Agile Perception, Precise Decisions: AI‑Driven Smart Network Operations

During the 20th GOPS Global Operations Conference in Shenzhen (April 7‑8, 2023), Huawei Operations Lab expert Liu Yuliang delivered a talk titled “Agile Perception, Precise Decision‑Making: AI‑ and Data‑Driven Intelligent Operations for Communication Networks.”

The presentation covered telecom industry trends, emerging operational challenges, and a vision of shifting from network‑centric to business‑centric, human‑machine collaborative operations powered by artificial intelligence.

Liu emphasized that AI can empower operators to inject expert experience into data foundations, turning data into knowledge, enabling machines to learn, make autonomous decisions, and execute actions, thus creating a closed‑loop, self‑driving network management model.

In response to the hype around ChatGPT, his team has been developing a foundational model for network operations, proposing a large‑scale machine‑learning framework that combines data and domain knowledge to give the model “common sense” for solving operational pain points.

He concluded with a call to build more universal, efficient operation technologies and models to help customers achieve lower costs, higher network quality, and intelligent‑operation transformation.

Huawei leverages 30 years of ICT experience through its Operations Lab, investing in key technologies such as ChatOps, EDNS, Knowledge Center, AABD Pro, low‑code development, and hyper‑automation, accumulating extensive knowledge assets and real‑world deployments in MBB/FBB/NFV/5G and industry‑specific operations.

Key Solutions

ChatOps intelligent semantic interaction : Using a 400 k‑term telecom‑operations lexicon and 300 k samples plus proprietary AI algorithms, the system achieves over 85 % accuracy in natural‑language understanding, enabling multi‑turn Q&A and automated information retrieval.

EDNS (Expected Demand Not Served) : Multi‑dimensional data and “mechanism + data‑driven” modeling create a digital twin of the network, quantifying business impact and reducing service loss by 5‑10 %.

AABD Pro : Adaptive spatio‑temporal mining combined with expert knowledge identifies fault propagation and clusters, reaching up to 90 % clustering completeness and 93 % root‑cause identification accuracy, thus improving fault‑handling efficiency.

Multi‑vendor cross‑domain topology reconstruction : Deep protocol understanding and AI‑based reconstruction generate high‑precision physical, logical, and service‑layer topologies, supporting intelligent hazard detection and problem localization across domains.