Genetic Programming based Algorithm for HW/SW Cosynthesis of Distributed Embedded Systems Specified using Conditional Task Graph

In this paper we propose a novel genetic programming based iterative improvement approach for hardware/software cosynthesis of distributed embedded systems. Unlike other genetic programming solutions for distributed embedded systems in this work the system is specified using conditional task graph. In such a graph every node represents a single task. The edge represents amount of data needed to be transferred between connected tasks, however some of the edges can be conditional. The data is transferred using those edges only if condition is satisfied. Proposed methodology is based on genetic programming. Therefore the genotype is a system construction tree. In each nodes of the tree are system building options. The next generations are obtained using standard genetic operators: mutation, crossover, cloning and selection.