logo_header
  • Topics
  • Research & Analysis
  • Features & Opinion
  • Webinars & Podcasts
  • Videos
  • Event videos

Defining automation

Automation can be described as operational shift but is “a fuzzy word” that’s difficult to define. We detail some level-setting definitions.

20 Apr 2020
Defining automation

Defining automation

This is an excerpt from our recently published report Network automation using machine learning and AI. Download the report now for the full insight. The telecom industry is always evolving. Sometimes this means moving from one generation of technology to the next, while other times the change is more foundational, like going from analog to digital or from circuit­switching to internet protocol. The shift to 5G is both, a generational shift in technology and potentially a foundational shift in the way communications service providers (CSPs) do business. Automation can be described as operational shift: It doesn’t so much change the fundamental processes of designing, creating, fulfilling, monitoring and managing networks and services as alter how they are implemented in profound and beneficial ways.

As one expert interviewed for this report explains, automation is “a fuzzy word” that’s difficult to define. Is closed-loop automation the same as autonomous networking? Is an automated process governed by policy the same as rules-based automation or simply running scripts? Is AI required for autonomous networks?

Below are some level-setting definitions: Network automation – a recent MIT Technology Review report, produced in partnership with Ericsson, defines network automation as “the elimination of repeatable manual tasks and their replacement by programmed tasks automated with the use of software.” Examples include the configuration of servers, scheduling maintenance, and adding or removing services. Network automation that goes beyond a single server or service to the configuration of several virtualized network functions often requires some orchestration to manage the workflows across the network. A good example of this is Vodafone’s successful test of full automation of its transport connectivity services. Closed loop – Blue Planet, a division of Ciena, offers a good definition of closed loop: “a continuous and repeating cycle of communications between the network infrastructure and software elements, including analytics, policy, and orchestration, to enable selfoptimizing capabilities.” Self-­optimizing capabilities – self-optimization takes closed loop a bit further, leveraging closed-loop processes to automatically adjust parameters and configurations to make optimal use of constrained resources such as computing resources, radios, transport and access facilities, and energy. Autonomous network – fully autonomous networks don’t exist yet, but TM Forum members define them as “providing the service lifecycle on demand with minimal or no human intervention.” Members are collaborating in the Autonomous Networks Project to develop a common understanding of and consensus about what defines autonomous networks and how to implement them. The idea is for autonomous networks to configure, monitor, maintain and repair themselves independently, providing a fully automated, zero-wait, zero-touch, zero-trouble set of network and ICT services for businesses in many industry verticals and consumers. As noted in the introduction to this report, most CSPs believe that autonomous networks will eventually require artificial intelligence (AI – see below). AI – the development of computer systems capable of performing tasks that normally require human intelligence. Machine learning and deep learning are types of AI: With machine learning, computers, systems and machines learn and improve from experience without being explicitly programmed; deep learning takes machine learning further by processing information in layers, where the result or output from one layer becomes input for the next. CSPs collect huge amounts of network and customer data, the volume and complexity of which is increasing rapidly because of a ballooning number of devices and experience-related data. AI, machine learning and deep learning are becoming necessary to analyze and use this data. AI in Operations (AIOps) describes the use of AI technology to automate CSPs’ operations. Read more about the state of AI and autonomous network in the industry today.