IoT Orchestration in the Compute Continuum: Integrating Kubernetes with Stack4Things

Managing distributed IoT infrastructures across the compute continuum poses significant challenges, especially regarding the orchestration of heterogeneous devices, real-time data processing, and reproducibility. This paper proposes an architectural solution that integrates Stack4Things (S4T) with Kubernetes, leveraging Custom Resource Definitions (CRDs) to model and manipulate IoT nodes as first-class Kubernetes objects. By aligning IoT-specific capabilities with the declarative management style of Kubernetes, our approach streamlines device provisioning, monitoring, and reconfiguration. Furthermore, we develop a Crossplane Provider that extends Kubernetes' control over edge-to-cloud services, simplifying multi-environment deployments. This integration ensures that IoT resources can be orchestrated alongside conventional containerized workloads. We present two deployment models, direct Kubernetes-S4T integration and S4T integration via Crossplane, showing how each accommodates a range of deployment needs. Throughout the paper, we discuss considerations related to latency, particularly in the choice of integration model, and explore runtime updates through the S4T plugin mechanism. This strategy equips researchers and operators with a unified framework to administer IoT nodes, accelerating the adoption of edge computing and fostering reproducible experimentation in the compute continuum.