Reinforcement Learning-Based Deep Q-Network Traffic Routing for IoT-Enabled Smart Networks with QoS Constraints
Summary
No summary yet. VIRAT can read this paper and write a plain-language summary of its aim, methods, key findings and limitations.
Summarise with VIRAT Free to try with your Science Catalogs account.Abstract
The radical increase in the numbers of Internet of Things (IoT) devices in smart networks has posed a considerable challenge in traffic and scalability management, and Quality of Service (QoS) provisioning. The conventional routing protocols are also static and do not respond to the dynamic traffic loads, energy and heterogeneous devices. In an effort to overcome these shortcomings, this paper presents a Reinforcement Learningbased Deep Q-Network (DQN) routing architecture that models traffic routing and makes it a Markov Decision Process (MDP). The framework uses network state data such as latency, queue occupancy, residual energy and packet loss to make adaptive routing decisions. A QoS rewarding functionality was created to trade off delay reduction, maximization of the ratio of packets delivered, and energy efficiency. The results of the simulation prove that the proposed DQN framework is superior to the Shortest-Path and the QoS-aware protocols as it significantly reduces the latency by 23 %, increases the ratio of packet delivery rates by 18 percent, and minimizes the energy consumption by 15$\%$. Besides, both ROC and confusion matrices analysis prove that the approach is reliable and robust. It is possible to note that the results proved that reinforcement learning-based routing is a prospective paradigm enabling self-optimizing and QoS-awarded IoT-enabled smart networks.
Fields of Study & Subjects
Energy Efficient Wireless Sensor Networks
Network Traffic and Congestion Control
Software-Defined Networks and 5G
Publication details
- Journal / source
- International Conference on Intelligent Sustainable Systems (ICISS)
- Pages
- 272–277
- Publication date
- 4 March 2026
- Language
- EN