Dawn of the AI era for Mobile Networks
Artificial Intelligence & Machine Learning (AI/ML) will impact every industry is a foregone conclusion and the Telecommunications industry is no exception. Widespread 4G/LTE implementations and the evolution to 5G networks create tremendous opportunities for AI/ML to play a role in design, operations and service lifecycle management of mobile networks. In fact, AI/ML will be a key competitive factor for Mobile Network Operators (MNOs)in the 5G era.
MNOs compete with each other horizontally to win new subscribers and retain current ones (a.k.a churn management) with the price being the key competitive factor. MNOs also try and compete vertically with Over-the-Top (OTT) players to play a role in the value chain beyond connectivity. 4G/LTE networks came with expectations that would allow MNOs to compete vertically with OTT. Not only have these expectations been unfulfilled, but MNOs have also undergone consolidation in some very large markets including the United States and India in the last decade.
5G networks offer yet another opportunity for MNOs to level the playing field and compete vertically for additional revenue streams from their subscribers. MNOs also have the opportunity to monetize data from their networks and participate in the value chain beyond connectivity on an equitable basis with OTT players in Autonomous Driving, Content Delivery, Retail, and other verticals. 5G networks promise low latency connectivity, edge cloud capabilities, cloud-native network function (CNFs) with hyperscale capabilities and more.
5G networks also come with complex radio access network (RAN) challenges in the orchestration of macro and micro cells, radio frequency interference management to name a few. Beamforming is an interesting AI/ML use case that helps with radio frequency interference management. This is a good tutorial on beamforming if you want to know more.
The orchestration, management and dynamic resource allocations to these VNFs (a.k.a. VNF Autoscaling) per changing subscriber demand is another interesting AI/ML learning use case that will result in optimal Telco Cloud CapEx investments, differentiated SLA management with profit maximization intent and network operation efficiencies. DaVinci Networks is building a deep neural network based solution to address this use case.
Deploying any AI/ML solution on a network with multi-tiered legacy platforms and disjoint management systems with vendor locked-in data is a challenge that MNOs are facing as they try to cross over to distributed, disaggregated and open architectures. Upcoming blogs will address open source software innovations that are gaining traction to support AI/ML microservices, AI/ML skill shortage in the Telecommunications industry and state of the art for machine learning for mobile networks.