IPv6 Network Dynamic Label Distribution and Fast Path Reconfiguration Mechanism Based on Deep Reinforcement Learning
Main Article Content
Abstract
Against the background of large-scale commercial deployment of IPv6, SRv6 (Segment Routing over IPv6) has become the mainstream forwarding technology. Traditional static label distribution suffers from low resource utilization, while passive fault path reconstruction has high delay and cannot adapt to the graded QoS requirements of services. To realize intensive label management and millisecond-level network self-healing, this paper proposes a joint management and control mechanism based on the time-series weighted improved PPO (RW-PPO) algorithm. Relying on the IPv6 extension header to construct a multi-dimensional network decision space, the algorithm optimizes the weight of training samples and solves the convergence oscillation problem of native PPO under dynamic traffic. A hierarchical distributed label distribution architecture is built to realize on-demand allocation, recycling and aggregation of labels. A three-level reconstruction architecture consisting of fault prediction, path prestorage and local switching is constructed to reduce the overhead of network-wide routing recalculation. Comparative experiments are carried out based on the Mininet simulation platform. The results show that compared with the traditional OSPF-SR scheme and the native PPO scheme, the proposed mechanism reduces the label occupancy rate by 18.72%, achieves a single-link fault reconstruction delay as low as 9.2 ms, and significantly reduces the interruption duration of high-priority services, which is suitable for the operation and maintenance scenarios of operators’ IPv6 heterogeneous service networks.
Downloads
Article Details
This work is licensed under a Creative Commons Attribution 4.0 International License.
Authors who publish with this journal agree to the following terms:
- Authors retain copyright and grant the journal right of first publication with the work simultaneously licensed under a Creative Commons Attribution License that allows others to share the work with an acknowledgement of the work's authorship and initial publication in this journal.
- Authors are able to enter into separate, additional contractual arrangements for the non-exclusive distribution of the journal's published version of the work (e.g., post it to an institutional repository or publish it in a book), with an acknowledgement of its initial publication in this journal.
- Authors are permitted and encouraged to post their work online (e.g., in institutional repositories or on their website) prior to and during the submission process, as it can lead to productive exchanges, as well as earlier and greater citation of published work (See The Effect of Open Access).
References
J. Gao, Y. Wang, and R. Liu, “Analysis on deployment and application of data center network based on IPv6,” Electron. Technol., vol. 54, no. 9, pp. 54–55, 2025.
M. Mikami, “Segment Routing IPv6 Based Low Latency Mobile Network for Vehicular Communications towards Automated Driving Era: Special Section on Emerging Communication Technologies in Conjunction with Main Topics of ICETC2024,” IEICE Trans. Commun., vol. E109-B, no. 2, pp. 147–160, 2026, doi: 10.23919/TRANSCOM.2025CEI0002.
Y. Tian, Z. Wang, X. Yin, et al., “Traffic Engineering in Partially Deployed Segment Routing Over IPv6 Network With Deep Reinforcement Learning,” IEEE/ACM Trans. Netw., vol. PP, pp. 1–14, 2020, doi: 10.1109/tnet.2020.2987866.
J. Qu, Y. Yang, L. Gao, et al., “Deep Learning-Based Fast Security Detection Method for Cross-Domain Access Behaviors in IPv6 Networks,” J. Circuits Syst. Comput., vol. 35, no. 17, 2026, doi: 10.1142/S021812662 6501343.
E. A. Ksimi, C. Leghris, S. Lafraxo, et al., “ICMPv6-based DDoS Flooding-Attack Detection Using Machine and Deep Learning Techniques,” IETE J. Res., vol. 70, no. 4, pp. 3753–3762, 2024, doi: 10.1080/03772063.2023.2208549.
E. O. E, A. Mohammed, H. Shady, et al., “Deep-Learning-Based Approach to Detect ICMPv6 Flooding DDoS Attacks on IPv6 Networks,” Appl. Sci., vol. 12, no. 12, p. 6150, 2022, doi: 10.3390/APP12126150.
C. Filsfils, P. Camarillo, J. Leddy, et al., “Segment Routing over IPv6 (SRv6),” RFC 8986, pp. 1–40, 2021, doi: 10.17487/RFC8986.
D. Ma, X. Zhao, C. Wang, et al., “Review on application and challenges of SRv6 technology in data forwarding plane,” J. Softw., vol. 37, no. 1, pp. 378–397, 2026, doi: 10.13328/j.cnki.jos.007422.
D. Yuan, H. Yu, and C. Jiang, “Cloud-Native SRv6-SDN Architecture for Traffic Inference and Rerouting,” in Proc. 2025 IEEE Int. Symp. Parallel Distrib. Process. Appl. (ISPA), 2025, doi: 10.1109/ISPA67752.2025.001 76.
Y. Zhao, Q. Guo, Y. Zhao, et al., “An improved deep q-network method for fault diagnosis based on deep reinforcement learning,” Measurement, vol. 279, p. 121779, 2026, doi: 10.1016/J.MEASUREMENT.2026.121779.
N. Alyaoui, B. A. Guiloufi, O. Baccouch, et al., “Intelligent Congestion Control and Scheduling in VANETs: An SDN-Based PPO Approach,” J. Netw. Syst. Manag., vol. 34, no. 4, p. 104, 2026, doi: 10.1007/S10922-026-10088-9.
B. Liang, F. Li, and N. Zheng, “AutoSRv6: Configuration Synthesis for Segment Routing Over IPv6,” IEEE J. Sel. Areas Commun., vol. 43, no. 2, pp. 473–483, 2025, doi: 10.1109/JSAC.2025.3528808.
P. Loreti, A. Mayer, and P. Lungaroni, “Implementation of Accurate Per-Flow Packet Loss Monitoring in Segment Routing over IPv6 Networks,” in Proc. IEEE High Perform. Switching Routing Conf. (HPSR), 2020, doi: 10.1109/HPSR48589.2020.9098988.
Z. Ding, Y. Huang, and H. Yuan, “Introduction to Reinforcement Learning,” in Lect. Notes Electr. Eng., 2020, doi: 10.1007/978-981-15-4095-0_2.
H. Tran-Dang and D. S. Kim, “DQN-Based Resource Allocation and Offloading (DQN-RAO) Algorithm for Multi-User Multi-Server MEC Network,” in Proc. 2026 Int. Conf. Artif. Intell. Inf. Commun. (ICAIIC), 2026, doi: 10.1109/ICAIIC68212.2026.11454139.
Q. Qiu, F. Zeng, and H. Yang, “Memory-Augmented Deep Deterministic Policy Gradient,” in Lect. Notes Comput. Sci., pp. 41–52, 2025, doi: 10.1007/978-981-97-8963-4_5.