Driven by increased numbers, types, and operational behaviours from customers, the telecommunication industry have spent over a decade transitioning from hardware-based to software-based infrastructure. In doing so, these networks have become large end-to-end programmable distributed systems, ready to meet the challenge through automation.
Much of this automation is based on the use of ML applied to use cases for the purpose of control, prediction, or classification. Despite impressive achievements, the design, development, and maintenance of such solutions is based on human-effort. With the global 2021 telco labour spend at $273 billion (vs $326 billion on infrastructure), using humans in this way per use case, per context across the network is not scalable.
This talk proposes a different vision of the future for telecommunications. In this vision, the network itself is responsible for the creation, validation, and deployment of the its own software. Drawing on examples over the last 5 years from research, standards, open source, and industrial activity, this talk will outline the challenges, opportunities, and practical steps underway and required towards scalable autonomous operation of the network.