SchEdge: A Dynamic, Multi-agent, and Scalable Scheduling Simulator for IoT Edge
Ali Hamedi, Amirali Ghaedi, Amin Soltanbeigi, Athena Abdi
TL;DR
SchEdge introduces a lightweight, modular Python-based simulator for dynamic IoT edge scheduling that supports online, multi-agent scheduling with cycle-accurate execution. By separating a workflow layer (environment, scheduler, state) from a dataflow layer (data generator, WindowManager, Preprocessor), it achieves realism, configurability, and scalability for evaluating online and RL-based scheduling strategies. Experimental results demonstrate stable runtime performance, efficient resource usage, and robustness under dynamic device churn, as well as favorable scalability compared to Java-based peers. The framework enables rapid experimentation with diverse topologies, workloads, and adaptive scheduling schemes, offering a practical platform for advancing edge computing research. Future work includes larger-scale datasets, latency-aware and predictive scheduling, and interfacing with real IoT devices and hybrid edge-cloud systems.
Abstract
This paper presents a dynamic, adaptive, and scalable framework for simulating task scheduling on the edge of the Internet of Things called "SchEdge". This simulator is designed to be highly configurable to reflect the detailed characteristics of real-world IoT. This framework focuses on online task scheduling and its multi-agent nature provides multiple schedulers to implement various scheduling schemes in parallel. SchEdge consists of two main parts the workflow and data flow. The workflow manages the schedulers' interaction with the application and environment while the data flow deals with the input application and its preprocessing. Combining these sections provides scalability, adaptability, and efficiency in the SchEdge. To validate the efficiency of this simulator, several experiments categorized as behavioral and technical analysis are performed to show its efficiency, scalability, and robustness.