Toggle Main Menu Toggle Search

Open Access padlockePrints

[PhD Thesis] Incorporating faults and fault-tolerance into real-time networks: a graph-transformational approach

Lookup NU author(s): Daniel Owen

Downloads

Full text is not currently available for this publication.


Abstract

Investigating the specification of faults and fault tolerance strategies is a challenging task, particularly for a real-time systems (RTS) architecture. Guaranteeing that a design will satisfy its specification in an environment that may exhibit faults at its interfaces, using potentially faulty components, presents an interesting problem. We take advantage of a specific RTS architecture, that of Real-Time Networks (RTNs). The RTN specification language (RTN-SL) has a graphical and concrete grammar and an existing axiomatic semantics. We first deal with the specification of faults which are feasible for RTNs, extending the existing axiomatic semantics of RTN-SL to encompass faulty behaviours. From this, we propose several classical fault tolerance strategies for RTNs which can be incorporated in RTN designs. To show the soundness of our fault axioms, we show the existing axiomatic semantics sound, which requires the definition of a new operational semantics for RTN-SL. We also show that the fault axioms are a conservative extension to the existing axiomatic semantics, using a conservativity result for structural operational semantics (SOS). Building upon this semantic framework, we outline a transformational design methodology based on a graph grammar syntax for RTN-SL. Each of the fault tolerant strategies we consider are presented as graph grammar productions which can be applied to RTN-SL designs to tolerate component faults. We show the applicability of our methodology on a realistic study of a classical RTS implementation. As our main contribution, we show the merits of a transformational design methodology which allows for classical fault tolerant strategies to be instantiated in a design, preserving the original design’s functionality.


Publication metadata

Author(s): Owen DJ

Publication type: Report

Publication status: Published

Series Title:

Year: 2005

Institution: School of Computing Science, University of Newcastle upon Tyne

Place Published: Newcastle upon Tyne


Share