A Cyber-Physical Systems Approach to Optimizing Internet of Vehicles Architecture with Rapidly Evolving Technology

Abstract

There is significant work underway to make the internet of vehicles a reality, however the required infrastructure is substantial and the technologies will quickly evolve during the early adaption phase. While an internet of vehicles architecture must be optimized for its key performance metrics, because of the technological volatility, additional objectives measuring the sensitivity of the key performance metrics to technological change should also be included. This would minimize the impact of evolving technologies although it would also effectively double the number of objectives. A large number of objectives creates a significant difficulty for optimization due to Pareto breakdown where a large percentage of the solution space is non-dominated. An approach is developed that casts this as a cyber-physical systems architecting problem using a meta-architecture that represents the problem in terms of systems and interfaces. The systems are formally characterized in manner allowing the objectives to be calculated given an architecture which is a particular instance of the meta-architecture. The meta-architecture is structured as a chromosome to allow the use of a many-objective evolutionary algorithm to optimize the architecture despite the large number of objectives present. An evolutionary algorithm was chosen because it does not require a continuous solution space and can readily enforce constraints on solutions. A guide to problem formulation and models of representative objectives and feasibility constraints are provided to demonstrate this approach.