Generative-AI for AI/ML Model Adaptive Retraining in Beyond 5G Networks

TL;DR

This work tackles AI/ML model performance degradation in Beyond 5G networks by introducing a Generative-AI–based predictive retraining framework that uses VAE and GAN to generate traffic resemble observed data and KS tests to decide when retraining is needed. The approach is integrated into the O-RAN architecture (Non-RT and Near-RT RIC) and evaluated on two fronts: QoS prediction on the OSC platform and a network-slicing scenario using the Colosseum dataset, where it outperforms threshold and LOF baselines in terms of retraining trigger latency (MRTT) and re-training replacement time (MRPT). Key contributions include (i) a system model for Gen-AI–driven retraining within O-RAN, (ii) two KS-test–based predictive algorithms using VAE and GAN, and (iii) comprehensive QoS and NS evaluations showing earlier and more reliable retraining to maintain SLA adherence. The findings suggest that Gen-AI–driven adaptive retraining can significantly reduce SLA violations and improve resource efficiency in dynamic B5G environments, with promising future directions in reinforcement learning and explainable AI for decision transparency.

Abstract

Beyond fifth-generation (B5G) networks aim to support high data rates, low-latency applications, and massive machine communications. Artificial Intelligence/Machine Learning (AI/ML) can help to improve B5G network performance and efficiency. However, dynamic service demands of B5G use cases cause AI/ML model performance degradation, resulting in Service Level Agreements (SLA) violations, over- or under-provisioning of resources, etc. Retraining is essential to address the performance degradation of the AI/ML models. Existing threshold and periodic retraining approaches have potential disadvantages, such as SLA violations and inefficient resource utilization for setting a threshold parameter in a dynamic environment. This paper proposes a novel approach that predicts when to retrain AI/ML models using Generative Artificial Intelligence. The proposed predictive approach is evaluated for a Quality of Service Prediction use case on the Open Radio Access Network (O-RAN) Software Community platform and compared to the predictive approach based on the classifier and a threshold approach. Also, a realtime dataset from the Colosseum testbed is considered to evaluate Network Slicing (NS) use case with the proposed predictive approach. The results show that the proposed predictive approach outperforms both the classifier-based predictive and threshold approaches.

Figures (11)

• Figure 1: VAE architecture
• Figure 2: GAN architecture.
• Figure 3: System model.
• Figure 4: Proposed Approach Workflow.
• Figure 5: QoS experimental setup.