DeepCAS: A Deep Reinforcement Learning Algorithm for Control-Aware\n Scheduling

2018/03/08 by Burak Demirel, Demirel, Burak, Arunselvan Ramaswamy +5
Computer Science · Engineering · #Age of Information Optimization #Artificial Intelligence (cs.AI) #FOS: Computer and information sciences #FOS: Electrical engineering #Real-Time Systems Scheduling #Smart Grid Security and Resilience #Systems and Control (eess.SY) #electronic engineering #information engineering

paper · pdf · doi:10.48550/arxiv.1803.02998

openalex publication_date 2018/03/08 · openalex created_date 2022/08/06 · openalex updated_date 2026/07/28

Abstract

We consider networked control systems consisting of multiple independent\ncontrolled subsystems, operating over a shared communication network. Such\nsystems are ubiquitous in cyber-physical systems, Internet of Things, and\nlarge-scale industrial systems. In many large-scale settings, the size of the\ncommunication network is smaller than the size of the system. In consequence,\nscheduling issues arise. The main contribution of this paper is to develop a\ndeep reinforcement learning-based \control-aware scheduling\n(\DeepCAS) algorithm to tackle these issues. We use the following\n(optimal) design strategy: First, we synthesize an optimal controller for each\nsubsystem; next, we design a learning algorithm that adapts to the chosen\nsubsystems (plants) and controllers. As a consequence of this adaptation, our\nalgorithm finds a schedule that minimizes the \control loss. We present\nempirical results to show that \DeepCAS finds schedules with better\nperformance than periodic ones.\n

Related

• Deep reinforcement learning for scheduling in large-scale networked control systems
• Control-Aware Scheduling for Low Latency Wireless Systems with Deep\n Learning
• Semantic-aware Transmission Scheduling: a Monotonicity-driven Deep Reinforcement Learning Approach
• Deep Reinforcement Learning for Wireless Scheduling in Distributed Networked Control
• Neural Heterogeneous Scheduler