Conceptual application of the MAPE-K feedback loop to opportunistic sensing

Abstract

Opportunistic sensing is an adaptive technique for context recognition that aims to make use of available sensors instead of requiring the deployment of specific sensors for a specific recognition goal. Methods implementing the opportunistic sensing approach need to adapt themselves to the availability of data sources and the context recognition goals of the system. Opportunity Framework, a reference implementation of opportunistic sensing, provides this adaptability with two black box components. In the scope of autonomous computing the explicit use of feedback loops to control self-adaptation has been proposed. Several authors have argued that this concept is central to self-adaptive systems in general. Therefore, applying an explicit feedback loop can be useful to conceptually compare and combine Opportunity with other self-adaptive systems. In this paper we first applied MAPE-K feedback loop for the adaptability of the Opportunity Framework on a conceptual level. Our findings are that this feedback loop could be applied to Opportunity but that multiple feedback loops could provide clearer separation of the adaptive components. Second, using the DYNAMICO model for this purpose, a conceptual model that proposes three interacting feedback loops, yields further insights and leads us to propose separation of components for the adaptability in Opportunity Framework.