Automated Online Experiment-Driven Adaptation–Mechanics and Cost Aspects

As modern software-intensive systems become larger, more complex, and customizable, it is desirable to optimize their functionality by runtime adaptations. However, in most cases infeasible fully model predict behavior advance, which a classical requirement of self-adaptation. To address this problem, we propose self-adaptation based on sequence online experiments carried out production environment. The key idea evaluate each experiment data analysis determine the next potential via an optimization strategy. feasibility approach illustrated use case devoted traffic navigation where Bayesian optimization, grid search, local search are employed as strategies. Furthermore, cost discussed three components examined—time cost, adaptation endurability cost.