A Cloud Continuum Research Infrastructure for Distributed CPS Experimentation

This repository contains the research infrastructure and experimental artifacts presented in the paper "A Cloud Continuum Research Infrastructure for Distributed CPS Experimentation".

Abstract

Cloud continuum applications require experimental environments capable of combining heterogeneous Edge, Fog, Cloud, and high-performance computing resources while preserving reproducibility, observability, and control over distributed deployments. This work presents a two-level reference architecture built on top of the SLICES Cloud Continuum Blueprint. The approach separates the research-infrastructure layer (managing distributed resources) from the application layer (organizing cyber-physical workflows according to an Edge-Fog-Cloud pattern).

The architecture enables researchers to deploy, customize, and compare alternative control and monitoring strategies over a programmable infrastructure substrate, treating placement, timing, and data provenance as first-class experimental concerns.

Architecture: A Two-Level Reference Approach

As described in the paper, this infrastructure follows a two-level reference architecture designed for Cloud Continuum experimentation:

1. Research Infrastructure Layer (Layer 1): Manages the distributed resources (Edge, Fog, Cloud) through a programmable substrate. It leverages Stack4Things for orchestration and Crossplane for Kubernetes-native resource management.
2. Application Layer Workflow (Layer 2): Organizes Cyber-Physical workflows according to an Edge-Fog-Cloud pattern. This layer treats placement, timing, and data provenance as first-class experimental concerns.

Repository Structure

• core/ (Research Infrastructure Layer):
  • stack4things-improved/: Core orchestration framework and management plane.
  • crossplane-provider/: Declarative resource management for distributed infrastructures.
• experiments/scenarios/ (Application Layer Workflows):
  • airwatch/: Urban monitoring pipeline (5-stage workflow: Ingest, Normalize, Detect, Aggregate, Alert).
  • rec-simulation/: Renewable Energy Community (REC) management focusing on Digital Twin synchronization.
• experiments/runner/ & tools/: Orchestration and metric extraction for systematic validation (40 automated runs).
• docs/: Replication protocols and research-ready guides.

Validated Use Cases

The infrastructure is validated through two representative scenarios:

1. Renewable Energy Community (REC): Distributed Digital Twin coordination and time-window-based energy control.
2. AirWatch: An urban monitoring pipeline focused on anomaly detection and cloud-side aggregation.

Both workloads were evaluated through a systematic campaign of 40 automated runs, comparing virtualized (POD) and physical (RASP) edge deployments over the geographically distributed SLICES infrastructure.

Getting Started

To replicate the experiments or explore the infrastructure, please refer to the following guides:

• Replication Guide: Step-by-step instructions to run the experimental campaigns.

---

For more details on the architecture and performance evaluation, please refer to the full paper.