An Optimization Modeling Method for Adaptive Systems Based on Goal and Feature Models

Abstract

Adaptive Socio-Cyber-Physical Systems (SCPSs) need a comprehensive requirements modeling approach to embed social concerns (goals) in their development activities. Since these kinds of systems often involve complicated and dynamic interactions with their environment, they must react to environmental changes using different potential solutions that satisfy social concerns as well as system objectives and qualities. This paper presents an optimization modeling method that monitors an SCPS's environment and qualities to provide design-time and runtime solutions that satisfy required goals of the system and its stakeholders, as well as imposed correctness constraints specified in a feature model. We combine arithmetic functions generated automatically from goal and feature models as an objective function input to an optimization tool (IBM CPLEX) in order to compute, at design time, optimal solutions for common situations. Runtime optimization can also be used for unforeseen situations. An illustrative example is used to assess the feasibility of the method. The results show that optimizing the mathematical functions of goal/feature models together is beneficial in exploring SCPS requirements and detecting weaknesses in common adaptation situations.