paper a systematic compositional verification scheme which permits to verify the safety of real-time systems with compositional adaptation and an unbounded number of structural configurations.

There is an increasing need for today’s autonomous systems to collaborate in real-time over wireless networks. These systems need to interact closely with other autonomous systems and function under tight timing and control constraints. This paper concerns with the modeling and quality assurance of the timing behavior of such network embedded systems. It builds upon our previous work on run-time model checking of temporal correctness properties and automatic white-box testing using run-time assertion checking. This paper presents an architecture for the network embedded systems, a lightweight formal method that is based on formal statechart assertions for the design and development of networked embedded systems, and a process of using run-time monitoring and verification, in tandem with modeling and simulation, to study the timing requirements of complex systems early in the design process.

Inheritance in real-time object-oriented programming is a young subject for research, let alone for practice. Issues in inheritance design are discussed in the context of TMO () scheme for real-time distributed object programming. The TMO scheme guides programmers to incorporate timing specifications in natural, modular, and easily analyzable forms. The scheme thus makes it relatively easy to practice inheritance design. Some TMO structuring rules and styles that enable efficient design of inheritance are presented. A GUI-based approach for TMO-framework programming with exploitation of inheritance is also discussed.