• Nvidia is working to ease adoption of new AI RAN applications with tools for training, simulation, and deployment that leverage the power of AI to speed up and scale these processes.
• New offerings include automatic code translators, digital twin simulators, and an “AI-native 6G lab in a box.”
Source: Nvidia
Artificial intelligence (AI) chip giant Nvidia is continuing its drive toward ushering in a new age of AI-enhanced radio access networks (RANs) – tackling some of the hurdles facing this young ecosystem and working to make it easier for software application developers to innovate in the space.
• RAN vendors are exploring how AI can improve RAN operations, including through the use of both RAN-specific AI operations and multi-agent coordination to assess faults and recommend improvements.
• AI RAN is also being used to support new services, as equipment vendor ZTE illustrated with an AI-enhanced small-cell solution in a car dealership that supports livestreaming in the building.
RAN vendors are pushing further toward achieving a RAN that makes greater use of AI – to support new services and to improve network operations, including by coordinating with AI in other parts of the network.
Some vendors have begun proposing the use of multi-agent AI platforms to improve network performance. For example, training separate AI agents to analyze performance degradation in different areas of the network – RAN, core, and transport – may allow those agents to develop specialized areas of expertise to maximize their accuracy. These specialized agents could each work directly with a coordinating agent tasked with overseeing and orchestrating their activities and findings.
ZTE is one of the equipment vendors proposing such a model. And it is exploring other ways to improve RAN operations and support new services using AI. This year, ZTE began marketing its AIR Engine solution to make better use of AI in the RAN. The solution includes a general-purpose server card inserted into an otherwise traditional baseband unit – the processing unit of a mobile base station. (ZTE had previously introduced a similar offering aimed more at private networks; AIR Engine primarily targets public mobile-broadband networks.) AIR Engine is designed to use AI to enhance RAN performance by, for example, improving power efficiency and spectral efficiency as well as beamforming – closely tracking users’ movements, identifying the applications they’re using and applying the best network resources for those specific applications. The solution doesn’t currently use graphics processing units (GPUs), which have been widely deployed for AI tasks but have thus far been deemed by many telecom-industry members to be too expensive for distributed deployments in mobile base stations.
ZTE has also discovered a real-world use case applying AI RAN to small cells. The vendor has deployed its QCell distributed indoor small-cell solution at a car dealership that regularly hosts auto-industry influencers who want to stream videos from inside its showrooms. In addition to providing 5G connectivity, this version of QCell uses AI – running on a general-purpose compute plug-in card – to support live streaming in three ways: It optimizes the network for streaming (including supporting uplink speeds), it minimizes the impact of the streaming on other network users and it uses self-learning to reduce power consumption.
Mobile operators taking such a solution to market are likely to face some familiar challenges, including (a) understanding specific enterprise verticals well enough to know how AI-enhanced 5G could add value and (b) confronting the clash between traditional telco business models – which benefit from highly scalable offerings like consumer mobile broadband – and enterprise solutions like the one described above, which may require high customization without offering high scalability.
Mobile operators should also explore whether enterprise AI RAN solutions work better as virtual RAN solutions. ZTE’s use of a plug-in card for AI compute could impose some limits on capacity. Could a solution that uses general-purpose compute across AI and RAN workloads alike be more efficient? ZTE, which does not embrace vRAN in its RAN strategy generally, argues that using vRAN in the case of the car dealership could drive up the cost of the solution for a use case that is quite cost-sensitive.
Whatever challenges these new models present, vendors are increasingly taking them on and engaging the learning process to move further toward visions of a more AI-centric RAN, which is expected to be a significant defining characteristic of the next generation of mobile networks in a few years, when the 6G era arrives.
• Operators want 6G migration to be more software-centric and require less expensive radio installation than 5G.
• Stakeholders developing industry standards for 6G are making progress toward such goals.
As the mobile telecom industry continues to work on defining 6G – the next generation of mobile technologies, expected to meet commercial reality before the end of this decade – operators have been vocal about the ways they want 6G to be unlike 5G. In particular, they want to avoid the disappointing return on their 5G network rollout investments, including the expensive nationwide installation of large massive MIMO radio/antenna units.
• Echostar, owner of Boost Mobile, will sell its wireless spectrum to AT&T and decommission its Open virtual RAN – which Open RAN proponents had long pointed to as a proof-point of Open RAN’s value.
• The news not only impacts Boost Mobile’s RAN suppliers but could also become a symbol of Open RAN’s long struggle to accelerate adoption and a bellwether for the sector’s near-term prospects.
Echostar’s move this week to sell its wireless spectrum licenses to AT&T and transition into a “hybrid mobile network operator” is a grim milestone for the Open RAN ecosystem, since the company’s Boost Mobile operator unit says it will wind down parts of its open virtual RAN (vRAN) as a result.
• IRIS² aims to enter service by early 2031, providing internet access and secure communications for different applications, similar to SpaceX’s Starshield and Starlink.
• The EUR10.6 billion project funded by the EU, ESA, and consortium companies, aims to increase Europe’s strategic autonomy and unlock a new area for economic growth.
The European Commission has signed a concession contract for IRIS², the European satellite constellation announced two years ago, which will oversee the development, deployment, and operation of a network of secure, multi-orbital satellites. Continue reading “IRIS²: Europe Launches a Satellite Communication Program”→
• Comcast, Charter, and Broadcom are collaborating to develop a unified DOCSIS 5.0 standard.
• Despite fiber’s advantages, existing cable infrastructure can be upgraded to meet future demands, delaying full-scale fiber adoption – at least in the US.
Just when some analysts felt comfortable proclaiming DOCSIS 4.0 as cable’s final chapter, the mirage of DOCSIS 5.0 now seems to be taking shape. The development comes after a phase marked by ambiguous comments on what lies beyond DOCSIS 4.0. These references often hinted at DOCSIS 5.0 (a.k.a., 25G DOCSIS, or 3 GHz DOCSIS) whenever the future of cable was in question, particularly in comparison to PON.
• NVIDIA introduced a new virtual RAN platform based on its GPUs
• The company argues that a converged RAN/AI network can make the RAN more efficient and enable new services to further monetize mobile networks.
Earlier this year, when AI giant NVIDIA and others announced the formation of a new industry consortium, the AI RAN Alliance, the lack of detail surrounding their plans was understandable. After all, they had promised not much more than to begin exploring the ways that AI could be used to transform RANs.
Summary Bullets:
• Ericsson is expanding its virtual RAN (vRAN) partnership with server maker Dell to include more joint solution development, customer engagement, and support.
• The deal shows each vendor taking a more committed approach to pursuing the vRAN market, interest in which has surged in the past two years.
• At Mobile World Congress (MWC), Ericsson introduced a new E-band radio with increased power output to extend the length of backhaul connections, and Intracom added a beam-tracking antenna to maintain link alignment.
• Wireless backhaul specialist Ceragon promoted a long-awaited product it hopes to unveil later this year, the first product based on the vendor’s new in-house silicon: an E-band radio promising a capacity of 100 Gbps.
• Nokia will use CPUs from Nvidia to bring more choices to the vRAN space, following the former’s ‘anyRAN’ concept.
• Nokia also aims to use Nvidia’s GPUs to transform telecom networks using AI – an aspirational future vision.
Nokia announced this week that it will collaborate with chipmaker Nvidia in two ways – both focused on virtual radio access networks (vRAN). Nokia will use Nvidia Grace CPU Superchip for processing vRAN functions at Layer 2 and above, together with Nokia’s RAN software and hardware accelerators (PCIe cards that boost vRAN performance). It will also use the graphics processing units (GPUs) Nvidia is most known for to handle AI applications and vRAN acceleration.
