Self-healing and self-adaptive management for iot-edge computing infrastructure

Abstract

Containerized micro-service oriented computing deployment strategies have proven to possess resilience, selfadaptive, and self-healing properties in cloud computing environments. The rapid growth of smart Internet of Things (IoT) deployments necessitates a similar approach to mitigate the challenges associated with manually managing large fleets of IoT devices. To address these challenges, we propose a novel software framework that extends Kubernetes(K8s) to collect and integrate IoT device performance metrics. By leveraging this framework, a set of self-healing and self-adaptive strategies can be deployed, taking into account the status of IoT devices. In our research, we evaluate the impact of IoT device-to-edge compute latency, bandwidth, and jitter information using the proposed software framework, including a metrics collection plugin and a custom scheduler. The results demonstrate significant enhancements in Quality of Service measures for a benchmark application scenario, emphasizing the framework’s ability to reduce manual intervention efforts through extended adaptation strategies.