COSCO
COSCO is an AI-based co-simulation and container orchestration framework for integrated Edge, Fog, and Cloud computing environments. Written in Python, it enables academics and practitioners to develop, simulate, test, and deploy scheduling policies across heterogeneous infrastructures. The framework introduces a gradient-based optimization approach using deep neural networks as surrogate functions combined with co-simulation to support intelligent decision making. COSCO consists of two main components: a discrete-event Simulator that runs standalone for testing algorithms against workloads like Bitbrains and Azure2017/2019 traces, and a Framework for deploying policies in real-time physical Fog and Cloud environments with real applications such as DeFog and AIoTBench workloads. Key features include hassle-free development of AI-based scheduling algorithms, seamless integration of scheduling policies with simulated backends, container migration support using the CRIU utility, and multiple deployment options i