MATrA: meta-modelling approach to traceability for avionics

Abstract

Traceability is the common term for mechanisms to record and navigate relationships between artifacts produced by development and assessment processes. Effective management of these relationships is critical to the success of projects involving the development of complex aerospace products. Practitioners use a range of notations to model aerospace products (often as part of a defined technique or methodology). Those appropriate to electrical and electronic systems (avionics) include Use Cases for requirements, Ada for development and Fault Trees for assessment (others such as PERT networks support product management). Most notations used within the industry have tool support, although a lack of well-defined approaches to integration leads to inconsistencies and limits traceability between their respective data sets (internal models). Conceptually, the artifacts produced using such notations populate four traceability dimensions. Of these, three record links between project artifacts (describing the same product), while the fourth relates artifacts across different projects (and hence products), and across product families within the same project. The scope of this thesis is to define a meta-framework that characterises traceability dimensions for aerospace projects, and then to propose a concrete framework capturing the syntax and semantics of notations used in developing avionics for such projects which enables traceability across the four dimensions. The concrete framework is achieved by exporting information from the internal models of tools supporting these notations to an integrated environment consisting of. i) a Workspace comprising a set of structures or meta-models (models describing models) expressed in a common modelling language representing selected notations (including appropriate extensions reflecting the application domain); ii) well-formedness constraints over these structures capturing properties of the notations (and again, reflecting the domain); and iii) associations between the structures. To maintain consistency and identify conflicts, elements of the structures are verified against a system model that defines common building blocks underlying the various notations. The approach is evaluated by (partial) tool implementation of the structures which are populated using case study material derived from actual commercial specifications and industry standards.