Flexible requirement satisfaction in adaptive systems

Student thesis: Doctoral ThesisDoctor of Philosophy (PhD)

Abstract

An adaptive system modifies its behaviour in response to changes in its environment or in the system itself. For the system to be able to perform adaptation, the system engineer must know the conditions that the system may experience. However, it is nearly impossible to enumerate all possible conditions and adaptation. Thus, the system requirements may need to be relaxed in order to accommodate adaptive behaviour.
To enable performance toleration, the RELAX approach classifies requirements into two categories; invariant requirements, which have to be fully satisfied no matter what, and relax requirements, which can be partially satisfied to maintain flexibility. To facilitate the adaptability of the system, fuzzy branching temporal logic (FBTL) is used as the semantics of the requirements language. The approach represents the relax requirements satisfaction using fuzzy sets. As a result, instead of saying whether the relax requirement is satisfied or fails, the approach says the requirement is satisfied to a certain degree.
Moreover, RELAX provides a DEP uncertainty factor to capture the relationship between requirements. DEP indicates whether relaxing a requirement will impact the satisfaction of another requirement positively, negatively or both. However, the previous RELAX approach does not take this relationship into account while doing the relaxation.
In this approach, we use RELAX requirement language and take into account the relationship between requirement while performing the relaxation. Thus, we can achieve optimal relaxation, the minimum level of relaxation where all requirements are satisfied, and present relaxation area/s where the requirements are always satisfied to a certain degree. To verify the relaxation result, we employ a UPPAAL model checker. The verification is conducted by translating the fuzzy requirements into crisp requirements using fuzzy alpha cut so the UPPAAL can verify if the requirement is satisfied or fail at a certain degree of relaxation.
Date of Award24 Jun 2021
Original languageEnglish
Awarding Institution
  • The University of Bristol
SponsorsIndonesian Endowment Fund for Education (LPDP) -Ministry of Finance of the Republic of Indonesia
SupervisorTrevor P Martin (Supervisor)

Cite this

'