Inexact End-to-End Response Time Analysis as fitness function in search-based task allocation heuristics for hard real-time network-on-chips

Search-based algorithms can be used to identify a task allocation that meets timing constraints in Network-on-Chip (NoC). The efficiency of these algorithms is directly related to their fitness functions, which are applied over a large solution space and are able to identify feasible solutions. In this paper, we introduce five speed-up schemes to reduce the running time of End-to-End Response Time Analysis, leading to a significant improvement in finding the mapping solutions by using this analysis as a fitness function within a search-based algorithm.