Graph-based approach for software allocation in automotive networked embedded systems: A partition-and-map algorithm

Complex automotive networked embedded systems require novel algorithms for exploring different design decisions at early stages of the design flow. The problem of allocating the software components on electronic control units lies at the core of these design decisions. This paper formalizes this allocation problem using graph theory. The proposed formalism allows the designer to use a wide variety of graph-theoretic optimization algorithms, which are capable of minimizing more than one criterion simultaneously. The proposed algorithm is then shown, by means of numerical examples, to give the same answer as mathematical optimization but is 15 times faster in computation time.