It’s become almost a mantra in the industry: 5G is the promised land. However, due to burgeoning Opex and Capex cost-effective 5G deployment will make automation a necessity without which the promised land cannot be reached. Automation is absolutely crucial to address the deployment and management challenges arising from the large number of network elements that will be deployed to ferry the huge amounts of data that will inundate networks. While 5G is still being planned and in initial testing and deployment phases it hasn’t been rolled out in most countries on a large scale, it seems that MNOs have not rushed to implement automation – even though every company involved in 5G understands the challenges.
One reason for this dissonance is the perceived “lack of maturity” in automated RAN technology that has been holding MNOs back from implementing 5G automation, according to a report by Rethink Research. One important factor holding companies back from fully embracing 5G technology, according to the study, is that “progress has been slower” in automating the network. That “immaturity in the technology” has “led to fears that operators would end up automating bad practice, or suffering trade-offs in the quality of their networks.”
Despite the need – and the attendant profits for companies that get to 5G first – operators are right to hold off until automation is up to par. The sheer number of network elements needed to run a 5G network – small cells, picocells, mini-macro base stations, complex antenna arrays, etc. will require an automated eye to deploy efficiently, keep things running. Repairing and maintaining all those systems is going to be a challenge, to say the least.
The efficient deployment is the first challenge that needs to be addressed. Automation will enable MNOs to more easily deploy resources as needed, optimizing configurations and adaptive network policies as needed in order to optimize customer experience. When 100,000 fans are streaming live HD video from the Super Bowl, communication service providers (CSPs) need to ensure that network capacity in the area of the stadium expands and is robust enough to ensure a perfectly smooth experience for everyone.
It all sounds great, but the question of how to get from where we are now has plagued the industry ever since large-scale deployment of 5G has been contemplated. Keeping up with all the equipment and addressing all the deployment scenarios using the age-old manual processes is going to be impossible, I believe – unless automation powered by artificial intelligence is part of the solution.
AI is largely based on machine learning – where algorithms compare thousands of data points in order to determine maximum efficiency or operating parameters – fits hand in glove with RAN automation.
In an automated deployment scenario all or most of the heavy pre-planning manual work needs to be eliminated. The AI system “gets to know” all the components in a network and how they work together – and is able to project how the entire network and all its components will perform under normal and stressful circumstances. Using that data, the system can ensure automatic deployment of new elements as needed in a CI/CD mode. Pre-activation is done seamlessly based on the adaptive AI powered configurations, while post activation optimization is done on the fly 24/7 with no manual intervention.
For ongoing optimization and maintenance machine learning AI-based systems can sweep equipment feeds of all types – all the different components of a 5G network “in the wild” – and examine their performance, determining if it matches the parameters that CSPs require and expect as well as the customer experience of the end users. The AI system learns how a network is supposed to operate in the multifaceted scenarios it is subjected to, and builds an adaptive model for each piece of equipment, each deployment algorithm. If something is not performing as it should, or appears to be headed in that direction (i.e., a piece of equipment cannot handle X amount of connections when Y conditions prevail), the AI system will be able to determine that in advance, and activate the systems that can correct/repair the problem.
The AI system “gets to know” all the components in a network and how they work together – and is able to project how the entire network and all its components will perform under normal and stressful circumstances. Using that data, the system can ensure automatic deployment of services as needed in a CI/CD mode, in advance of the need even appearing. Pre-activation is done seamlessly based on the adaptive AI powered configurations, while post activation optimization is done on the fly 24/7 with no manual intervention.
AI and machine learning is the answer to the conundrum of how CSPs can finally move forward with 5G – remaining confident that their systems will work, and that customers will remain satisfied.
