Operators are applying artificial intelligence and machine learning technologies to leverage the power of their new programmable, software-based networks.

Sandra O'Boyle, Senior Analyst – CEM & Customer Analytics, Heavy Reading

November 10, 2017

7 Min Read
AI & Machine Learning in NFV/SDN: Key Takeaways

At Light Reading's recent OSS in the Era of NFV/SDN event in London, I moderated a panel discussion on "Analytics, Machine Learning & AI in Next-Gen OSS/BSS" and wanted to share some key insights from the speakers:

  • Dima Alkin, VP Service Assurance, Teoco Corp.

  • Oliver Cantor, Business Network and Security Solutions, Verizon Communications Inc. (NYSE: VZ)

  • Ignacio Mas, Senior Expert in Programmable Network Architecture, Ericsson AB (Nasdaq: ERIC)

  • Mark Pendred, Control and Orchestration Lead, BT Media & Broadcast

  • Jay Perrett, Founder and CTO, Aria Networks Ltd.

Machine learning and AI can help, but set the right expectations
Machines can solve problems, learn what's happening and act on those learnings -- but they have to be programmed how to learn and they have to be taught rules and outcomes (e.g., how to route traffic through a network). However, to utilize machine learning (ML), companies first have to overcome barriers such as organizational distrust and a general resistance to ML and AI. Dima Alkin from TEOCO explained that in the case of service assurance, we are very forgiving of human error in root cause analysis and trouble-ticketing systems, but we expect machines to be absolutely right every time, even if it's unrealistic. The onus is still on us to give machines the requirements and get objectives right -- what are you trying to do?

Oliver Cantor from Verizon was cautious, and said the hardest part is applying machine learning to the multi-dimensional network layer; operators are not bad at the customer side of things. He advocated getting the data in one place and starting with machine learning to understand patterns and what the data is telling us.

Ignacio Mas from Ericsson, a network engineer at heart, discussed the issue of maturity of using machine learning and AI models in the network, stating that the models are already there and being used by operators to crunch data to predict customer behavior and prevent churn. With networks becoming programmable and software-based with SDN/NFV, operators should try these models out.

Dima Alkin also discussed why big data analytics projects have failed in the past. One issue was the amount of time and resources spent on reprocessing, normalizing and standardizing data to make it perfect and to make analytics work. He advocates consuming available data as is and moving as close as possible to live network environments to try out models, rather than wasting time in proof-of-concept trials.

AI – top down, bottom up or both?
There was debate and disagreement on this topic -- whether AI algorithms in areas such as policy and closed loop control needed to be top down -- decided by an intent or information model -- or bottom up -- through optimizing control loops at the infrastructure layers.

Jay Perrett, founder and CTO of Aria Networks, was adamant that networks should be a commodity -- an intelligent Ethernet cable that decides how to get from A to B -- and that network optimization should be handled top down based on what the service needs and what's right for the business.

Ignacio Mas from Ericsson argued we need both -- a top-down and bottom-up approach -- to create a network that will support services and make changes we want, and react to customer needs/what we need from the network.

Perrett also discussed how AI is important as it can scale, operate automatically (e.g., self-healing VNFs) and adapt to new requirements/changes in the model (e.g., to support new use cases). With an information model and plug-and-play algorithm system (e.g., topology optimization, route optimization), you can probe and change so that when a new use case comes on, it can adapt to the environment. This is important, as you cannot anticipate use cases that will be important in three years, for example.

Next page: The business value of AI

The business value of AI
Network operations are ripe for automation. In an earlier presentation, Brian Naughton, CTO of Accanto Systems , showed how Level 2 and 3 support roles would dramatically change with NFV and automated provisioning, monitoring and healing processes. A key consideration is what to do with the outcomes and information from AI and machine learning: They have to be integrated into processes in order to impact the business.

Perrett from Aria Networks also brought up the issue of knowing how much of your network is for revenue vs. how much is for resiliency. This is a business issue because you can engineer the network to the nth degree without taking into account revenue. Another caveat: because we talk about software and NFV/SDN as scalable networks where VMs can be added on demand, the panel said we need to remember that there are finite resources, and the AI engine needs to know the limits and boundaries -- before you get the license bill at the end of the month.

Orchestration function and AI
The discussion moved on to using AI to inform business logic and policy in network and service orchestration decisions. Marc Pendred from BT said that the orchestration function needs to have speed of reaction to customer needs, or BT will lose customers. When it comes to an orchestrator, you need to have something that is topology-aware and a number of integrated business rules to improve network operation and optimize the traffic flows.

Ignacio Mas from Ericsson was skeptical about a single engine taking decisions and expects that control loops will be occurring throughout the network at different levels. Using the example of cell stations and frequency-hopping, a policy of changing frequency could happen as control loops at the cell level, but an exception would be that if required to go to a different frequency band, approval would have to go up to a "higher level" because that could be a business decision. There is also the assumption that the system at any level will be able to teach itself, and that testing will be possible to predict the impact before going live (e.g., adding a new node or service to the network).

Oliver Cantor from Verizon also discussed proactive loops or open loops as a first stage before we get to control loops and automation, and jokingly referred to open loops as those requiring higher-order human involvement. For example: Do you really want to do this? Followed by: Are you really sure you want to do this? Also, the AI engine must be not just reactive but proactive, so as to prevent network conditions/issues by taking actions in advance.

Where will AI and machine learning be most disruptive? Some closing thoughts from the panel
"Security operations -- when reality forces you to do something, this is when you get to this point. We see advertising and customer micro-segmentation early successes, and only now CSPs are bringing it to network engineering and operations, and a lot of technology that enables it. We see opportunities to solve problems in the legacy network environment, we have operators asking what you are doing for NFV/SDN/5G -- can you do same on fiber network?" -- Dima Alkin, VP Service Assurance, TEOCO

"It will be interesting to see how this works in practice. I can see clashes, say, between a security policy engine that wants to shut things down due to a perceived threat versus a customer policy decision to increase bandwidth. Two people in this case make policy; it could be quite disruptive if you automate these things." -- Oliver Cantor, Business Network and Security Solutions, Verizon

"Business-driven optimization is the driver, and disruption is network-based on SDN/NFV." -- Jay Perrett, founder and CTO, Aria Networks

"In the operations base, changing from manual processes to trusting a machine -- what's also unclear to me is what happens to automation tomorrow when user habits change over time?" -- Mark Pendred, Control and Orchestration lead, BT & Broadcast Media

"We expect to see more disruption at the bottom layer 1 and physical layer, e.g., frequency-hopping, lambda traffic optimization in routers, we also expect to see more disruption. In the areas of customer interaction and CEM, churn prediction, behavioral prediction, concrete aspects of customer behavior -- in the last few years, a lot has happened." -- Ignacio Mas, senior expert in Programmable Network Architecture, Ericsson

— Sandra O'Boyle, Senior Analyst, Heavy Reading

Read more about:

Omdia

About the Author(s)

Sandra O'Boyle

Senior Analyst – CEM & Customer Analytics, Heavy Reading

Sandra leads Heavy Reading's research on customer experience management and customer analytics related to the network and services, customer care, billing and marketing. Sandra also looks more broadly at how service providers are reinventing digital operations with a "customer first" focus and adopting big data strategies. Sandra brings to these areas an excellent understanding of the competitive issues and market trends shaping the telecom and IT sectors. Sandra joined Heavy Reading from Rohde & Schwarz's ipoque, a network traffic and subscriber analytics vendor, where she worked in strategic product marketing. Prior to that, Sandra spent more than ten years as Research Director for the global business network and IT services practice at Current Analysis covering enterprise cloud and network services, and advising operators, IT service providers, vendors, and enterprises. She has also held editorial research positions at PC World and The Industry Standard in San Francisco. Sandra is based in Amsterdam.

Subscribe and receive the latest news from the industry.
Join 62,000+ members. Yes it's completely free.

You May Also Like