As the complexity of nowadays systems continues to grow, we are moving away from creating individual components from scratch, toward methodologies that emphasize composition of re-usable components via the network paradigm. Complex component interactions can create a range of amazing behaviors, some useful, some unwanted, some even dangerous. To manage them, a “science” for network design is evolving, applicable in some surprising areas. In this paper, we consider a few application domains and discus the design challenges involved from a methodology standpoint. From large-scale hardware/software systems, to dynamically adaptive sensor networks, and network-on-chip architectures, these ideas find wide application.
Radu Marculescu, Jan M. Rabaey, Alberto L. Sangiov