.png){kind=icon}
.png){kind=icon}
Return
AICon: Agentic Intelligence for the Computing Continuum
.png){kind=icon}
PRE
AI
DOI：10.1109/MIC.2025.3645964.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
Computing continuum systems combine Internet of Things devices, edge, fog, and cloud resources to reduce latency, cost, and energy consumption in highly heterogeneous environments. However, existing simulators that rely on centralized architectures have tightly coupled components, fragile artificial intelligence/machine learning integrations, steep learning curves, and oversimplified agent models. These limitations make it difficult to develop realistic decentralized and resource-aware management approaches. We introduce AICon, an open source Python framework built on YAFS (Yet Another Fog Simulator) that overcomes these issues through a clean, modular three-layer design: infrastructure, application, and a novel management layer composed of autonomous agents. These agents are treated as first-class simulated workloads with realistic features: configurable work–sleep cycles, partial observability, and quality-of-service-aware actions (e.g., scaling application intensity or tuning node performance). AICon significantly lowers the barrier to entry, improves code reuse, and accelerates prototyping and validation of adaptive, decentralized management strategies that optimize service-level objectives under real-world constraints.
Keywords:
Energy consumption
Cloud computing
Costs
Codes
Computational modeling
Learning (artificial intelligence)
Computer architecture
Internet of Things
Observability
Tuning