Autonomous Function Composition and Evolution for Adaptive Information Network

Information networks have been introducing new or improved protocols one after another to satisfy diverse requirements of a variety of emerging applications, which makes network systems too complex to control and even fragile. In this paper, to accomplish a sustainable network system which autonomously adapts to diverse requests, we propose an autonomous function composition method and evaluate its effectiveness. More specifically, each node in our proposed architecture has small pieces of networking functions, called function modules, and adaptively combines appropriate function modules to answer each service request. Combinations are refined taking into account the degree of satisfaction of users by a genetic algorithm. Then, the latest child combination is applied to a new request. Through simulation experiments, we show that our proposal can successfully achieve the maximum fitness adapting to request changes.