That's it, Mobile World Congress (MWC) 2023 is over. It was an incredible, long week rich in meetings, discussions, and above all — announcements.

A key topic of this year’s MWC was open and virtualized radio access network (Open vRAN).

On the show floor, we demonstrated our infrastructure innovation that culminated in Qualcomm 5G RAN Platform portfolio and Qualcomm Edgewise Suite. A comprehensive infrastructure portfolio to support the network of the future. One of our key infrastructure demonstrations was our industry-leading, energy-efficient virtualized and Open RAN demo. 

But before jumping into the details of the demo, let’s take a step back to understand its significance in leading the industry towards 5G virtualized networks. 

With the desire to move away from being tied to legacy equipment from a single supplier, and the need to quickly scale networks capacity when and where is needed, the use of commercial off-the-shelf (COTS) hardware and virtualized software functions is gaining momentum. 

However, there are doubts on Open vRAN’s capability to run latency-sensitive and high-complex layer-1 RAN functions on COTS servers in an efficient and economical fashion. Executing full baseband processing functions on COTS central processing units (CPUs) is not cost or energy-efficient, especially for high-capacity large-scale deployment. 

And here comes the role of Accelerator Cards — offloading CPU cores from latency sensitive and compute-intensive 5G baseband L1 (PHY) Layer functions which can increase performance, reduce the number of processors required, and significantly reduce power consumption.

In-line vs look-aside architecture

There are two primary architectures for hardware acceleration technology: look-aside and in-line architectures. 

• Look-aside accelerators: Whether they are in the form of separate cards or the latest version, which is integrated in the main processors offload CPUs only from select L1 functions, mostly Forward Error Correction (FEC). This implies that majority of L1 functions are still implemented in the CPU. 
• Inline accelerators: Unlike in the Look-aside architecture, in-line accelerators offload all the L1 processing from COTS CPUs, providing a much more energy efficient L1 implementation and freeing up valuable CPU resources for upper layers, such as L2 and L3. This approach eliminates the need for higher core CPU processors resulting in direct CAPEX and OPEX savings. These savings increase as we go to high-capacity scenarios such as Massive multiple-input, multiple-output (MIMO), which requires significant L1 processing and therefore, significant CPU cores. The technical motivation of the increased energy and cost efficiency of in-line accelerators relies on using DSP computing architecture, which is known to be the most efficient architecture for low-latency signal-processing workloads. 

In addition, in-line accelerators deliver significant system scalability benefits. As more L1 capacity is needed, networks operators can add more accelerator cards independent of COTS CPUs. Lookaside, by contrast, means reliance on adding more COTS CPUs as capacity increases.

While early deployments of vRAN were done based on current available look-aside acceleration solutions, in-line accelerator cards are rapidly gaining momentum and have been adopted by global operators.

However, when it comes to performance, advanced features, and power-efficiency, not all in-line accelerators are created equal. And this brings us back to our demo at MWC.

The most energy efficient Open RAN solution ever built

With the understanding that the best open RAN solution is the one that offers the best performance, most advanced features, and highest capacity at the lowest power consumption, we made the decision to demonstrate our industry-leading, energy-efficient Open RAN solutions at MWC 2023.

The demo showed 64T64R Multi-user MIMO 100MHz with 16 layers Radio Unit — powered by Qualcomm QRU100 5G RAN Platform — connected to a Hewlett Packard Enterprise Proliant 110 server with a Qualcomm X100 accelerator card.  The system simultaneously serves four mobile devices each with four layers in the downlink — showcasing cutting-edge performance with total energy consumption of the entire X100 card less than 18 watts (W). 

The X100 card is an in-line accelerator card that runs entire 5G baseband L1 layer and offloads CPUs from the latency sensitive and compute-intensive functions, such as demodulation, beamforming, channel coding, and Massive MIMO computation. It is designed to support a throughput of 24Gbps at less than 40W peak power consumption. This truly unmatched power consumption makes this solution the most energy efficient ever built.

With operators’ commitments to net zero and desire to save cost, energy efficiency have become a strategic priority for them. Based on a study by GSMA Intelligence, more than 70% of energy is consumed in the RAN — in other words, Qualcomm X100 improvements in energy efficiency will have a significant positive impact on operator’s operational and financial targets¹.

Collaboration with ecosystem players

We are bringing innovative, cost-effective, and power-efficient Open RAN solutions to the marketplace, boosting the infrastructure ecosystem. However, it’s only through industry collaborations that we will unlock the full potential of 5G Open RAN. We have played a crucial role in the adoption of 5G Open RAN through collaborations with world’s key ecosystem players. We’ve announced many exciting milestones to advance open RAN commercialization across the globe. This includes:

Dell Technologies: Collaborating on the development next-generation 5G vDUs for Open RAN 5G deployments.

Fujitsu: This collaboration will commercialize 5G open and virtualized integrated distributed unit (DU) and RU solutions with powerful 5G mmWave performance capabilities.

Hewlett-Packard Enterprise (HPE): Together with HPE, we will deliver the industry’s first fully-optimized (vDU) solution with up to 60% lower total cost of ownership (TCO).

Mavenir & Vodafone: Announcing Next-Generation, artificial intelligence-powered Massive MIMO Radio Unit powered by Qualcomm QRU100 5G RAN Platform.

NEC, Hewlett Packard Enterprise (HPE), & NTT Docomo: Introducing latest NEC 5G virtual distributed unit (vDU) powered by the Qualcomm X100 5G RAN Accelerator Card.

Rakuten Mobile, Inc. (Rakuten Symphony’s parent company): We’re collaborating to develop a next-generation 5G RU with massive MIMO capabilities and DUs.

Viettel Group: Unveiling prototypes of Viettel’s distributed unit (DU) and radio unit (RU), powered by Qualcomm 5G RAN Platforms.

Vodafone: Collaborating to develop, test, and integrate next-generation 5G DUs and RUs with massive MIMO capabilities, ultimately to deliver the commercial deployment of Open RAN in Europe.

Zain: Collaboration to drive cloud-native, virtualized and O-RAN compliant 5G infrastructure in Saudi Arabia.

Winning the debate

Whoever delivers highest performance, advanced features, and most power-efficient solutions will win the race. Our industry-leading 5G RAN platforms is setting new benchmarks and laying the foundation for the next generation of high-performance, Massive MIMO-capable, and power-efficient 5G networks with low total cost of ownership (TCO). We’re looking forward to commercializing this solution with all our partners this year.

1. GSMA. (2021). Going Green: benchmarking the energy efficiency of mobile. GSMA Intelligence. Retrieved from https://data.gsmaintelligence.com/api-web/v2/research-file-download?id=6062… on March 14, 2023.