CableLabs marches toward virtualization with TIP, SNAPS

CableLabs is moving forward with several projects that were designed to create frameworks or use cases for virtualization.

While telcos, such as AT&T and Orange Group, have moved publically and openly down the virtualization racetrack, the cable industry has been a little slower out of the starting blocks due to variations across MSOs' networks.

CableLabs creates specifications and performs tests for cable operators from around the world. CableLabs has been diligently working on two areas of virtualization: SNAP and TIP.

CableLabs started its SDN/NFV Application Development Platform and Stack Project (SNAPS) two years ago. SNAPS, which is based on OpenStack, was designed to provide a common framework for MSOs' virtualization efforts, which include blending legacy systems with newer ones without disrupting services to customers.

RELATED: Cable playing the ‘fast follower’ toward virtualized, distributed network architectures

CableLabs created the SNAPS project to fill in gaps in the open source community and to help with the adoption of SDN and NFV by its cable members. While cable operators have pursued NFV and SDN on their own, including the rollout of Comcast's Active Core SDN-based network and the launch of its OTT SD-WAN service, they need a common framework that the industry as a whole can design towards.

"The ultimate goal is to get to a place where our operators could deploy the SNAPS infrastructure and run their virtual CTMS, their virtual CCAP, their virtual baseband units for 4G or for 5G on the same SDN and NFV infrastructures so that the headend really does start to look like a data center," said CableLabs' Joey Padden, principal architect, wireless. "The headend requires dedicated hardware but, hopefully, in the future it will all share general compute servers and an orchestrator will start up an extra CCAP instance if it looks like its necessary or an extra vRAN instance."

To date, CableLabs has used SNAPS to deploy a virtual Converged Cable Access Platform (CCAP) core by employing a vCCAP from Casa Systems along with hardware from Intel and Case.

SNAPS could be used to more quickly onboard virtual systems or virtual network functions (VNFs.) SNAPS also takes on integration testing by using CableLabs' subsidiary Kyro, which, is a for-profit division of CableLabs.

With SNAPS, CableLabs developed an OpenStack API abstraction library that it contributed to the Linux Foundation's Open Platform for NFV (OPNFV) project.

SNAPS includes integration with NFV orchestrators using the OpenStack NFV orchestrator as well as support for containers and Kubernetes. CableLabs is also looking at how to use SNAPS for server-less computing.

The SNAPS project is led by CableLabs' Randy Levensalor, lead architect, wired technologies. CableLabs posted a guest blog last week that outlined a specific use case for a SNAPS deployment. 

TIP moves towards field trials

Also on the virtualization front, CableLabs announced its Telecom Infra Project (TIP) Community Lab in November. TIP, which is a global initiative that was founded by Facebook in 2016, is working on creating a more decentralized approach to building and deploying infrastructure for telecom networks.

In its lab, CableLabs is working with its members on developing a non-ideal fronthaul interface using DOCSIS that would enable wireless radio networks to work with wired infrastructures. Virtual RAN highlights the importance of leveraging the fixed infrastructure to deliver a mobile solution by virutalizing part of the network

"At CableLabs, 27 of the 60 operator members we have globally have mobile operations as well as the traditional HFC infrastructures," Padden said. "The TIP virtualized RAN fronthaul projects started up specifically to address non ideal IP links."

Non-ideal links have less throughput than gigabit and fiber services, as well as more latency than gigabit point-to-point fiber. Padden said that DOCSIS, which includes DOCSIS 3.0, 3.1 and, eventually, full duplex, qualified "pretty squarely as a on-ideal fronthaul IP link."

"One goal of the project is to create a fronthaul protocol that is completely open so multiple vendors can produce products for either end of the fronthaul, either the remote radio or the virtualized baseband unit," Padden said. "Because it's interoperable, you can get your baseband unit from vendor A and your remote radios from vendor B or vendor C and it will all work. That's pretty different from previous fronthaul links like CIPRI where even though the CIPRI spec was published, more or less, it didn't actually interoperate. So the goal of this project is to create an open spec that is complete enough that it is actually interoperable."

"The second goal is to produce one fronthaul protocol that goes from non-ideal links all the way to the other end of the spectrum. If an operator deploys some mobile infrastructure over fiber for something like macrosites and some other mobile infrastructure that goes over say a DOCSIS plant, you can use the same fronthaul protocol."

While CableLabs is leading TIP's non-ideal IP links for DOCSIS, BT's TIP Lab is focusing on fronthaul for G.Fast and managed Ethernet. Telecom Italia's lab is looking at fronthaul for PON networks and Bharti Airtel's piece of the fronthaul protocols is working with multihop microwave transport.

"So each fronthaul for each lab had different throughput characteristics and different latency characteristics, but pretty much all them—for one reason or another—would classify as non ideal," Padden said. "So we're trying to create one protocol that works across all of these labs."

The TIP requirements for the fronthaul protocols included working with 4G LTE while also being 5G radio ready.

According to a TIP whitepaper, each lab started at "phase zero" with nonproduction-grade proof-of-concept solutions, which the labs tested for six months. After the successful testing, the labs recently started "phase one," which will also last six months.  The goal of phase one is to turn the nonproduction-ready proof-of-concept solution into deployable products.

"So if everything goes smoothly, by the end of the year we should be setting up field trials with operators," Padden said. 

CableLabs has other projects underway that were created to make mobile applications easier to deploy over cable operators' HFC plants. One of those projects is the mobile backhaul project, which is led by CableLabs' Jennifer Andreoli-Fang, distinguished technologist, Office of the CTO and R&D.

One of the goals of Andreoli-Fang's mobile backhaul project is reducing the latency in DOCSIS so that it's more mobile backhaul and fronthaul friendly. Reducing latency in the DOCSIS network will unlock additional potential for cable operators, according to a blog by Andreoli-Fang.

"While both Jennifer's mobile backhaul project and the TIP vRAN front haul project work separately, they work even better if you deploy both of them and get the benefits of vRAN over DOCSIS while also getting less latencies from the DOCSIS network," Padden said.