Towards Adaptive Monitoring Services for Self-Adaptive Software Systems

In order to deal with the great diversity of execution contexts, modern software systems rely on feedback control loops and external monitoring services for observing them and their environment, and respond to context changes through adaptation. In this process, the monitoring services play a crucial role since the quality of the monitoring data (e.g., timeliness, freshness, accuracy, availability, etc.) affects directly the self-adaptation decisions. Most of the current approaches supporting monitoring for self-adaptive systems (SASs) assume that the monitors are static components and they do not change at runtime. Due to the dynamism of execution contexts mentioned before, this vision is not valid anymore. Nowadays, monitoring services need to be adaptive as well, in order to respond to context changes, e.g., new measures to collect are required or a monitor service failure occurs at runtime. The complexity of handling monitoring services adaptation in coordination with SASs operation challenges and offers new opportunities to software engineers. In order to address this challenge, this research proposes to extend the adaptation logic of modern SASs with an external MAPE-K loop for managing the adaptation process of the monitoring services participating in the SASs adaptation logic. Different algorithmic, statistical, modeling and stochastic analysis and decision-making techniques are being explored for implementing this loop. Moreover, a reusable architecture for enacting the adaptation decisions in the monitoring services is being developed. The approach will be evaluated in two systems: a self-adaptive smart vehicle and a real self-adaptive video streaming service.