Why complex networks need additional support

Network ops, maintenance issues don’t solve themselves – AI-powered automation can help deal with them

Sponsored feature Organizations today face considerable challenges when it comes to operating and maintaining complex network infrastructures. It’s an issue which stems in part from the historical deployment of different radio access technologies in campus, office, and datacenter environments. And it’s made worse by the pressing need to deliver more bandwidth to a wider variety of end user devices, often in densely populated built-up areas prone to signal interference.

Campus networks can experience congestion because there are too many devices attempting to connect to the same signal, for example. The resulting saturation often impacts browsing speed and application latency. It’s usually a similar situation in enterprise offices, datacenters and server farms, especially where digital transformation and a greater reliance on cloud-hosted applications services have led to a rise in device numbers and data traffic.

Those tasked with the operation and maintenance (O&M) side of wireless network management in these locations have additional challenges. They have to maximize network uptime in different environments – including office buildings, conference rooms and libraries for example - each of which has its own structural characteristics and bandwidth needs.

They may also have to cope with poor coverage in out of the way locations such as warehouses and remote offices. They might be dealing with buildings that extend to multiple levels, or that have thick walls that lead to signal dead zones. All of which means that radio calibration has to accommodate multiple factors, including optimal access point (AP) siting, sources of environmental interference, power management, roaming and bandwidth allocation.

Dealing with all of this has traditionally pushed organizations into recruiting personnel with a high level of O&M expertise – people who can bring their rich professional knowledge to bear alongside a deep familiarity with diverse network devices. This level of experience is difficult and expensive to find though. And even when present hard-pressed staff can still be hampered by fragmented data from multiple sources, inefficient log analysis, and delayed fault identification and resolution. Which means that when network faults occur, fixes might take several days to be addressed rather than minutes or hours, adversely impacting both the end-user experience and service quality for the organization’s customers.

Embedding automated intelligence into O&M

Huawei believes that building greater automation and artificial intelligence into analytics systems can go some way to helping solve these network O&M management issues. To that end the company launched NetMaster. It’s designed to emulate the experience of more than 10,000 experts and uses “advanced semantic understanding capabilities” to provide AI-assisted “knowledge Q&A, interactive service analysis and informed decision-making” to improve network O&M efficiency by a factor of around 100, says the company.

A year later saw Huawei add further capabilities, using MWC Barcelona 2025 to launch an upgraded NetMaster Network Agent which is designed to provide even greater levels of network assurance and fault handling. To do so, the new version of the product featured two new AI-powered components, each supporting a different networking environment.

OptimSpirit

The first AI agent, OptimSpirit, is an automated network optimization agent designed to overcome many of the problems inherent in wireless networks. It employs a “multi-objective decision-making algorithm” to optimize coverage, interference, bandwidth and load, detecting incidents in milliseconds to cut optimization times from hours to minutes and improve network traffic usage. It’s been customary for network optimization solutions to separately address coverage, roaming, capacity and interference issues. But these are traditional methods only meant to deal in part with the coverage problems caused by interference, according to Huawei. OptimSpirit’s “multi-objective decision making” is engineered specifically for complex campus environments, using automatic detection to make sure most wireless faults can be addressed within minutes, simplifying O&M processes and ensuring user experiences, says the company.

Students need education, and education needs the internet and network access to services including online courses and videos. That sort of content can consume a lot of bandwidth. But campus networks are notorious for environmental characteristics such as short distances between dormitories, dense student populations and the existence of numerous obstacles which can block Wi-Fi signals.

All of this has a habit of translating to slow, unreliable network speeds that lead to end user complaints and subsequently lost business as students go elsewhere to study. OptimSpirit can go some way to getting around those issues by helping to identify potential performance risks in advance and applying automated policies to optimize bandwidth and coverage where needed to eliminate problems before they arise.

Reducing the average handling time for network faults on campus is also an important goal for NetMaster. For example, aside from optimizing wireless campus network performance AI also helps the product to resolve around 80 percent of wireless campus network faults with no human intervention, says Huawei. This means that a single network engineer might well be enough to keep a campus of tens of thousands of users up and running and fully maintained. And unlike a human, this kind of AI-centric intelligent O&M is, of course, operational round the clock.

AssurSpirit

NetMaster’s other new AI agent, is a scenario-specific troubleshooting agent built to handle the unique challenges of operating networks in a datacenter environment. One of these challenges is handling the many alarms that incidents can trigger in such a complex, data-intensive networking environment.

In its research, Huawei identified over 100 different types of alarm, each with their own handling processes, which are buried in over a thousand pages of documents. Not only does this complexity make alarm handling time consuming for datacenter network administrators, but it also creates risk of error in what has until now been a manual process.

AssurSpirit changes that alarm handling process by automatically generating alarms. It autonomously mines and analyzes data to conduct root cause analysis and make informed decisions, helping it to resolve common issues such as interface anomalies, traffic irregularities and entry exceptions.

The agent also offers troubleshooting suggestions pulled from an LLM containing customers' proprietary expert knowledge bases. And it is built with constant improvement in mind; users can update existing “chains-of-thought” using natural language, which can significantly reduce development and maintenance costs.

These capabilities can help customers cut mean time to repair (MTTR) for the datacenter network to under five minutes, calculates Huawei.

Why AI matters to the future of connectivity

Huawei has been hard at work integrating AI into networks for a number of years now, developing AI applications designed to transform traffic consumption habits and help create new business models for their owners. It’s been over 10 years since the company started using AI algorithms to find anomalies in network performance and identify attacks on networks. Later it worked on the use of neural networks to predict traffic flow on the network.

Now, NetMaster is a pillar in the company’s intelligent network O&M solution, which it calls its ‘Autonomous Driving Network’ or ADN. In the same way that modern self-guided vehicles fall into different levels of autonomy, the ADN offers different levels of autonomous operation for datacenter and enterprise networks,. These include automatic identification of network risks before they become issues for O&M, along with automated fault location and service provisioning.

Huawei and IEEE jointly published an L4 data center autonomous driving network white paper designed to offer high levels of network automation. Many of its customers are using ADN, especially in the financial industry and education sector. Beijing’s Tsinghua University for example is one of the leading academic institutions in China. It uses ADN to manage the whole of its campus network, supporting around 16,000 network devices.

Tsinghua University’s experience suggests that adding extra intelligence to network infrastructure can quickly provide return on investment in the form of added efficiency and reduced costs. With the latest additions to its NetMaster Network Agent, Huawei has again shown its commitment to harnessing AI to help meet the expanding network O&M demands of organizations everywhere.

Sponsored by Huawei