[PROFILE] One step B.Yond: Startup is using AI and automation to map networks

 

This editorial by Mike Robuck first appeared in FierceTelecom, Oct 4, 2018


  B.Yond is using AI and machine learning to map out service providers' network elements. (Pixabay)

B.Yond is using AI and machine learning to map out service providers' network elements. (Pixabay)

B.Yond is blending artificial intelligence and automation to provide carriers with real-time snapshots of their network assets.

Having real-time visualization of network assets can help carriers detect faults before they happen, keep track of VNFs as they are moved around a network, and provide the "smarts" to keep up with the dynamic changes in cloud environments.

"B.Yond is applying artificial intelligence or AI to network automation," said Chief Product Officer Rikard Kjellberg, in an interview with FierceTelecom. "We're doing that by building a software platform that is completely cloud native. It runs on multiple different types of clouds, anything from AWS to an on-premise OpenStack deployment, whether it's Red Hat or Mirantis or anything else. It is, like I said, designed for cloud, and it runs in a 'five nines' type environment.

"When you go this cloud route, the dynamics change fundamentally in your network, and the traditional legacy models for operationalizing this are simply obsolete. The reactive model and rules-based model won't be able to handle the new dynamics. You have to have more smarts, and that's where the machine learning and AI come in.

B.Yond's first product, Integrity, is an automated inventory reconciliation tool that helps ensure inventory systems provide an accurate, live view of underlying hardware and virtual network assets. Kjellberg said B.Yond's feature set includes an integrated lifecycle management pipeline and continuous integration and continuous delivery pipeline

"The first use case that we're going after is really foundational to everything we do from there," Kjellberg said. "That first use case is network topology visualization. By going into the network and collecting all the information, all the data from all the network elements, from RAN into the data center, we're able to create a real picture of what the telco has in its network.

"I think on average we figured out that about 6% of their network assets are actually invisible to the telco today. That leads to lost revenue. It leads to capital investments sitting unused and being written off without generating any revenue."

To date, Integrity has been deployed by a Tier 1 carrier in the U.S., but Kjellberg said other telcos are taking a hard look at it. B.Yond is working on its second use case, which has yet to be deployed but it's in trials and proof of concepts with 5 of the 10 largest telcos.

"Now with that very accurate visibility into a network, we can then apply AI to look at the dynamics of the network, and, for example, help our carriers to make better investment decisions and proactively make more targeted capital investments down to the cell sites," Kjellberg said. "That's the second use case that we're currently working on. We're just going to keep adding use cases over time. The platform is very open in that regard."

One of the hot topics this year has been the cloudification of networks, which B.Yond has evangelized since its founding last year.

"As we look into virtualization and cloudification, we're looking at a much more rapid evolution, in terms of your network's going to have a lot more vendors and a lot more contributions from companies big and small," said B.Yond's Johnny Ghibril, VP of data science and solution architecture, during the FierceTelecom interview. "Cloudification is what gives you that benefit of a software managed, software operated network. That means you have network elements and services and products that are being spun up and spun down at any given time, much of it automated, without giving it direction.

"Making sure that you're continuously tracking that live snapshot of the network at any given time is very foundational. Having that accurate snapshot is absolutely essential to be able to actually trace, troubleshoot, build, and manage the network."

B.Yond’s origin story

B.Yond was founded a year ago by CEO Ned Taleb, who was a co-founder of Nexius. Nexius provides software-based telecommunications network deployment services and was one of the first members of Facebook's Telecom Infra Project (TIP) that made its debut two years ago at Mobile World Congress.

"Ned saw an opportunity with 5G and edge computing to really disrupt the status quo of the telco/vendor relationships that are in existence today, and create a new type of a product company and a new type of relationship with the telcos," Kjellberg said.

RELATED: Editor's Corner—Open source is not 'one size fits all'

In addition to other AI and automation startups, B.Yond also faces competition from larger vendors. B.Yond's competitors in the elements management systems space include Cisco, Nokia and Ericsson.

In April, B.Yond announced was one of the members of the Linux Foundation's LF Deep Learning Foundation, which was designed to facilitate artificial intelligence, machine learning and deep learning. Kjellberg said B.Yond has adopted Kubeflow, which was put into open source by Google.

Kjellberg has been vocal about the status of open source communities and standards bodies. He said a company of B.Yond's size needed to carefully choose which open source group to put its efforts  and resources behind.