YAIFS: Yet (not) Another Intelligent Fog Simulator: A Framework for Agent-Driven Computing Continuum Modeling & Simulation

Researchers Isaac Lera and Carlos Guerrero have unveiled YAIFS (Yet (not) Another Intelligent Fog Simulator), a significant evolution of the existing YAFS framework for modeling distributed cloud-edge systems. The core innovation is a paradigm shift: treating simulation not as a static, one-off tool but as an interactive, service-oriented environment. YAIFS achieves this through a layered architecture that exposes the simulation's internal state and controls via a unified API and service interface. This design fundamentally allows external entities—primarily AI agents—to observe, control, and modify the simulation's execution in real-time.

A central technical contribution is the integration of the Model Context Protocol (MCP) as a standardized communication layer between agents and the simulation. MCP provides a common set of tools, decoupling complex agent experimentation from the simulator's internals. The paper demonstrates this with two compelling scenarios: an LLM-based assistant that accepts natural language commands to control simulations, and a multi-agent system where AI agents monitor conditions and dynamically adapt resource placement decisions. This transforms simulations into programmable testbeds for adaptive, AI-driven experimentation.

The implementation, publicly available on GitHub, opens new avenues for research and development in the computing continuum. By providing a structured, agent-friendly interface, YAIFS lowers the barrier for applying advanced AI techniques to optimize complex distributed systems. It enables researchers to test how ensembles of agents can manage latency, bandwidth, and compute resources across cloud, fog, and edge nodes in response to unpredictable, real-world workloads.