Intel sees the combination of 5G and AI as “transforming dreams into reality”
Intel has been an ecosystem and market leader as operators turn to new virtualized and open 5G RAN architectures. With silicon support, partnerships and reference designs for vRAN and Open RAN, Intel is now looking at how fast-developing artificial intelligence (AI) technologies can be brought to bear to drive innovation in the 5G RAN. Speaking at RCR Wireless News’ Open RAN Global Forum, available on demand here, Intel’s Cristina Rodriguez, vice president of the Network and Edge Group and general manager of the Wireless Access Network Division, went through six ways AI can deliver 5G RAN innovation.
Setting the stage, Rodriguez said, “We need a network architecture and infrastructure that allows to implement new functions quickly, and we need all the bright minds and developers to drive innovation, and make all these dreams a reality. Open RAN and virtual RAN lower the barrier of entry, and bring a richer ecosystem. That’s why the RAN needs to be virtualized all the way to layer one. Open, virtual RAN delivers extraordinarily flexibility and web scale benefits through secure containerized microservices align operators to deploy compostable applications. It positions operators to save costs through more efficient resource utilization, and get more from the infrastructure by harnessing the latest artificial intelligence techniques. These advanced intelligent networks will help operators elevate user experience, and rapidly deliver new services, as well as automate service deployment, and maximize energy efficiency even as networks become more complex and disaggregated. With a full cloud-native approach, you get the maximum flexibility, scalability, software agility, and the greatest possibilities for innovation. Only with this approach, we can harness the latest and best of artificial intelligence techniques.
Specific 5G RAN “pillars of innovation” covered by Rodriguez include:
—Dynamic network configuration: “AI enables dynamic hardware allocation for RAN functions through software-driven resource allocation making fine-grain, real-time resource provisioning and control, nearly effortless. Dynamic resource allocation can deliver very real operational and capital cost saving as network rightsize the resources in real-time.
—Traffic steering: “The network traffic can be balanced across multiple cells, and correct coverage and capacity problems on the fly.”
—Optimized spectrum allocation: “Spectrum efficiency will be automatically optimized to improve network throughput and reliability.”
—Energy efficiency: “We have seen significant reduction in power using AI in conjunction with dynamic power management techniques.”
—Predictive analytics and controls: “AI is not only about reacting to changing network conditions, but also about predicting them. Predictive analytics leverage insights, algorithms to study KPIs, and compare against learned baselines. When anomalies are detected, control algorithms proactively and preemptively take action. Networks will become self-correcting and resilient.”
—Network slicing: “This is another example of AI driving business innovation. AI applications unlock the opportunity for network slicing where operators can develop new revenue streams based on customer specific SLAs…with AI backed guarantees on throughput, latency, and reliability.
For 5G RAN combined with AI, “zero-touch automation is the ultimate use case”
Whether it’s related to virtualization, cloud-native, AI, or a combination of those three (and other technology sets), the big goal for operators is automation. The goal here is to automate what can be automate thereby reducing opex, then tune those automation capabilities to more effectively monetize the network through responsive service delivery that makes optimal use of compute, network and spectral resources.
Rodriguez traced a path, using AI, from self-learning to self-correcting to “eventually full self-sufficiency. Zero touch automation is the ultimate use case. And to make all this happen, we need the right technology..To handle the demanding needs of RAN workloads including Layer 1, Intel has built an architecture based on flexible, programmable, general purpose CPUs integrated with acceleration for the most demanding tasks that enables operators to deploy a fully virtualized 5G RAN without compromising, and being able to reap the full benefits of a software defined network, end-to-end, all the way from core, to edge, to RAN.”
She ended her presentation by taking a long view on AI that starts with 5G RAN architectural and other investments today, and sets the stage for ongoing advancements. “AI-driven automation will be a game changer and a critical technology for future generations of RAN standards. For this to happen, we need a cloud-native, software-defined, fully programmable network infrastructure, and we need the right technology underpinning the architecture. We need a platform that is ready for innovation today, right now, and that has the flexibility and the scalability to grow into the future. Virtual RAN and its cloud-native architecture[s]… are helping operators to create a software programmable platform where operators can begin integrating AI capabilities into their networks today while providing the flexibility for continued innovating in the future.”
