Integrating Reinforcement Learning and Cognitive AI for Low-Latency and Resilient High-Speed Network Optimization

View at publisher Summarise with VIRAT
Authors: Guru Kesava Dasu Gopisetty, Balajee Maram, Manikonda Srinivasa Sesha Sai
Cited by: 0
References: 10

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

High-speed networks are handling a growing amount of information that requires adaptive optimization at timelines of a few milliseconds in order to satisfy ultrareliable low-latency communication bassoses (URLLC). Classical rule-based orchestration and static optimization tend to fail in the face of partial observerability and dynamic workloads resulting in tail-latency failures and sluggish response to disruption. The proposed hybrid framework combines the reinforcement-based learned behavior with cognitive AI reasoning and enables resilience in intent driven control of networks. We formalize the problem as an optimization of p95/p99 latency, mean time to recovery and control overhead under firm capacity and SLA constraints. We apply the use of multi-agent policies edge deployment combined with the coordination of dialectical inference to support real-time adaptation. Experimental evaluations on a variety of topologies and workloads report consistently reductions of 30+ percent in p99 tail latency and faster recovery over heuristic baselines. The system is published with reproduced artifacts, such as code, datasets, and containerized implementations, in order to achieve a sense of transparency and to have a verifiable progress.

Fields of Study & Subjects

Software-Defined Networks and 5G Network Traffic and Congestion Control Advanced Optical Network Technologies

Publication details

Journal / source
Biennial International Conference on Nascent Technologies in Engineering (ICNTE)
Pages
1–8
Publication date
16 January 2026
Language
EN