Skip to main content


Why 5G Networks Need AI at the Edge

5G promises increased capacity, lower latency, and whole new classes of applications and services. But capitalizing on these improvements will require unprecedented levels of automation attainable only through artificial intelligence (AI).

For network operators, AI complements the automation of their network functions virtualization (NFV) and software-defined network (SDN) infrastructure. At the same time, enterprise and industrial users can leverage this distributed intelligence to reduce costs, enhance security, and deliver new low-latency, data-driven services.

5G can be deployed without AI, but leveraging AI in 5G mobile edge compute will be essential to streamlining operations as the number of network endpoints, use cases, and overall complexity grows. The first step in its implementation is defining “the edge.”

Defining the 5G Network Edge For AI Compute

The “edge” of a 5G network can mean many things, depending on the end user/operator or application. This is especially true given how dramatically the edge can shift in SDN and NFV deployment scenarios. At a high level, for instance, a network operator, enterprise organization, or device manufacturer/end-user each could define the “edge” differently:

  • Operator Network Edge—The operator network edge, often referred to as multi-access edge compute (MEC), can be positioned in radio towers, base stations, edge routers, or central offices that comprise a distributed data center. These platforms offer the compute power to host AI stacks for both the operator and its clients/partners.
  • Enterprise Premises Edge—The enterprise edge is capable of supporting AI compute at any point within a facility or system. Data from smart devices is collected and processed at the enterprise edge, though some amount of cloud connectivity is required for updating machine learning (ML) models.
  • Device Edge—The device edge refers to smart devices developed by OEMs or being operated by users that send and receive data from AI systems such as routers or gateways. These gateways have more computational power or access to aggregate analytics from many smart devices.

Once defined, operators and enterprises can use 5G AI edge compute for targeted outcomes.

AI at the Edge for Network Operators: Dynamic Network Slicing

Operators gain the most from AI in MEC platforms, where the technology can make traffic optimization and network resource management more efficient. Dynamic network slicing, for instance, includes real-time selection of optimal data rate or choosing the best access point or network through which traffic can be routed.

Using AI in dynamic network slicing will prove critical to 5G operators as it enables them to offer different qualities of service (QoSs) for common users, enterprise customers, and vertical industries. Besides automating the process of efficiently provisioning the network based on the amount and type of traffic, this also allows operators to better monetize their infrastructure by offering multiple service tiers at different prices.

AI at the Edge for Enterprise: New Service Opportunities

Users in the enterprise and vertical markets also can capitalize on AI-enabled 5G edge compute for service offerings of their own.

AI at the edge places intelligence closer to data sources, which enhances security, minimizes latency, reduces backhaul costs, and puts AI closer to industry domain experts than in cloud-centric topologies. For domain experts working on ML and deep learning (DL) models, this proximity assists in engineering solutions that send only the highest-quality edge data back to the cloud for model training.

Intelligent manufacturing use cases provide a good example of how AI and 5G edge compute add value for enterprises and end users. With AI at the edge, not only can network nodes ingest a range of sensor, video, and other input data and output real-time decisions, the millisecond latency of 5G technology allows those responses to be integrated in control applications.

As shown in Figure 1, enterprise and industrial organizations can leverage this position at the edge to provide security, analytics, and other AI-as-a-Service offerings.


Figure 1. Deploying AI across the 5G edge network infrastructure can provide solutions to a range of needs in the retail, industrial, smart city, and other markets. (Source: Intel® Corporation)

5G AI Mobile Edge Compute: Open-Source, Standards-Based, and Off-The-Shelf

Of course, technical challenges exist before AI at the 5G network edge can become a practical reality. How to federate and update AI stacks at diverse edge points at massive scale is high on this list. One avenue to ensuring the success of AI in 5G edge compute is the use of hardware standards and open-source software.

From a hardware perspective, companies like Joinus Technology are delivering the ETSI and PICMG ATCA Solution Based on Intel® Xeon® Gold Platform. Such solutions provide up to 22 cores and 44 threads per socket, while also integrating Intel® Advanced Vector Extensions 512 (Intel® AVX-512) to accelerate AI workloads in enterprise premises and operator network edge environments. The platform can be used at the operator network edge to automate traffic orchestration and resource management, or on-premises in the enterprise for AI-as-a-Service.

On the software side, Intel and its partners are open-sourcing solutions that will allow ecosystem players to develop AI stacks quickly and cost-effectively. For example, Wind River Titanium Cloud was built from OpenStack, and provides a platform for supporting AI in 5G MEC.

The Intel® Network Edge Virtualization (Intel® NEV) SDK is packaged with Titanium Cloud, providing network architects with a suite of reference libraries and APIs for different edge networking deployment scenarios. Both are supported by platforms like the Joinus Technology ATCA Solution based on the Intel Xeon Gold Platform.

AI Automation for 5G

In addition to traditional communications, 5G networks will connect billions of Internet of Things (IoT) sensors. With these come new machine-type communications (MTC) and networks, as well as broader use of the RF spectrum.

The ambitious goals of 5G require higher levels of automation, which can be increased only by distributing intelligence throughout the network infrastructure. AI at the edge can continually optimize network performance for operators by monitoring usage trends over time, and also act as a platform for new services and business opportunities for carriers, the enterprise, and industry.

With 5G network trials already underway, the time to bake AI into the mobile edge infrastructure is now.

About the Author

Brandon is a long-time contributor to going back to its days as Embedded Innovator, with more than a decade of high-tech journalism and media experience in previous roles as Editor-in-Chief of electronics engineering publication Embedded Computing Design, co-host of the Embedded Insiders podcast, and co-chair of live and virtual events such as Industrial IoT University at Sensors Expo and the IoT Device Security Conference. Brandon currently serves as marketing officer for electronic hardware standards organization, PICMG, where he helps evangelize the use of open standards-based technology. Brandon’s coverage focuses on artificial intelligence and machine learning, the Internet of Things, cybersecurity, embedded processors, edge computing, prototyping kits, and safety-critical systems, but extends to any topic of interest to the electronic design community. Drop him a line at, DM him on Twitter @techielew, or connect with him on LinkedIn.

Profile Photo of Brandon Lewis