ETSI: AI Will Minimize Network Over-Engineering

Network operators are looking at artificial intelligence first and foremost to help minimize the resources used in their networks, according to ETSI's Ray Forbes. However, use cases like automation are still a few years out.

Huawei Technologies Co. Ltd 's Forbes is the Convener of European Telecommunications Standards Institute (ETSI) 's Industry Specification Group (ISG) on Experiential Network Intelligence (ENI), which launched with a kick-off meeting in April. The group, made up primarily of vendors but also Verizon Communications Inc. (NYSE: VZ) and China Telecom Corp. Ltd. (NYSE: CHA), was set up to define a Context Aware System using AI based on the "observe-orient-decide-act" control model. More simply, it's crafting how to apply AI to software-based network management. (See OrbTV: ETSI Goes Sci-Fi With AI or the video below for more.)

Networks are becoming hard for humans to understand at a glance, Forbes says, and a lot of cost is deployed by the network operators who are over-engineering their networks and running network services at their least efficient times. He estimates that networks are about 50% over-engineered today, and AI can bring them down to only about 10%. That's why priority number one for operators is to apply AI to use their equipment more efficiently. (See The Automation Taboo: Let's Talk About Jobs.)

"It's making resource management less human dependent in terms of knowledge of how they can do that," Forbes adds. "You want to have a business analyst who just performs data models; you don’t want to have a lot of pre-thinking and management analysts that analyze what to do."

These are the goals he's hearing from ENI members Verizon and China Telecom. Forbes says ETSI is also in the process of talking to others like NTT DoCoMo Inc. (NYSE: DCM), Orange (NYSE: FTE) and Deutsche Telekom AG (NYSE: DT) about joining, but hasn't gotten firm commitments yet.

For more on automation, visit the dedicated automation content channel here on Light Reading.

It's still the early days for most telcos to even be doing AI at all, so using AI to automate their networks is even further out on their roadmaps. For ETSI, Europe's standard organization, its goal is to develop use cases and requirements for AI this year, model them next year and then tackle testing and automation a year or two after that, Forbes says.

"We don’t want to jump in the deep end," Forbes adds. "Taking inspiration from things like the AlphaGo model -- it spent awhile learning. It lost games quite well at first, but it took 18 months to master it. You need to design a model and spend a bit of time teaching the model how to do things."

Forbes called moving to "fully AI-based autonomics" -- whether everywhere in the network or in a few places -- a "pipe dream," but one that ETSI's ENI working group is taking a step towards. He adds, "The ultimate aim is to improve the user experience and simplicity to have intent-based network indications in line with customer demand for new revenue."

— Sarah Thomas, Circle me on Google+ Follow me on TwitterVisit my LinkedIn profile, Director, Women in Comms

Sign In