Optimum Architecture Development Using Evolutionary Programming

This paper introduces an innovative approach to architecture design and optimization, which is inspired by the evolutionary processes of natural species. The objective is to reduce the architecting search space with structured architectural models that can handle ambiguity at different stages of the conceptual design. A key aspect of architecture development is to select a proper set of design primitives so that the problem can be formulated effectively and solved using tools available. Inspired by the famous heuristic of systems engineering that the greatest leverage in architecting is at the interfaces, an interface-based architecture design approach was developed that facilitates both the graph representations of system architectures and the applications of evolutionary algorithms. The basic idea is to choose an appropriate interface for each functional component from possible design space so that the overall system can be integrated in a way that optimizes the performance measures of interest. The interface design then drives the overall architecture design. This paper illustrates this method using the example of on-board intelligent transportation systems