AI burden in RAN - 5G Wins!

-


Does 5G wear the AI burden more lightly in the RAN than earlier ‘older’ generations?

Most certainly! We aver in our upcoming report, “AI ML and DL in the Mobile Core”.

To be sure, there was no shortage of motivation for the usage of AI in 4G, or for that matter ANY generation of RANs. Let us look at 4G networks alone.

4G networks offered a quantum leap in terms of throughput over its predecessors. That aspect alone would have incentivized telcos to view their RANs from the prism of AI-enabled insights. To be sure, they did.

Then there was energy management. Practices and the possibility of using AI and ML technologies in energy management precede the advent of 5G RAN. The case for traffic optimization in the pre-5G era was advanced greatly by the hybrid mode of deployment - the 5GC coupled with old radios as the hybrid mode offered a spectacular rise in throughput, enough to keep network planners interested in AI and ML-based constructs.The principle driver for AI however is not throughput alone, it is complexity. In terms of network complexity, 4G networks took only baby steps in form of CUPS. What was the end-result? The comparatively rigid and bounded nature of 4G networks as well as the limited number of use-cases led the operators to merely scratching the surface when it comes to harnessing AI and ML constructs in coding.

A full-blown 5G network is a different entity altogether. At its absolute pinnacle, which we are still approaching, 5G promises to be a smoothly functioning ecosystem of multiple networks – each with its optimal coding pattern, paving way for advanced AI and ML-based management technologies.

5G is not merely a quantum jump in data throughput, it is also a spectacular expansion of the bouquet of services that telcos can offer. This aspect makes energy management considerably more pressing and acute as a concern, necessitating exploration of AI and ML technologies.

And then there is network slicing.

Are we there with real network slicing yet? Not really, skeptics may emphatically say.

Are we going to be there in the near future? There can be no doubt about it though.

Network slicing will be almost singularly responsible for 5G network planners to turn towards intensive application of AI and ML technologies in traffic optimization. Just look at how closely the RIC partner ecosystem of leading lights resembles a real marketplace. It’s not just the diversity of options, it is also the depth to which some of them go.Such is the seminal facilitating impact of the RIC that we have a very real possibility of disaggregation and programmability of the physical layer from the DU.

So yes, 5G rules!

What do you think are the additional reasons for and against the impact of 5G RAN on AI adoption? Let us know in the comments below.

Comments

Post a Comment

Popular posts from this blog

SDN, NFV and Indian Telcos

-
  I returned with useful insights from the SDN-NFV India earlier this month. There were two discernible trends – optimistic solution vendors and cautious telcos. Indian telcos are worried about the following: Indian telcos, including the largest ones, are genuinely concerned about the solidity and stability of SDN and NFV products. Telcos think that vendors do not completely appreciate the difference between a service outage suffered by Facebook and an outage suffered by a licensed cellular operator. The former can be irritating, the latter can be catastrophic. Indian telcos are also looking towards concrete ways of making money from SDN and NFV. What works with telcos in the US may not work with Indian telcos. The above points were recurring themes. You could say that Indian telcos were never among the early adopters of technologies. You could also say that Indian telcos are bleeding far too heavily in the present to think rationally about overhauling their networks. Plenty has be...
Read more

India - Don't forget 5G!

-
I recently conducted a poll on the prospects of 5G roll-out in India. I basically tried to gauge the perception about the availability of 5G in India - on or before 2021; or after that. Expectedly, the overwhelming response was the latter. India has too many challenges to surmount before 5G can make an appearance - the auction process, the operator finances and the choice of vendors. Each of these challenges is a topic for separate discussion. The uncertainty surrounding 5G in India is unfortunate. 5G does promise higher data rates than its predecessor, but that is not its only advantage. 5G institutionalizes network segmentation or 'slicing' according to the traffic profile. It is thus possible for low throughput IoT sensor traffic to co-exist with data-intensive video without the network designer breaking into a sweat. For India, which has ambitions of developing world-class manufacturing, logistics, transportation, healthcare, finance and governance frameworks; it is importa...
Read more

AI, RAN and Delay-Doppler - How Cohere does it

-
Image
Agreed. The title has too many nouns! Each of the nouns is important though, as we will find out. Allow me to explain. So this article is a profile of Cohere Technologies . Cohere features in our report “AI and RAN – How fast will they run”. As you would now guess, Cohere Technologies specializes in using the Delay-Doppler channel model. It uses this model in software to MU-MIMO enable existing FDD and TDD networks. What does the model do? This model leverages AI/ML for detection, estimation, prediction and precoding of channels. What does the term ‘Delay-Doppler' mean? It is easy to misinterpret this phrase as a special use-case of the Doppler effect – it is not. Cohere uses the word ‘delay’ as a shorthand for distance and the word ‘Doppler’ for velocity. Cohere postulates that tapping the channel simultaneously for time delay and Doppler effect provides a unique profile of the signal path from the antenna to the user – specifically, the scattering of signal and its causes can b...
Read more