Authors:
Nadezhda Varzonova
1
and
Melanie Schranz
2
Affiliations:
1
University of Klagenfurt, Klagenfurt, Austria
;
2
Lakeside Labs, Klagenfurt, Austria
Keyword(s):
Agent-Based Simulation, Edge-Fog-Cloud Continuum, Swarm Intelligence.
Abstract:
The rapid expansion of Internet of Things (IoT) devices and the increasing demand for data-intensive applications have driven research into distributed computing models such as the edge-fog-cloud continuum, which integrates real-time edge processing, collaborative fog layer management, and highly scalable cloud infrastructure. In this paper, we present COSMOS (Continuum Optimization for Swarm-based Multi-tier Orchestration System), a Python-based simulation framework built on the Mesa multi-agent library, designed for implementing and evaluating self-organizing scheduling algorithms in distributed systems. The framework provides modular components for swarm coordination dynamics, constraint-aware scheduling, and real-time optimization, enabling flexible experimentation with various scheduling scenarios. We designed the system architecture to be highly configurable and observable, allowing for flexible experiment setup and comprehensive data collection. Its extensible API enables rese
archers to implement and evaluate alternative orchestration strategies for resource allocation, facilitating the integration of both classical and learning-based scheduling approaches. We demonstrate the effectiveness of COSMOS through case studies on diverse scheduling paradigms, including nature-inspired approaches such as hormone-based orchestration and ant colony optimization. These studies showcase its capability to model and optimize real-world distributed computing scenarios.
(More)