Search-based Adaptation Planning using Post-pareto optimality methods

Qing Dai, Qingshan Li, Lu Wang

Abstract: With the increase of the complexity of modern software and the instability of its running environment, researchers combine the search-based software engineering and self-adaptive software, propose the theory of Search-based Adaptation Planning [1] to assist decision making of self-adaptive software, which would yield a Pareto Set of numerous strategies. Thus, how to analyse and select strategies from the Pareto set need to be studied. Meanwhile, the search-based adaptation Planning theory lacks practical verification. In this paper, a self-adaptive software system with Search-based Adaptation Planning is used as the test case. Three post-optimization methods are applied in the system to analyse and filter strategies. The experiment compares differences between elapsed time, CPU utilization and memory utilization of these methods. The work of this paper provides some suggestions for developers to filter and select strategies from search-based adaptation Planning system.