Trend
the costs of building and training ML and AI models, the risks of training using poor or biased data, and looming regulatory uncertainty around AI. As a result, automation efforts in 2024 will focus on lab consolidation, cloud migration, automating specific network lifecycle and DevOps processes, and where automation will become increasingly important as service providers embrace more complex multi/ hybrid cloud environments. Improving energy efficiency is also
on the agenda this year, as intelligent automation provides early wins for service providers seeking to lower energy costs and reduce CO2
emissions. As
these efforts evolve, especially paired with more advanced AI, operators will have the opportunity to achieve huge sustainability gains across their entire network and operational footprints in 2025 and beyond.
4
AI will drive data centre evolution AI is already transforming data centres as we see massive investments by
cloud service providers to introduce AI fabric in order to keep up with demand. In 2024 there will be intensive focus and investment in graphics processing units and other back-end infrastructure for large learning clusters. Cloud service providers will also be looking to upgrade front-end inferencing and faster interconnection technologies. Ethernet will dominate these front-end networks, with 800G adoption growing rapidly among Tier-1 cloud service providers next year, when 51.2Tbps switches reach the market. Even in the back end, during 2024 and
2025 we will see new network designs for high-speed Ethernet as viable alternatives to InfiniBand. Aimed at high- performance computing, proprietary InfiniBand technology delivers the needed throughput and latency, but Ethernet offers a secure, ubiquitous and cost-effective option. Initially we’ll see cloud service providers start using Ethernet in parallel with InfiniBand;
indeed, more cloud service providers will view Ethernet as essential for the long-term evolution of AI infrastructure within the data centre, simply due to the huge volume of processing that AI needs.
5
Growing use of network digital twins Across the telecoms industry, service providers are advancing toward end-to-end,
AI-powered, self-driving networks, with 2024 being the year they start adopting network digital twins. Digital twins allow the operator to
identify and overcome operational challenges, especially where time- consuming, error-prone manual processes dominate, such as maintenance engineering. Traditionally, maintenance upgrades and changes to a live network must be performed during specific windows, which disrupts operations. But, through network digital twins, highly-accurate, real-time virtual models will give service providers the flexibility to thoroughly validate changes before introducing them. Paired with automation, they can experiment, evaluate and learn more quickly how to best optimise new service deployments, all the while reducing errors and outages in the live network.
6
Private networks will continue to gain ground Te market for enterprise private networks saw slow but steady growth throughout
2023. In 2024, we expect them to see significant growth and become a meaningful source of revenue, buoyed by demand from enterprises, governments and military customers. Last year commercial private networks
typically achieved a return on investment within six months. For certain 5G-enabled applications, some customers achieved reduced downtime and improved productivity: • Ultra-high-definition video monitoring: Customers are finding private network “AI vision” solutions extremely valuable in areas like factory supply-line monitoring
and fault detection. Ultra-HD video monitoring can detect damage to manufactured components in real time, which otherwise would be difficult or impossible to notice. Tese solutions can prevent costly production errors or damage to manufacturing equipment, avoiding hugely-expensive repairs and downtime. • Augmented reality: Remote maintenance is very expensive, yet with communication networks, pipelines, power grids and other infrastructure with a large, distributed footprint, faults or breakages typically require expert engineers to be dispatched to remote locations. As a result, repairs oſten translate to multiple days of expensive downtime and lost productivity. With private networks that support AR, onsite personnel can overlay schematics on equipment and perform guided repairs in real time. AR will allow less skilled local employees to fix more problems, quickly and cheaply. • Government and military: Military agencies have also been exploring 5G private networks and AR tools for maintenance and field training activities. With private networks and AR, military and government agencies can have enhanced capabilities and insights anywhere, even to the most remote, inhospitable locations. A recent survey by STL Partners revealed
great willingness by enterprises to pay a premium for private networks – provided that operators can guarantee business outcomes under SLAs. As demand continues to grow, we’ll see
service providers building up their SLA management and monetisation capabilities for private networks. And in enterprise private networks
we’ll see increased investment in edge computing, with tighter security and enhanced data privacy. Te most compelling private network
applications, especially those that are latency-sensitive or AI-/ML-driven, will benefit from local application processing. To meet this growing demand, look for service providers that can bundle private network connectivity with edge computing.
www.electronicsworld.co.uk March 2024 05
Page 1 |
Page 2 |
Page 3 |
Page 4 |
Page 5 |
Page 6 |
Page 7 |
Page 8 |
Page 9 |
Page 10 |
Page 11 |
Page 12 |
Page 13 |
Page 14 |
Page 15 |
Page 16 |
Page 17 |
Page 18 |
Page 19 |
Page 20 |
Page 21 |
Page 22 |
Page 23 |
Page 24 |
Page 25 |
Page 26 |
Page 27 |
Page 28 |
Page 29 |
Page 30 |
Page 31 |
Page 32 |
Page 33 |
Page 34 |
Page 35 |
Page 36 |
Page 37 |
Page 38 |
Page 39 |
Page 40 |
Page 41 |
Page 42 |
Page 43 |
Page 44 |
Page 45 |
Page 46 |
Page 47 |
Page 48
