Harnessing AI for fast, reliable networks

How to build, configure and support efficient, secure networks with help from artificial intelligence

Sponsored feature More than ever before, modern organizations from the smallest branch office to the largest shopping mall rely on digital services to serve their users.

That puts considerable strain on networks operation teams fighting to connect growing numbers of employees, devices and customers whilst simultaneously reducing downtime and maximizing operational efficiency. While on the one hand, they must have fast, scalable, and secure networks that serve everything from local area networks through to datacenters, and everything in between, on the other, they need to keep always-on access to mission critical applications and services and deliver a platform for future expansion and digital transformation as and when the business requires.

By automating some aspects of management using AI, network operations teams can do more to safeguard those ambitions without drowning in alerts and manual workflows. In 2024, Huawei launched its Xinghe Intelligent Network Solutions to address those requirements. It spans four domains: Campus network, Data center network, Network Security, and WAN.

Rethinking campus networking

Wi-Fi plays a big role in a campus network, and issues like congestion or poor data signal hand-off are toxic for users. Huawei launched its first five-radio Wi-Fi 7 access point as part of its Campus offering, which it says can deliver 4K video smoothly over up to 60 channels. This is where the company applies AI to help smooth out wireless traffic, even in high-volume networks. It uses AI-powered Coordinated Scheduling and Spatial Reuse (CSSR) technology to help queue traffic, identifies applications to determine those with the lowest latency requirements, and analyses the user experience on the fly to ensure that they're getting the best performance possible.

Meanwhile, a 'VIP Line' gives those with privileges can be created to deliver the most reliable access of all. Huawei's own research suggests this approach can achieve a 75% reduction in latency and a 30% increase in bandwidth for VIP users of its campus networking technology.

The campus solution also harnesses AI for fault resolution, using machine learning to learn more about network faults in order to help resolve them. Root-cause analysis is often the bane of the network technician. Many try to fix the network with one hand while fielding calls from angry users with the other. Huawei says that this system can fix 80 percent of wireless networking faults within minutes, enabling a beleaguered network tech to manage 10,000 users solo.

Elsewhere the Wi-Fi Shield feature improves security by scrambling wireless channels.. This complements the protection already provided by WPA (Wi-Fi Protected Access). Wi-Fi Shield combines with media access control security (MACsec), to encode network traffic throughout the entire network - including both wired and wireless portions

Wireless Channel State Information (CSI) technology monitors fluctuations in radio waves to detect human presence. This feature does double duty as an energy saving technology by monitoring the state of the environment and using it to adjust the energy devoted to the Wi-Fi network. Huawei reckons this can slice 30 percent from campus network energy consumption.

Modernizing networks in Ekurhuleni and Dragon City

The City of Ekurhuleni implemented Huawei's Wi-Fi 7 campus network across over 2,000 square kilometers to support an electricity network that serves over two million people in South Africa. The urban authority had been using Wi-Fi 5 access points, but had been experiencing various problems including low bandwidth, limited coverage, and security issues. The existing network had also suffered multiple cyber attacks which led to information leaks, and lacked a unified network management platform, meaning that operations staff had to find and fix network faults manually, one at a time.

Huawei installed its CloudEngine 10GE-capable switches to coordinate traffic through Ekurhuleni’s wired and wireless networks, connected to AirEngine Wi-Fi 7 access points running Wi-Fi Shield to avoid packet capture by would-be eavesdroppers. The company enabled operators to manage the network using iMaster NCE-Campus and iMaster NCE-CampusInsight tools. This gave the operations team real-time visibility into network quality and users' experiences of applications on the network. Huawei also says that the ability to remediate network faults in a single click improved operations efficiency by 90 percent.

Huawei also used its campus networking capabilities for Dragon City - one of the largest shopping centers in South Africa - to complete its 330,000 square meter Smart Mall. The first five phases of its construction had left the mall with an inadequate network which slowed down electronic payment times to as long as two minutes. Security problems were also rife, with cyber criminals found to be stealing merchants' customer information. It was difficult for operators to manage these issues, as they could only deal reactively with complaints as they came in.

Huawei installed its High-Quality 10Gbps Campus Network running through Cloud Engine switches, again connected to AirEngine access points running Wi-Fi 7 and Wi-Fi shield. The iMaster NCE-Campus+ and iMaster NCE-CampusInsight tools allowed Dragon City’s operations team to take a more proactive approach when resolving network issues, cutting the time to find the causes of faults down to minutes. Overall, the network slashed waiting times for electronic payments and improved service efficiency by 80 percent, Huawei says.

Fabric, WAN, and SASE solutions

While the Campus products focus on the broader user environment, Huawei's Xinghe Intelligent Fabric Solution concentrates on high-speed network fabrics within datacenters. This system marries the management of network and security devices into a single interface, and like other domains in the Huawei rollout it applies AI heavily to datacenter traffic management.

It monitors application experience and conducts fast root cause analysis to isolate faults while using its iMaster NCE-Fabric device to automate network configuration, for example. And to help combat outages, the fabric technology uses fast switches at the link, device, and network levels to keep network services running in the event of a problem.

Huawei Xinghe Intelligent Fabric Solution has three components. The first is what the company refers to as the ‘AI Brain’; a topological network map capability along with NetMaster that together help to automatically resolve network issues. This layer of the solution also now features network security integration that automatically deployed network security policies.

The second layer is the AI Connection network load balancing algorithm that optimizes network path planning to help eliminate performance bottlenecks. Huawei believes this can improve efficiency in AI training (which is highly data intensive) by over 10 percent. AI Connection also includes a technology that Huawei calls iReliable, which automatically switches over at the link, device, and network levels to keep connections up and running continually.

Finally, AI NEs (network elements) use the company’s CloudEngine datacenter switches and StarryLink optical modules to visualize module packet loss and delay. This is also a security tool, points out Huawei, bringing features such as group isolation capabilities to the table.

The growth of AI technologies, government digitalization, IP-based production networks, and technology innovation in finance is putting pressure on WANs to meet latency standards. Huawei's newly upgraded Xinghe Intelligent WAN solution consists of AI routers, AI new connections, and AI new brains. It injects AI capabilities into IP networks to improve their capabilities and performance.

AI routers can accurately identify encrypted flows to maintain reliable performance metrics for IP networks. The AI new connections implement flow-level scheduling, ensuring quality of service levels on a per-application basis. The AI new Brains component helps enterprises simulate changes, optimize network performance in near-real time, and resolve 90% of network faults automatically, improving O&M efficiency.

Xinghe Intelligent WAN uses intelligent optical modules and network slicing technologies to improve service reliability and ensure quality of service in production networks.

With Huawei's product portfolio now covering so many facets of networking, from campus through to SD-WAN, a single-vendor solution to securing all facets of the network from the edge to the center has come within reach, says the company.

With the Huawei Xinghe Intelligent Unified SASE Solution, AI is integrated into network security detection - specifically 18 small AI models built to identify help identify threats more quickly and accurately, and achieve as high as 91 percent accuracy in detecting unknown threats. Additionally, AI models are embedded into security devices to process AI inferencing workloads in microseconds.

The SASE solution previously focused on the network edge. But in 2024 the company embedded its HiSec Endpoint endpoint detection and response (EDR) agent as a native component to support individual laptops, desktops, and mobile endpoints. HiSec Endpoint now extends security protection from the branch level down to the device level says Huawei, offering real-time behavior analysis, threat detection, and automated remediation as a baked-in service. That enables it to detect threats like ransomware, fileless malware, and privilege escalation attacks on devices, even when they're offline or disconnected from the corporate network. It also helps IT to respond more quickly to these incidents by corresponding alerts on the endpoint with network anomalies. Huawei expects to be able to reduce network security risks by up to 75 percent in branch office networks as a result.

Keeping back a flood of attacks around the world

Huawei's network security capabilities have already helped major clients including Mexico's Caminos y Puentes Federales de Ingresos y Servicios Conexos (Federal Roads and Bridges and Related Services, CAPUFE). It’s a government agency that operates federally-owned roads and bridges and has over a dozen branch offices, all of which must stay in regular contact with headquarters.

CAPUFE was concerned about what looked like a rising tide of cyber attacks on government organizations around the world, including ransomware, and wanted to upgrade its network to defend itself. With only a small security operations team in-house, it didn’t have the budget to protect branch-level networks – a significant risk since branch devices cached some content from the head office.

Huawei hardened the network for the agency by installing a system at headquarters to manage branch firewalls and analyze branch security and service information to identify and locate any threats. This was done by deploying HiSec-equipped USG firewalls, improving efficiency by an estimated 50 percent according to Huawei. The firewalls block over 500 threat events daily, and integrating multiple functions reduces costs by 30 percent, the vendor says.

Halfway around the world, Huawei also helped Severius, a datacenter operator in the Netherlands, to solve an ongoing problem with near-constant distributed denial of service (DDoS) attacks on its Amsterdam hosting facility.

Serverius decided to turn the problem into an opportunity by developing security as value-added service, which was a relatively rare offering in the Netherlands datacenter community. To do this, it needed a set of customizable services that it could integrate with its own operations platform. It chose Huawei's anti-DDoS and operations services as the foundational platform for this offering, and says that it blocked over 40,000 DDoS attacks in the first operation, successfully fending off malicious traffic volumes which amounted to hundreds of Gbit/s.

In the Philippines, Huawei's SASE system also helped direct sales company Personal Collection to secure its 650-branch network. The retailer, which is heavily involved in online shopping, faced security risks from multiple branch offices accessing cloud-hosted SaaS applications. Its operations team had to handle those risks manually, and the workload was increasing as services expanded and traffic volumes spiked traffic, impacting performance and degrading its security protection.

Personal Collection needed both a fast network and the flexibility to quickly set up new sites as more branch offices were added, but concluded that it didn't have a central management platform which could support those ambitions. That flexibility requirement could be satisfied by an SD-WAN, but the retailer’s legacy network security infrastructure didn’t support this or the firewall it wanted.

The company instead deployed Huawei’s Xinghe Intelligent Unified SASE Solution, which supports SD-WAN connections and centralized management of network and security devices, including remote security gateways at the branches managed by iMaster NCE-Campus. The gateways handle automated service provisioning and AI-powered traffic optimization to help deliver the best user experience. And the built-in firewall capabilities use AI to conduct machine learning-based analysis of endpoints connecting to the system, processing traffic at speed up to 700Mbit/sec.

Huawei claims that this can detects 10 percent more threats than rival products, spotting 91 percent of unknown threats. The gateway also works out of the box, can be up and running at each branch in minutes and includes the capability to map and manage thousands of sites, it says.

With administrative and industrial networks increasingly operating alongside each other, Huawei hopes to bring these operational efficiencies to all aspects of public and private sector networks - whether that involves processing customer payments quickly in-store or detecting and stopping malicious traffic.

Sponsored by Huawei.