Trial 7: OptiPAL. As planned, the input to the process will be an EAST-ADL model that has been augmented with variability elements to define the potential design space (by establishing possible alternative architectural choices to be experimented with) as well as all data necessary for the analyses/evaluation of each object (e.g. an error model for safety analysis, etc.). An initial function called the Optimization Space Definition Module (OSDM) will take this model and determine the possible design space. The Central Optimization Engine (XXX) then uses this as the basis of the optimization process, experimenting with different possibilities. Each design candidate is passed to the Variability Resolution Mechanism (VRM), which generates a new model variant with all optimization variability resolved for that given design candidate. This model variant can then be analysed by the external analysis tools and evaluated on the basis of their results. The XXX then tries new possibilities, keeping good candidates and discarding poor candidates, until it is told to stop. It then presents the list of good candidates it has kept so far, which form a Pareto front (a set representing optimal trade-offs between the various multiple optimization objectives - e.g. minimize cost, maximize safety etc.). Each 'optimal' candidate can be generated on demand from the original model by the VRM, so the optimization is not creating and storing a huge set of variant models. Key points for the analysis Availability of language construct for the definition of cost constraints Correctness of Cost evaluation Correctness of Energy consumption evaluation Correctness of Current Consumption evaluation Correctness of cable length evaluation Correctness of power distribution evaluation Correctness of Safety, Timing analysis etc. Ensure the generated 'optimal' candidate models are coherent and plausible Analysis strategies ITEM Evaluation Steps
