The increased use of AI is changing the nature of the way traffic flows through mobile telecommunication networks. These changes force mobile operators to upgrade their networks to better serve the most current needs.
The transmission of tokens – the data that AI models process – is prompting an explosion in the volume of data coursing through mobile networks. And token traffic does not flow the way that traditional communication traffic has historically flowed, which poses problems for today’s mobile networks, which were not designed to serve AI needs.
Go deeper with GlobalData
Access deeper industry intelligence
Experience unmatched clarity with a single platform that combines unique data, AI, and human expertise.
Transmitting tokens creates two significant challenges for mobile operators. One is the fact that common protocols for transporting data over the internet (TCP/IP) fragment data into packets rather than grouping them into files. Another is that network resources tend to be allocated equally to a range of traffic types, which results in data from a particular source being transported in separate batches, potentially affecting the reliability and continuity of token transmission.
The rise of AI is also increasing the need for enhanced uplink connections in mobile networks, for example, because many AI use cases require frequent real-time multimodal data transmission to the cloud for inferencing, processing, and agent interaction, and some of this transmitted data is bandwidth-intensive, such as video. And it may increase the need for network duration, to the extent that AI agents consume network resources for more hours of the day than human beings do.
One way for mobile operators to meet the new burdens that AI is placing on their networks is to deploy and activate 5G-Advanced technologies.
5G-Advanced is a set of standardised network software features and technologies that represent the next half-step in evolution for the 5G network technologies that arrived just before the start of this decade. Just as 5G-Advanced technologies were set to debut commercially, the topic was largely overshadowed in industry discourse by the rapid rise of AI and the topic of using AI to enhance networks. That AI would steal 5G-Advanced’s thunder is somewhat ironic, since 5G-Advanced could help operators address the requirements that AI is placing on the network.
5G-Advanced’s multi-dimensional capabilities can allow operators to increase uplink and downlink speeds, reduce latency and improve coverage, among other things. It can also help provide differentiated services – for example, giving some users a temporary speed boost or a certain service assurance.
5G-Advanced is enhancing the performance of mobile networks with larger antenna arrays than the first wave of 5G networks, higher power output levels and simultaneous support for multiple frequency bands. New generations of 5G-Advanced gear also offer lightweight radio designs for easier installation and minimised site fees, as well as greater energy efficiency (with fewer radios needed to support multiple spectrum bands).
New innovations in commercial base station equipment also help evolve the network to better address its latest needs. New power amplifiers and advanced shielding techniques make next-generation radios more efficient. New software algorithms help reduce signal interference to make uplink connections stronger. And network configurations made in parallel rather than in series offer even more efficiencies that translate into improved network performance. These innovations are appearing in commercially available products, such as Huawei’s Super Uplink MetaAAU and EasyAAU radio/antenna units.
Another way that operators may enhance their networks to better serve AI applications is with more spectrum. Many of the world’s operators deployed 5G in what is known as “mid-band” spectrum between 1GHz and 6GHz; this represents an effective combination of coverage and capacity. But the ability to harness additional spectrum can increase network capacity. Regulators in some countries have allocated parts of the upper 6GHz frequencies to mobile operators, while some have left this spectrum unlicensed, to be used by WiFi traffic, for example. China has been proactive in allocating upper 6GHz spectrum, which offers high bandwidth for increased capacity. And some networking equipment providers – for example, Huawei — have developed products to harness this spectrum. However, it will take time for an ecosystem of consumer devices supporting these frequencies to become commercially prevalent. In the meantime, operators need to scale the steep learning curve inherent in understanding both how AI is changing their networks and how AI can be used to improve them. The industry is perhaps less than five years away from introducing the next generation of mobile networking technologies: 6G. Although 6G is currently at an early stage of development and largely undefined, a key principle held by industry stakeholders is an interest in making 6G an AI-native technology from day one. That means operators must use the time between now and then to invest in AI technology and understand how it will change mobile networks by the time 6G arrives.
