Supporting composition in symbolic system synthesis

Typically, state-of-the-art approaches in system synthesis do not consider the trend in embedded systems design towards systems-of-systems where optimized subsystems exist from previous projects or as 3
^rd
party IP. In this paper, we propose a novel top-down system synthesis approach with additional support for the composition of subsystems that is based on the use of hierarchical mapping edges and a list-based scheduling algorithm using distributed priority queues. The proposed method not only enables composition of existing subsystems, but experimental results also show a significant reduction of the design space while maintaining a good quality of the implemented systems. Especially for large network-on-chip systems (NoC), our approach outperforms an existing top-down methodology in solving time by nearly 50% and in average quality by 11%.