Dependability Explicit Metadata: Experimental Results and Research Issues

  1. Lookup NU author(s)
  2. Dr Carl Gamble
  3. Dr Stephen Riddle
Author(s)Gamble C, Riddle S
Publication type Report
Series TitleSchool of Computing Science Technical Report Series
Year2011
DateMay 2011
Report Number1249
Pages50
Full text is not currently available for this publication.
This report presents the results of from the development of prototype tool support to demonstrate metadata-based dynamic resilience, and a set of research issues that outline further work to improve the fidelity of the tool support and mature the resilience framework. The problem addressed by the work is the provision of support for dynamic reconfiguration in the context of Network Enabled Capability, by maintaining resilience dynamically. Resilience is defined as the ability of a system to maintain a dependable service while assimilating change, without loss of functionality. To achieve the required level of resilience requires the availability at run-time of dependability explicit metadata, information about system components which can govern decision-making about reconfiguration. This report is in two parts. The first part describes the experiment, which is used to illustrate how metadata-based reconfiguration can be more resilient to change than a system with a static configuration. This experiment is a deterministic simulation which is instrumented to allow the properties and configuration of the system to be read and altered. The second part discusses the simplifications used in the experiment, describes the issues which need to be considered in order to improve the fidelity of the experiment, and presents a set of research issues for further development of the resilience approach in terms of the metadata, provenance and policy definitions which make up the framework. The key results from the experiment efforts are the understanding of the properties that would be required of a simulation platform to gain a realistic confidence in the approach. The main requirements are a high fidelity simulation of a system-ofsystems (SoS), with a valid model of the component failures. This would allow the effectiveness of the transfer of metadata and the decisions made upon it to be assessed. A set of six properties is provided that may be used to guide the definition of metadata for a component or SoS. The main research issues are the generation of ontologies of metadata by examination of existing systems, and the exploration of protocols used to ensure that components within the SoS get the metadata they need in a timely manner.The means for generating policies is seen as the biggest research issue in that particular area, with the policies presented in an earlier deliverable been programmatic and too fine grained for use in a battlespace situation. An approach using natural language terms is suggested and related work from the security community is discussed. The final research issue relates to provenance metadata. Here the provenance structure used during in the experiment is presented and shown to be of use for detecting data incest and data repudiation. The work reported in this document remains at an early stage and is still relatively low Technology Readiness Level (TRL). The research issues section addresses the developments in the approach which would be required in order to increase its maturity
InstitutionSchool of Computing Science, University of Newcastle upon Tyne
Place PublishedNewcastle upon Tyne
URLhttp://www.cs.ncl.ac.uk/publications/trs/papers/1249.pdf
ActionsLink to this publication