Self-Adaptive Systems

About

Self-adaptive and Self-repairing systems pose a serious challenge when there is a need for verification of the system. Any system that has been verified must be re-verified when a change occurs. Re-verification cannot be performed automatically in most cases, as often there are formal requirements specified that cannot be violated and there is no set method of confirming that a change will not violate one of the requirements at some point in the future. Thus, there is a need either to predict the entire future of a system and how the adaptation will impact the system in order for a verified system to continue operating after self-adapting or self-repairing.

Our work focuses on the creation of a verification workflow using Petri Nets. The Petri Net of the system is imbedded within the system itself, providing an architectural awareness to the system. Then, when a system needs to make a change, it can run its proposed change through its embedded Petri Net using its critical requirements to determine a) if the plan will fail a requirement and b) likelihood of the proposed plan violating the requirements. If a plan is determined to fail a requirement, a risk token used in the verification workflows is used to determine the risk of choosing each adaptation. Should all adaptation options fail, it is important to choose the one that provides the least risk to the system without causing the system to completely fail.

This work is sponsored by the Air Force Research Laboratory.

Publications

  • Self-Adaptation Strategies to Maintain Security Control Assurance Cases, S. Jahan, A. Marshall, and R. Gamble, Proceedings of the 12thIEEE International Conference on Self-Adaptive and Self-Organizing Systems,  2018.
  • Toward Evaluating the Impact of Self-adaptation on Security Control Certification, A. Marshall, S. Jahan, and R. Gamble, to appear in the Proceedings of the 13thInternational Symposium on Software Engineering for Adaptive and Self-Managing Systems (SEAMS), May 2018.
  • Assessing the Risk of an Adaptation using Prior Compliance Verification, A. Marshall, S. Jahan, and R. Gamble, in the Proceedings of the 51stHawaii International Conference on System Sciences, Jan. 3-6, 2018.
  • Embedding Verification Concerns in Self-Adaptive System Code, S. Jahan, A. Marshall, and R. Gamble, in the Proceedings of the 11th IEEE International Conference on Self-Adaptive and Self-Organizing Systems, Tucson, AZ, Sept. 2017.
  • Visualizing Self-Adaptive Plan Simulations given Embedded Verification Concerns, S. Jahan, A. Marshall, and R. Gamble, in the Proceedings of the11th IEEE International Conference on Self-Adaptive and Self-Organizing Systems (Demo Track), Tucson, AZ, Sept. 2017.