Architecting self-adaptive Internet of Things (IoT) systems pose a lot of challenges due to heterogeneity, resource constraints, interoperability, etc. Although microservice architectures (MSA) emerged as a popular solution for developing next generation IoT systems, they further increase these challenges. This can be attributed to the complexity involved in managing adaptation concerns arising at different levels: i) IoT devices level, due to open and changing contexts, resource constraints, etc; ii) microservices level, due to dynamic resource demands; iii) application level itself, due to the changing user goals. In fact, recent studies have shown that traditional self-adaptation techniques are not flexible enough to be applied to MSA based systems. Moreover, what proposed in the literature handles adaptation either at the architectural level or at the application level. Towards this direction, we propose a self-adaptive architecture for microservice-based IoT systems. In particular, the architecture supports data-driven adaptations, by also leveraging machine learning techniques, and handles adaptations at different levels in a different manner: i) at device level, through a fog layer; ii) at microservice level, by leveraging the use of service mesh; iii) at application level, by means of dynamic QoS-aware service composition.
|Titolo:||Data-driven Adaptation in Microservice-based IoT Architectures|
|Data di pubblicazione:||2020|
|Appare nelle tipologie:||4.1 Contributo in Atti di convegno|