D4.2 - Design, Development and Evaluation of the Cloud-to-Edge Simulator

This deliverable presents the design, development, and evaluation of a comprehensive Cloud-to-Edge simulation environment that enables the modelling and assessment of decentralised, self-organising orchestration mechanisms within the Swarmchestrate project. Addressing the limitations of traditional centralised orchestration frameworks, Swarmchestrate adopts a swarm-based, application-centric approach to resource management across the dynamic and heterogeneous Cloud-to-Edge continuum.

The simulation environment, built upon and extending the DISSECT-CF-Fog simulator, plays a crucial role in validating the project's innovative concepts before deployment. We chose to use the DISSECT-CF-Fog simulator because it offers a flexible, modular, and energy-aware simulation environment that is well-suited for modelling complex edge, fog and cloud computing scenarios. Key achievements include:

• Extension of DISSECT-CF-Fog to support TOSCA (Topology and Orchestration Specification for Cloud Applications) for describing applications and infrastructure.
• Implementation of sophisticated models of components involved in the deployment phase, including capacity modelling, self-organisation, and dynamic connectivity.
• Development and refinement of broadcasting algorithms for efficient resource selection and allocation during application deployment.
• Integration of time-series analysis and predictive models to anticipate system behaviour during runtime, supporting dynamic task offloading decisions for more efficient resource usage.
• Implementation and evaluation of distributed algorithms based on Ant Colony Optimisation (ACO) for decentralised orchestration, demonstrating enhanced
  responsiveness and improved energy efficiency in managing large-scale applications.

The simulator has been evaluated using realistic scenarios inspired by real-life use cases, including high-density IoT (Internet of Things) deployments and applications running on edge devices. The results validate the effectiveness of the decentralised orchestration approach, showing reductions in deployment time, energy consumption, and operational costs, while maintaining high system performance even under increased load and complexity.

This work forms the foundation for the Digital Twin concept in the second phase of Swarmchestrate, where simulated insights and predictions will feed directly into live system management. By offering a robust and extensible simulation platform, this deliverable significantly advances the project's goals of enhancing orchestration in the Cloud-to-Edge continuum.