5G and Artificial Intelligence
Salt and pepper. Day and night. Fred and Ginger. Some pairings create an exquisite experience that’s simply not otherwise imaginable. The same is true of artificial intelligence (AI) and 5G networks. As all types of 5G-capable devices become more widely available, 5G networks roll out globally delivering greater than 10 times the speed at a fraction of the latency of 4G. Meanwhile, AI use cases for consumers and enterprises also continue to mature quickly. These two trends are significantly interrelated, and together have huge implications for consumers, enterprises and communication service providers (CSPs) alike.
5G Networks will Accelerate AI Use Cases
The increased bandwidth and lower latency made possible by 5G technologies promises to enable a wide variety of new business models and new applications. Emerging technologies in augmented and virtual reality, cloud-assisted gaming, real-time language translation and image processing, remote surgery, connected drones and more, all require the delivery of consistent high-quality network connectivity. Additionally, enterprises and the public sector will increasingly take advantage of 5G and AI use cases to deliver to consumers enhanced entertainment, retail, financial services, mobility, public safety, security, healthcare, education and many other connected experiences. (See our earlier “AI Conversations” blogs on CIO.com for more on AI use cases.)
To achieve the ultra-low latency required for some of these AI-based tools, many of the applications that enable these new and emerging experiences and services will need to operate within the CSP’s network. This new operating environment is referred to as the mobile-edge-cloud, and is another essential ingredient to the success of 5G innovation. In addition, these business model innovations will result in intelligent devices and applications consuming and generating data like never before, creating a demand for data lakes and data analytics capabilities that are exponentially larger and more widely available to data consumers than what is available today. For consumers and enterprises alike, these realities of the growing AI/5G ecosystem create new challenges that need to be considered as new solutions and services are launched.
Impacts of 5G on CSPs
The 5G network represents a tremendous leap forward in a long cycle of accelerating demand and innovation for data mobility — with ever-increasing cravings for more and faster data delivery. For CSPs, the advent and rapid expansion of the 5G network signals both unprecedented opportunities and enormous challenges.
To explore the impacts of 5G on CSPs, I turned to my colleague Chris Falloon, a Senior Partner in Dell’s Global Transformation Office. Chris works with Dell’s largest service provider customers globally to help them transform their IT, workforce and network operations.
According to Chris, 5G networks open the door for CSPs to offer to a wide range of new service offerings, from enhanced mobile broadband and rich user experiences to lightning-fast data transfer speeds and connectivity for the massive numbers of intelligent devices flooding the Internet of Things. CSPs will help enterprises capitalize on unprecedented AI-centric opportunities by providing two key elements: the bandwidth and speed of 5G networks, along with new in-network application hosting environments in the mobile-edge-cloud and the massive data lakes they will require.
On the challenges side, 5G wireless networks — by their nature — require a much higher density of CSP network devices, and all the management complexity that comes with it. In moving from 4G to 5G, a CSP’s network with tens of thousands of nodes might morph into one with hundreds of thousands of nodes. The corresponding increases in preventative maintenance and the potential for network incidents can be a significant concern.
Chris points out that this increase in nodes and the need to manage new mobile-edge-cloud environments and provide customers with large-scale data management capabilities create a dramatic increase in both network size and complexity. These new realities demand a new end-to-end network and services architecture that is fully software-defined and that relies heavily on highly automated, AI- assisted operating models.
According to a Dell Technologies white paper, Dell Technologies and 5G: Analysis and Strategy to Capture the 5G Mobile Opportunity, the operational imperatives for CSPs center around three technology shifts:
- To disaggregate hardware and software stacks so that workloads can run on general-purpose compute, such as x86 servers
- To decouple core infrastructure and networking services from applications and protocols, so that these services can be delivered as a platform to applications
- To replace bespoke processes and infrastructure scripts with DevOps and “AIOps” style operational processes.
The authors say that “5G networks will be the first true end-to-end network built around these three paradigms: extending virtualization into the radio access network and network edge, virtualizing the network core, and extending end-to-end network overlays for network and service slicing.” 2
AI Use Cases for Successful Network Operations
Chris further explains that the use cases for AI span the landscape of 5G CSP operations. He provides a few key examples of the ways in which AI can help CSPs address the opportunities and challenges inherent in the adoption of 5G.
With the data-driven insights from AI systems, CSPs can monitor and analyze the customer experience, predict customer complaints, and proactively optimize services to keep customers happy and reduce churn. For example, AI-driven systems can predict heavy traffic and scale capacity to meet the expected demand. To maintain service levels, an automated system might make adjustments in the backhaul network or proactively add resources from other carriers.
AI-driven network operations tools help teams automate and streamline event and incident management, from problem reporting to event and incident responses and intelligent self-healing operations that replace expensive and time-consuming manual processes. This next level of operational automation is known as “AIOps,” in which AI empowers software tools to respond to operational events and incidents, act on changes in the network, address patching and security issues, and enable predictive maintenance — all without manual intervention.
Design and planning
AI can help CSPs improve the design of networks and enable better capacity prediction for near-term and long-term needs. It can give them the insights they need to see trends in traffic and the use of certain facilities, and then better understand how to make optimum use of network capacity to support targeted service levels.
To help CSPs see the decisions ahead more clearly, Chris offers these considerations for capitalizing on AI in their environments.
Consider network ops automation as an essential part of 5G.
Chris points out that, with today’s manual processes, it will be impossible to scale to the numbers of devices and incidents on the scale of those in a 5G network. The only feasible way forward is automation with AI-driven tools. “In 5G environments, networks need to manage themselves with zero-touch, self-healing capabilities powered by AI,” Chris says. “This is the North Star for next generation network and service management.”
Build a tool strategy.
As AI is deployed across an environment, CSPs could end up with hundreds of software tools. This can be overwhelming without a strategy for managing and governing AI on a large scale. “Operational simplicity is an essential part of keeping costs down,” Chris says. “Companies need to understand the end state of the tools ecosystem they’re building.”
Chris points customers trying get a handle on this emerging area to the TM Forum and its initiative called “AIOps Service Management.”2 This framework provides structure for the reengineering of multiple processes of the software lifecycle and service operations management to handle and govern AI software at scale.
Develop a data management and governance strategy.
A 5G network generates massive amounts of data. This data must be captured, stored and made accessible to various applications. To deal with the onslaught and gain the greatest value from captured data, CSP operations teams require a data management and governance strategy that makes heavy use of automation.
The CSP’s world is changing in dramatic ways. With the rise of 5G networks and the associated complexity of managing millions of new devices, CSPs’ operating models must change. In this new world, virtually nothing can be managed with legacy manual processes. Instead, CSPs’ networks will increasingly be managed in a zero-touch manner, via automated AI-driven systems. The leaders in this emerging ecosystem will be both the enterprises and CSPs that embrace AI and intelligent systems that seek to manage themselves.
To learn more
For more on 5G IT transformation and key use cases, see Dell Technologies and 5G: Analysis and Strategy to Capture the 5G Mobile Opportunity.
Take a deeper dive into network optimization with Intel’s AI and machine learning: Why now? Network optimization in the age of 5G.
1 Dell Technologies. “Dell Technologies and 5G: Analysis and Strategy to Capture the 5G Mobile Opportunity.” November 2019.
2 TM Forum. “AI for IT & Network Operations (AIOps). Retrieved 9/16/2020.