6G has a lot riding on its shoulders. On the one hand, 6G will seamlessly fuse the digital, physical and human worlds to create extrasensory experiences and make human beings endlessly more efficient. On the other hand, sustainability will be one of the guiding aspects of the 6G era. In 2030, our networks will not only refine the way we live and work, but they must also directly impact how we care for the planet.
One obstacle to building a sustainable 6G system is the fact that 6G will need to deliver much more data at faster rates than today’s networks, while still fulfilling very stringent energy-efficiency goals. This means that the required energy for transmitting a bit must be significantly reduced.
To achieve that end, we need radical ideas. Nokia Bell Labs and its partners at the Hexa-X consortium are investigating every manner of idea to create powerful yet efficient 6G networks for the future. One technology clearly fitting the bill is AI/ML. While AI/ML is not a new concept in mobile networking, it has so far only been implemented on a limited basis. With 6G, we can build the network from the ground up around AI/ML. This will result in an AI-native air interface, which will dramatically improve the efficiency of the network, while allowing many more devices to connect to it than is possible today. This means that industries across a broader range of domains will be able to operate more efficiently, thanks to seamless 6G connectivity.
Designing the air interface natively around AI/ML will allow us to use spectrum more efficiently, which will in turn make the network inherently more energy efficient as it takes less power to send the same amount of information. 6G can achieve these greater efficiencies by changing the way the network encodes data.
Current mobile devices encode information into what are known as symbols before transmitting through the network. But the network doesn’t merely send our videos, emails and other data in this symbol format. For each transmission, it also generates a special set of reference symbols that serve as a kind of interpreter for all the information that accompanies it. Those reference symbols allow the base station to compensate for any distortion in the wireless channel that would otherwise garble the transmission. Reference symbols, however, create a lot of overhead: in 5G systems typically 15% of any given transmission is taken up by these reference signals. In our 6G research, the plan is to eliminate those reference symbols completely.
6G will accomplish this through machine learning. By utilizing ML, the transmitter can “learn” the data symbols it needs to properly encode and decode any transmission, eliminating the need to send those references through the wireless channel. We’ve found that in single-antenna scenarios, this improves the spectral efficiency of a network by nearly 20% compared to 5G, resulting in a higher data rate for any given amount of transmit power. We expect that further research will show similar power savings are feasible in multiple antenna systems as well, making it ideal for the extreme massive MIMO used in future 6G networks.
Neural networks will tolerate hardware flaws rather than avoid them:
Another way in which an AI-native air interface can lead to more energy-efficient networks in the 6G era is by changing how we think about the inherent hardware limitations of mobile systems. Every manufactured component in a mobile network contains physical flaws. Such imperfections lead to a distorted signal, which hinders the accuracy with which the receiver can detect the transmitted data. This is similar to the clipping of audio signals we hear when sound is overamplified.
Today, we approach these hardware impairments by avoiding them. For instance, networks limit the peak transmit power in power amplifiers to prevent signal distortion caused by minute design flaws. But a properly trained neural network can tolerate such distortion without significant impact on the link throughput. This means that an AI-driven 6G air interface can be much more lenient towards such impairments, allowing amplifier modules to operate more efficiently. Considering that the amplifier consumes a large part of the total power required by a radio transmitter, this leads to considerable power savings.
AI can learn better 6G protocols:
Finally, AI/ML can also be used to learn more optimal methods for accessing shared radio spectrum. Traditionally, this is done by following a carefully designed and standardized set of protocols, called medium access protocols, that define when and how individual mobile devices transmit and receive signals over shared radio spectrum. In an AI-native system, the network can learn medium access protocols in the same manner it learns the data symbols we discussed above.
This means that rather than following a predefined and limited set of rules for accessing shared spectrum, the network can autonomously determine the best protocol to use in any given situation, improving spectral efficiency, which in turn creates more energy-efficient links. In addition, explicit efficiency constraints can be introduced into the training algorithm to ensure that these learned protocols satisfy stringent sustainability requirements.
A 6G network tailor-made for sustainability:
Our initial findings suggest that by designing 6G around AI/ML-based solutions we can achieve as much as a 50% reduction in transmit power over 5G for the same bandwidth and data rate. Considering this will be on top of all the power savings due to other technological advances, such as enhanced semiconductor manufacturing techniques, the significance of that number shouldn’t be taken lightly. Looking at it another way, 6G will use the same amount of power as today’s networks to transmit a much larger amount of data. There is no doubt that data traffic over mobile networks will increase dramatically in the 6G era. So in order to make 6G sustainable, one of our goals must be to reduce the CO2 footprint of every packet of data we transmit.
But AI/ML could do more than make networks greener. Promising 6G research shows that AI/ML may create tremendous flexibility in how we design individual networks. Different power grids have different electricity mixes, which is why the CO2 footprint of a RAN node varies between nations. It is therefore nearly impossible today to build a one-size-fits-all cellular network that aligns its power consumption with the CO2 targets of the country it is deployed in. The AI-native air interface has the potential to change all this since it could eventually be trained to respect sustainability key value indicators of the country in which it is deployed and the power-efficiency goals of the service provider that operates the network.
Altogether, the AI-native air interface will be a crucial component in building sustainable 6G technologies, as it can maximize the efficiency of the whole network. The AI-native air interface can help in achieving ambitious performance targets while still adhering to global ecological goals. Nokia is working diligently, together with the Hexa-X project partners, to create a future where 6G and sustainability are synonymous terms.
Website : sciencefather.com
See more Info : - network.sciencefather.com
Nomination : https://x-i.me/network
Registration : https://x-i.me/prireg2
Contact us : network@sciencefather.com
Social Medias :
Instagram : https://x-i.me/net23m
Pinterest : https://x-i.me/net23p
Facebook : https://x-i.me/net23f
Linked in : https://x-i.me/net23l
Youtube : https://x-i.me/NQGu
#ScienceFather #ResearchAwards #6G #NetworkVirtualization #NetworkProtocols
#Communication #Networking #Impromptu #Technology #NextGenNetworks
#WirelessRevolution #IoT6G #ConnectivityBeyond5G #6GResearch #Beyond5G
#DigitalTransformation #EmergingTech #Hyperconnectivity #6GInnovation #AI6G
#6GStandards #SmartCities6G #HighSpeedWireless #NetworkEvolution
#5Gvs6G #6GInfrastructure #6GApplications #FutureTech
No comments:
Post a Comment