Constraint-Driven Simulation-Based Automatic Task Allocation on ECU Networks 2004-01-0757

With the increasing number of ECUs in modern automotive applications (70 in high end cars), designers are facing the challenge of managing complex design tasks, such as the allocation of software tasks over a network of ECUs. The allocation may be dictated by different attributes (performance, cost, size, etc.). The task of validating a given allocation can be achieved either via static analysis (e.g., for cost, size) and/or dynamic analysis (e.g. via performance simulation - for timing constraints). This paper brings together two key concepts: algorithmic and optimization techniques to be used during static analysis and virtual integration platforms for simulation-based exploration. The two concepts together provide the pillars for a constraint-driven / simulation-based approach, tailored to optimize the entire ECU network according to a cost function defined by the user.