Automated Scheduling for Tightly-Coupled Embedded Multi-core Systems Using Hybrid Genetic Algorithms

Deploying software to embedded multi- and many-core hardware has become increasingly complex in the past years. Due to the heterogeneous nature of embedded systems and the complex underlying Network on Chip structures of many-core architectures, aspects such as the runtime of executable software are highly influenced by a variety of factors, e.g. the type, instruction set, and speed of the processor an executable is allocated to as well as its predecessors, their location, ordering and the communication channels in between them. In this work, we propose a semi-automated Hybrid Genetic Algorithm based optimization approach for distributing and re-scheduling executional software to heterogeneous hardware architectures in constrained solution spaces, along with an evaluation of its applicability and efficiency. The evaluation is based on both, publicly available as well as real world examples of automotive engine management systems.