AI and Wi-Fi 6: Driving the Revolution in Home Wi-Fi

As fixed networks enter the F5G (fifth generation) era, home Wi-Fi technology has also undergone a foreseeable upgrade cycle to support new radio technologies and remote management capabilities. But as the upgrade cycle accelerates, service providers realize that improving customers' home Wi-Fi experience has become a key requirement, driven by increasingly fierce broadband competition and declining profit margins.

A major part of the current upgrade of home network equipment is cloud-based centralized management, as well as machine learning and artificial intelligence (AI) capabilities to more effectively understand home network needs and consumption and optimize services within the home to ensure a consistent user experience.

While introducing machine learning and cloud-based CPE management, Wi-Fi 6 is rapidly being incorporated into next-generation home gateways and routers to increase physical layer throughput in the gigabit era. The main goal of Wi-Fi 6 is to ensure that the customer's Wi-Fi network does not hinder the delivery of high-bandwidth, latency-sensitive services (such as cloud gaming, 8k video, and cloud virtual reality services), with a focus on providing a theoretical maximum throughput of 10Gbps.

This combination of machine learning, AI, and Wi-Fi 6 gives service providers a new toolset to not only improve how they deliver Wi-Fi services to home broadband subscribers, but also to customize broadband and Wi-Fi services for each subscriber based on their consumption needs.

Using AI to reduce latency

One area where the combination of machine learning and Wi-Fi 6 could be most effective is in reducing latency on home Wi-Fi networks, enabling high-value services like cloud gaming and cloud VR services.

Wi-Fi 6 introduces OFDMA (Orthogonal Frequency Division Multiple Access), which allows routers and access points to divide multiple channels into smaller resource units (RUs). Each RU is then divided into smaller channels to transmit designated traffic for multiple devices at the same time. Taking this approach reduces the waiting time when devices connect and increases the overall throughput of the entire wireless network.

Leveraging the granularity of OFDMA, service providers can apply network slicing capabilities that are already available on enterprise Wi-Fi networks to home networks. In this case, VLANs can be assigned to specific packet flows associated with specific users, devices, or services. Using machine learning, VLANs and network slices can be dynamically changed based on upstream and downstream packet flows and the latency requirements of end devices and specific services. VLANs can also be prioritized based on packets and security. In this way, traffic for services such as cloud gaming that are very latency-sensitive can always receive higher priority.

Some operators are already leveraging this type of service to promote a dedicated gaming WAN. Operators use network slicing to not only ensure traffic priority throughout the home network, but also to minimize latency throughout the home network. This is a form of WAN acceleration specifically for gaming services, and operators can charge extra for it.

As bandwidth consumption ebbs and flows during specific hours, operators can dynamically reroute high-priority traffic, thereby continuously improving the performance of gaming services. This dynamic routing can be deployed in public networks as well as in home networks, where less congested Wi-Fi channels can be quickly identified and latency-sensitive traffic can be passed.

These capabilities go beyond DPI (deep packet inspection) capabilities. While DPI enables routers to quickly identify and mitigate security threats in a home or IoT environment, it generally does not have the ability to characterize packets as specific services or applications. AI capabilities integrated directly into home CPE can predict security threats based on learned traffic patterns. AI on the home gateway can identify devices on the home network, detect incoming threats, and identify the websites and server addresses they come from.

AI can also enhance parental controls beyond the basic function of managing Internet access based on user profiles or device MAC addresses. AI can be used to provide web content filtering and text analysis to identify emails, social media posts, and text messages (on a home Wi-Fi network) that contain explicit content or are potentially harmful or threatening.

All of these AI-based capabilities can help service providers change the business model of home broadband in a positive way. First, enhanced security and service identification capabilities help improve operational efficiency and reduce the frequency of service calls. Second, advanced security, service and application identification and prioritization, and content filtering capabilities enable service providers to clearly differentiate themselves from their competitors. In addition, as broadband price competition intensifies, service providers can use these unique services to quickly master pricing and provide an advanced service menu designed to maximize the customer experience.

Improving IoT security using slicing and AI

While speed, throughput, and minimal latency are key requirements for high-end gaming and video services, security and stability are becoming equally important in home networks as the number of sensors and home security devices increases. So, just as service providers can offer Wi-Fi network slices based on overall performance, they can offer a single Wi-Fi network for mission-critical home IoT devices, including home security and surveillance systems. This network slice will rely on machine learning and AI to identify usage and data consumption patterns for all devices and predict when sensors and devices will require software and firmware upgrades to ensure these issues don’t arise.

In addition to creating dedicated network slices for high-priority home IoT devices, home Wi-Fi networks must also rely on machine learning to understand when and how often external clouds and mobile applications access IoT devices. Machine learning algorithms in home CPE can monitor both incoming and outgoing device traffic to build an overall device and/or ecosystem profile, which can quickly flag abnormal traffic to and from IoT devices for users. Users can decide how to deal with potentially malicious traffic and infected devices.

Since IoT devices typically do not have the processing power and storage capacity to maintain a library of malware signatures, the responsibility for security increasingly falls on application providers or network operators. As broadband users become more reliant on IoT devices in their homes, users are more willing to pay extra to ensure the security and reliability of these devices.

Second, network operators can combine the power of machine learning with the additional features of Wi-Fi 6 to package IoT device management services as part of, or in addition to, managed Wi-Fi services. Specifically, Wi-Fi 6 includes a feature called Target Wake Time (TWT). TWT allows a home gateway to set a schedule for IoT devices to ping them to report their current status. As a result, devices don't have to compete for channel spectrum to communicate. Each device is guaranteed an optimal time slot to ping the router, and it can remain in power-saving sleep mode for longer periods of time.

FTTH becomes a key focus area for AI and Wi-Fi

FTTH (Fiber to the Home) network deployments continue to expand around the world. The reasons for this are not just to increase speeds and future-proof the network. Service and application flexibility is made possible by the underlying stability of the network. Operators can offer a wider range of applications and services over FTTH networks because they can dynamically allocate bandwidth to individual applications.

However, without an ONT that can process, identify, and predict packets by service or application type, the end-to-end network cannot gain this flexibility. In most of today’s FTTH networks, service providers still rely on basic bridging ONTs for physical fiber termination and then rely on separate routers to provide intelligent routing and management of packets and applications. But service providers are now realizing that integrating these capabilities and AI capabilities into smart ONTs allows them to allocate bandwidth by application, identify, anticipate, and correct problems in the home network itself, and provide an additional layer of security for home users.

The combination of AI and Wi-Fi 6 can help reduce latency for specific services by more than 50%, which is critical for latency-sensitive, high-priority applications such as online gaming, 4k/HDR and 8k video, and remote learning. Operators that have this equipment in their networks and offer these services can have a key competitive advantage by providing guaranteed QoE (Quality of Experience).

AI and Wi-Fi 6: The future of home networking

Amid the hype of 5G, operators around the world are working hard to enhance the capabilities of Wi-Fi and extend it into the home environment. Today, home broadband and Wi-Fi have become synonymous with users.

The combination of emerging Wi-Fi 6 gateway devices and technologies, along with cloud-based management and machine learning principles, makes the goal of rock-solid home Wi-Fi networks a reality. This can not only enhance service providers’ reputation among broadband users, but also create new revenue opportunities through comprehensive managed Wi-Fi service offerings as well as separate service tiers for specific user profiles.

At present, there are endless Wi-Fi 6 products on the market. In addition to the well-known iPhone 11 series phones that are compatible with the Wi-Fi 6 standard, there are many other Wi-Fi 6 products, including the Samsung Galaxy Note 10 phone that has just passed the certification, and the first Wi-Fi 6 10G e-sports router ROG GT-AX11000 launched by ASUS in 2018. According to incomplete statistics, there are more than ten products in the WiFi router category on the market, including ASUS, NETGEAR, TP-Link, etc., with prices ranging from hundreds to thousands of yuan. Compared with the price of the previous generation of routers, the price is higher. All major mainstream laptops are equipped with wireless network cards that support Wi-Fi 6, such as the Intel AX200 wireless network card.

However, there are also companies that integrate AI technology into Wi-Fi 6 and innovate home network scenarios. Huawei provides the industry's first eAI ONT series, which can intelligently identify the service type of home users through AI technology. With the help of innovative Wi-Fi 6 slicing and optimization technology, ONT can reduce the latency of specific services by more than 50%, thereby achieving zero frame freezing for high-priority services such as gaming, e-learning, and home/SOHO offices. By taking advantage of this, operators can launch value-added services while ensuring the experience to increase the average revenue per user (ARPU) of home broadband users.

Currently, Thailand's 3BB project uses Huawei's eAI ONT to help 3BB build a home broadband network that provides the best gaming experience.

In order to give users the best Wi-Fi 6 experience, on April 5, China Unicom and Xiaomi jointly released the first Wi-Fi 6 router - Xiaomi AIoT Router AX3600, which uses a Qualcomm 6-core processor, 6 high-performance external signal amplifier antennas, 512MB of large memory, and an AloT smart antenna with experience. It can easily realize one-click network configuration of Xiaomi smart devices and support WPA3 network encryption for safer wireless connections.

Comcast-backed startup Plume has developed an AI-driven adaptive home Wi-Fi mesh system. It has raised $85 million in a mix of equity and debt financing. Founded in 2014, Plume is one of many companies seeking to improve Wi-Fi connectivity by placing multiple routers throughout the home. Plume continuously learns and adapts to each home - monitoring internet usage and allocating bandwidth based on the devices that need it most.

The company's investor Charter Communications recently announced it would adopt Plume's open source OpenSync framework, and Plume claims that "more than 650 million devices communicate with 16 million OpenSync switches in 14 million homes."

After talking about so many new technologies of Wi-Fi 6, I am both happy and confused. I am happy that I don’t have to worry about network speed and security in the future. Now when I watch TV at home, the network will still "spin in circles". I am confused about these high bandwidth, low latency, automatic identification of threatening websites... When will these features be realized? Because the broadband of operators generally cannot reach such a high speed at present, the network equipment needs to be replaced, and the terminal also needs to be replaced. This is a long process! But I believe that technology has challenges in the process of development, as well as poetry and distance.