From the course: AI in RAN (Radio Access Network): Transforming Mobile Networks

The evolution path of RAN technologies

From the course: AI in RAN (Radio Access Network): Transforming Mobile Networks

Start my 1-month free trial Buy for my team

The evolution path of RAN technologies

“

Imagine trying to upgrade a bicycle into a self-driving car overnight. It's impossible. That's exactly why telecom operators are evolving their radio access networks into different strategic stages, not just in one giant leap. Let's break down what each phase looks like. We understand the legacy hardware and ecosystem existing in current scenario where operators are stuck in vendor locking ecosystem. They are relying on proprietary hardware that is rigid and expensive. For example, like buying a locked smartphone that only works with one carrier. That's how the current legacy ecosystem looks like. There are different pain points related to performance which is being capped by physical hardware where operators sometimes over-provision which is wasting the capacity or sometimes under-provision the different functions and instances which are not the ideal scenario or if there is a new features to be add-on in the network it takes months and years depending upon when the vendor release the new update. Second stage where the operators can move their radio access network towards is the virtualization which is the first taste of freedom where software runs on virtual machines but still it could be vendor controlled here moving from dvd players for example to streaming but only on netflix that is this scenario where no competitors allowed so scaling resources via software would be helpful in this case but the underlying architecture may still remain closed with the vendor. The third part is the cloudification, where actually the magic starts. Here, the key performances are decoupled from the hardware. Here, the automation enters into the picture. It could lead to better resource utilization, and hence can lead to the better savings, maybe up to 30% to 40% in this stage, as compared to the legacy environment. We then have the AI driven self-optimizing networks with the cloud native ecosystem. Here are the containers, Kubernetes, they are replacing the virtual machines and have a faster deployment with lighter workloads. The artificial intelligence can lead to certain applications such as energy savings during low demand, or maybe for traffic optimization, which can enhance the overall efficiency of this ecosystem. Now we talked about these stages, but why don't operators skip these three stages and go straight away to stage four? Well, there are certain limitations, especially for the operators with the brownfield networks such as 2G and 4G networks already in place with the proprietary hardware. There the legacy systems won't talk to cloud native technology without gradual upgrades. Also the skill gaps where the engineers need time to master the cloud native AI machine learning environment. Also the risk management which is the biggest factor for the telecom operators because a big band change in the radio access network straight away from stage one to stage four could crash critical infrastructure which could lead to outages and hence could lead to penalty and could damage the brand. brand, so hence the operators are very cautious about adopting the evolution of RAN in one go and hence they are going in different stages.

Contents