Mapping the Landscape of Generative AI in Network Monitoring and Management

TL;DR

This paper maps how Generative AI, including transformers, diffusion models, and state-space approaches, can be applied to network monitoring and management across five use cases: traffic generation, traffic classification, intrusion detection, log analysis, and digital assistance. It provides a taxonomy of GenAI architectures, discusses training regimes, and explains input representations (Datagram-to-Token, Datagram-to-Image) needed to handle network data. The work aggregates datasets and platforms, emphasizes reproducibility, and offers a model-centric view of adaptations required for NMM tasks, while comparing open and closed models and highlighting code availability. It also outlines practical limitations—trust, robustness, compute demands, and privacy concerns—and proposes future directions like real-time efficiency, multimodal integration, interpretability, and secure deployment to guide safe, scalable adoption in real networks.

Abstract

Generative Artificial Intelligence (GenAI) models such as LLMs, GPTs, and Diffusion Models have recently gained widespread attention from both the research and the industrial communities. This survey explores their application in network monitoring and management, focusing on prominent use cases, as well as challenges and opportunities. We discuss how network traffic generation and classification, network intrusion detection, networked system log analysis, and network digital assistance can benefit from the use of GenAI models. Additionally, we provide an overview of the available GenAI models, datasets for large-scale training phases, and platforms for the development of such models. Finally, we discuss research directions that potentially mitigate the roadblocks to the adoption of GenAI for network monitoring and management. Our investigation aims to map the current landscape and pave the way for future research in leveraging GenAI for network monitoring and management.

Figures (5)

• Figure 1: Survey organization.
• Figure 2: Timeline of GenAI development: while introducing VAN-based and GAN-based solutions, this work primarily focuses on developments from the Transformer onward.
• Figure 3: Overview of the general workflow of Transformer-based models: (a) Full Encoder-Decoder, (b) Encoder-Only, and (c) Decoder-Only.
• Figure 4: Overview of the general workflow of Diffusion Models, including (a) forward and (b) reverse diffusion processes.
• Figure 5: Overview of the general workflow of Mamba: the Convolutional (Conv) layer extracts relevant features from input data, focusing on spatial or temporal patterns; the Selective SSM layer filters and selects the most relevant latent states from the extracted features.