AT&T and Ericsson Pioneer AI-Powered Cloud RAN with Intel Xeon 6 SoC

AT&T and Ericsson have achieved a significant breakthrough in the evolution of cloud-native telecom networks, successfully demonstrating Ericsson’s AI-native Link Adaptation on a Cloud RAN stack powered by Intel’s Xeon 6 SoC. This milestone underscores the potential of artificial intelligence (AI) to dramatically improve the speed, flexibility and performance of radio access networks (RAN).

The Power of AI-Native Link Adaptation

The demonstration, conducted over AT&T’s licensed frequency bands, marks the first instance of portable Ericsson RAN software operating on Intel’s new Xeon 6 system-on-chip (SoC) platform. AI-native Link Adaptation dynamically adjusts to real-time changes in signal quality and interference, optimizing data transmission rates and boosting overall RAN performance [1]. This capability is designed to work effectively on both purpose-built and cloud-based infrastructure.

Intel Xeon 6 SoC: A Foundation for AI in RAN

Intel Xeon 6 SoC is engineered specifically for network and edge deployments, delivering breakthrough AI RAN performance with built-in acceleration. The SoC integrates Intel Advanced Vector Extensions (AVX) and Intel Advanced Matrix Extension (AMX) technologies, eliminating the need for separate, discrete accelerators while maximizing capacity, efficiency, and total cost of ownership (TCO) optimization [2].

Key Benefits of the Collaboration

• Open Architecture & Flexibility: By decoupling RAN software from vendor-specific hardware, AT&T gains the ability to integrate innovations more rapidly and avoid vendor lock-in [3]. This portability allows for faster scaling of new capabilities from the lab to live network deployments.
• First-Mover Advantage: AT&T is the first in the industry to demonstrate the portability of a commercial-ready AI feature on a Xeon 6 SoC-based cloud platform.
• Performance Gains: Tests have shown up to a 20% increase in throughput compared to legacy, rule-based link adaptation, alongside improvements in spectral efficiency [4].
• Scalable AI Deployment: Ericsson and Intel are jointly benchmarking AI models on commercial off-the-shelf hardware, delivering measurable improvements in AI inference speed and a scalable path for AT&T to deploy AI where it drives the most value.

What is Cloud RAN (vRAN)?

vRAN, or virtualized RAN, represents a shift in how RAN baseband processing functions are handled. Instead of relying on dedicated hardware at cell sites, vRAN utilizes general-purpose processors (primarily from Intel) to run baseband processing as software. The “functional split” within vRAN determines how processing is divided between centralized processors and the radio unit, impacting fronthaul bandwidth, latency, and cost.

Looking Ahead

This successful demonstration represents a crucial step towards more intelligent, efficient, and adaptable mobile networks. The collaboration between AT&T, Ericsson, and Intel paves the way for wider deployment of AI-powered RAN solutions, promising enhanced network performance and a better user experience. Continued benchmarking and optimization of AI models will be key to unlocking the full potential of this technology.