10.3.1 Classification of Systems from Component Characteristics

In recent times there is a growing interest in systems architectures in domains like biological systems and social networks leading to useful insights and generalizations. These developments have opened up possibility of investigating architectures of complex engineering systems on similar lines. Architecture of a system can be abstracted as a graph, wherein the nodes/vertices correspond to components and edges correspond to interconnections between them. Graphs representing system architecture have revealed motifs or patterns. Motifs are recurring patterns of 3-noded (or 4, 5 etc.) sub-graphs of the graph. Over-represented motifs have offered insights into the basic functionality of systems in some cases. Concept of motif significance profiles (i.e., proportions of various motifs present in a system) has also given rise to interesting insights. These profiles are seen to be highly correlated across systems of the same family (i.e., very similar proportions of motifs are present in systems of same type). Recently these profiles are proposed as classifiers for system architecture. We now show that the same classification of systems can be arrived at by merely looking at characteristics of components/nodes from which systems are synthesized. In other words, we argue that the motif significance profile of a system is due to the properties of the individual components that form the system. We have shown this by considering a vast variety of systems (38 systems arbitrarily chosen) ranging from – biological systems, languages, electronic circuits, software systems and mechanical engineering systems.