Model transformation for multi-objective architecture optimisation of dependable systems

The promise of model-based engineering is that by use of an integrated and coherent system model both functional and non-functional requirements may be analysed, implemented and tested in a rigorous and cost-effective manner. An important part of model-based engineering is the use of analysis and design languages. The Architecture Analysis Design Language (AADL) is a new modelling language which is increasingly being used for high dependability embedded systems development. Such languages are ideally suited to model-based engineering but the use of new languages threatens to isolate existing tools which use different languages. This is a particular problem when these tools provide an important development or analysis function. System optimization is such a function. System designers seek an optimal trade-off between high dependability and low cost. For large systems, the design space of alternatives with respect to both dependability and cost is enormous and too large to investigate manually. For this reason automation is required to produce optimal or near optimal designs. HiP-HOPS is a mature, state of the art, dependability analysis and optimisation method and tool. HiP-HOPS requires, as input, the local failure behaviour of the system components together with the inter-component failure propagation behaviour. For optimisation, component variability information is also required. The integration of tools such as HiP-HOPS into a model-based engineering environment requires that these tools have suitable access to the system model. Without proper integration, additional system information must be input at additional cost and risk of inconsistency. This paper shows how model transformation may be used to integrate a multi-objective optimization method and tool into a model-based engineering environment. To illustrate the transformation method it is applied in a case study; where, drawing from the results of the optimisation, we highlight the potential value of this work for model-based design