AI Copilots Powering the Future of Autonomous Networks

By Dimitris Mavrakis | 4Q 2024 | IN-7569

Registered users can unlock up to five pieces of premium content each month.

Log in or register to unlock this Insight.

Mobile networks have been going through an automation transformation for years, which is now accelerated by Generative Artificial Intelligence (Gen AI). Closed-loop automation is the long-term goal of this process, with vendors like Huawei, Nokia, Ericsson, and ZTE leading the charge in developing intelligent copilots (or agents) to manage complex network Operations and Maintenance (O&M). Agents could be considered as domain-expert AI-driven tools to speed up complex network operations and maintenance and often utilize Large Language Models (LLMs) trained from years of vendor experience in operating cellular networks.

These networks will only become more complex as more devices, advanced network elements, and cloud-native aspects become mainstream. These solutions mark a shift from traditional, equipment-oriented O&M, which heavily relied on human-machine collaboration, to an intelligent and automated approach that integrates many aspects, including service provisioning with network resource optimization. Historically, business operations and network O&M have operated independently, often resulting in inefficiencies, slow response times, and even service disruptions. Intelligent agents are designed to integrate across domains, enabling seamless integration between business operations, network maintenance, and service provisioning. These new tools can achieve a closed-loop, fully autonomous network system, and the vendors mentioned above certainly have significant experience and data to train and implement these models.

For instance, ZTE has been pioneering a multi-agent system that integrates business operations with network resource management across multiple network domains. Its solution leverages years of expertise in network O&M and utilizes distributed computing at the edge, allowing the intelligent agents to make rapid, autonomous decisions based on real-time network conditions. By doing so, ZTE's system responds to service assurance needs within seconds, ensuring a seamless experience for customers, even in complex, multi-vendor environments.

AI agents will surely revolutionize mobile network management in several ways, especially if multiple agents are integrated across the network. In the long term, this will minimize human intervention at a supervisory capacity, minimize operational costs, and improve network performance with very quick maintenance operations. This will allow mobile operators to focus on creating new services and allow these agents to allocate resources in real time, adapt to user requirements almost instantly, and even optimize the network in near-real time during peak usage times.

The possibilities that come after deploying these agents are many, especially for new types of services that include network slicing and edge computing. For example, agents may be used to optimize slices in near-real time, adapt to user requirements and network conditions, and make faster, data-driven decisions about the network configuration. With edge computing, this process may be decentralized to reduce processing load on central systems and may even reduce latency by placing these optimization workloads near the end user. At the same time, the closed-loop nature of these systems will redefine how networks are managed. Autonomous workflows allow for continuous monitoring, diagnosing, and adjusting of network operations, reducing the time needed to detect and resolve faults, while also improving long-term performance.

Mobile operators must consider the transition to agent-driven and autonomous networks an essential step toward 6G and future-proofing their infrastructure. Leading vendors like Ericsson, Nokia, Huawei, and ZTE are providing the tools needed to build more agile, efficient, and intelligent networks. To ensure a transition that doesn't disrupt their network, operators should take a phased approach to implementing multi-agent systems. Starting with pilot programs in specific regions or for particular use cases can help operators fine-tune the system's performance before scaling it across the entire network. This will allow them to identify and address any challenges early on, ensuring a more seamless adoption of the technology. Moreover, a more important priority should be training and upskilling their workforce. While AI agents can take over many tasks traditionally performed by human operators, personnel will still need to oversee and manage these autonomous systems.

In conclusion, the shift to multi-agent integration represents a major step forward for the mobile networks and the introduction of advanced concepts such as network slicing and 6G. By embracing this new concept, operators can unlock new levels of efficiency, improve service quality, and future-proof their networks.