Article: Design and performance analysis of multipath DQN routing mechanism in high-dynamic power optical networks Journal: International Journal of Information and Communication Technology (IJICT) 2026 Vol.27 No.77 pp.20 - 39 Abstract: High-dynamic power optical networks are characterised by frequent topology changes and resource constraints, which challenge traditional routing mechanisms in meeting quality-of-service requirements. This paper proposes a multi-path deep Q-network (DQN) routing mechanism tailored for such environments. By integrating multi-path transmission with deep reinforcement learning, the mechanism optimises delay, packet loss rate, and load balancing through a multi-objective reward function. Built upon a layered graph model of the optical network, the approach incorporates a multi-path selection strategy and employs DQN for intelligent routing decisions under dynamic conditions. Simulation results demonstrate that the proposed mechanism effectively reduces end-to-end delay and packet loss under high traffic loads, while improving resource utilisation and network robustness, making it suitable for efficient data transmission in high-dynamic power optical networks. Inderscience Publishers - linking academia, business and industry through research

Title: Design and performance analysis of multipath DQN routing mechanism in high-dynamic power optical networks

Authors: Jing Zhang; Lin Guo; Shaolei Chen; Jiang Xie; Xuedong Sun; Xiaohang Li

Addresses: Power Dispatch Control Center, State Grid Yibin Electric Power Supply Company, Yibin, Sichuan, China ' Power Dispatch Control Center, State Grid Sichuan Electric Power Company, Chengdu, Sichuan, China ' Power Dispatch Control Center, State Grid Sichuan Electric Power Company, Chengdu, Sichuan, China ' Power Dispatch Control Center, State Grid Yibin Electric Power Supply Company, Yibin, Sichuan, China ' Power Dispatch Control Center, State Grid Yibin Electric Power Supply Company, Yibin, Sichuan, China ' Power Dispatch Control Center, State Grid Yibin Electric Power Supply Company, Yibin, Sichuan, China

Abstract: High-dynamic power optical networks are characterised by frequent topology changes and resource constraints, which challenge traditional routing mechanisms in meeting quality-of-service requirements. This paper proposes a multi-path deep Q-network (DQN) routing mechanism tailored for such environments. By integrating multi-path transmission with deep reinforcement learning, the mechanism optimises delay, packet loss rate, and load balancing through a multi-objective reward function. Built upon a layered graph model of the optical network, the approach incorporates a multi-path selection strategy and employs DQN for intelligent routing decisions under dynamic conditions. Simulation results demonstrate that the proposed mechanism effectively reduces end-to-end delay and packet loss under high traffic loads, while improving resource utilisation and network robustness, making it suitable for efficient data transmission in high-dynamic power optical networks.

Keywords: high-dynamic optical networks; multi-path routing; deep Q-network; DQN; deep reinforcement learning; load balancing.

DOI: 10.1504/IJICT.2026.154958

International Journal of Information and Communication Technology, 2026 Vol.27 No.77, pp.20 - 39

Received: 28 Nov 2025
Accepted: 16 Feb 2026
Published online: 20 Jul 2026 *

Title: Design and performance analysis of multipath DQN routing mechanism in high-dynamic power optical networks

Keep up-to-date