The push to deploy fiber in the local loop is accelerating, given all the gaga over gigabit, but network operators are still facing definite and sometimes difficult choices about when to deploy and how to most effectively put fiber into their networks. Technology changes, customer behavior shifts and evolving economics are all playing a role in those decisions, which influence massive capex budgets in what is arguably the most expensive part of the network.
The result is a wide range of varying strategies that incorporate the newest in fiber, hybrid fiber-coax, copper and wireless technologies. The primary reason for that diversity, even in the face of fiber's popularity is two-fold: cost and customer demands/expectations.
"The significant percentage of our annual network capex goes into access network," Jeff Finkelstein, executive director of access architecture at Cox Communications Inc. told a Denver audience at Light Reading's Next-Gen Cable event in March. "We satisfy customer demand growth by managing available plant bandwidth and the subscribers sharing the bandwidth."
So while satisfying growing customer demand for bandwidth is the clear goal, accomplishing that at peak usage times means careful planning to make the most of today's network. As a result, Cox will add carriers and split service nodes to reduce the number of customers sharing bandwidth, while preparing for next steps such as DOCSIS 3.1 and a fiber deep strategy, he noted. Even DOCSIS 3.1 will have low penetration rates for its 1Gbit/s services over hybrid fiber-coax initially.
Achieving higher penetration rates of higher bandwidth services "requires hard business decisions," Finkelstein noted. The current access network strategy is built on "node +5" or a cascade of five amplifiers between the subscriber and the home. Pushing the fiber deeper to a node +1 or node +0 strategy "is the most expensive thing we can do" to raise bandwidth at higher penetration levels, he said.
Reducing those amplifier cascades does simplify the physical network and complexity and offer some operational cost savings, he notes, but that doesn't put Cox in a rush to rip out its HFC plant entirely to install fiber. Instead, the company looks constantly at the mix of subscribers in a given service group, the service tiers, fluctuations in peak hour usage and many other factors in deciding where and when to push fiber closer to the customer. This complex series of decisions lets the operator make the best use of its existing plant and be positioned for what it knows is likely to be a fiber-based plant at some point in the future, Finkelstein said.
Similar thinking was on display a month later, as a panel of operators at Light Reading's Gigabit Cities event in April discussed access strategies. All see the path to deploying fiber-to-the-premises in their future, but the strategies vary widely. Even AT&T Inc. (NYSE: T), which is pushing its Gigapower service to approximately 12.5 million customer locations by 2019, is also rolling out G.fast on twisted pair and coax, said Eddy Barker, AVP-technical design and architecture, where those technologies make more sense.
Comcast Corp. (Nasdaq: CMCSA, CMCSK) is gearing up to deliver DOCSIS 3.1 across its footprint for gigabit services, starting in Chicago, Detroit, Miami and Nashville, but has to prioritize how and where it scales that, Rob Howald, VP-network architecture said at the Gigabit Cities event. "We have to look at a lot of factors, including how much network congestion is in that area, the distribution speeds in that area, how different municipalities have different rules and regulations, and we can't ignore the competitive factor," he said. "Our focus in a [given] footprint that is ours is to satisfy our customers as best as we can."
That kind of market-by-market focus explains why FTTP varies by operator, even as everyone agrees the ultimate goal is to push fiber as deep into their networks as possible.
And while today's FTTP deployments are largely GPON-based -- with the exception of cable operators looking at Ethernet PON -- operators are looking ahead to the next technologies, including XGS PON and NG-PON 2.
The former is a symmetrical 10Gbit/s option that was introduced in the Full Service Access Network (FSAN) last year, says Jay Wilson, SVP-technology and strategy for Adtran Inc. (Nasdaq: ADTN). NG-PON 2, on the other hand, brings a new level of optical networking to the access arena, using technology once reserved for the core, he comments in an interview with Light Reading.
Verizon Communications Inc. (NYSE: VZ) is already moving aggressively on the latter front, issuing a request for proposals with vendors, according to industry reports, after testing the technology in 2015. (See Verizon Preps Next Major Broadband Upgrade and Verizon Revs Up Wireline Race With NG-PON2.)
Wilson sees this latest standard as a possible end game for PON and local access fiber because of new flexibility it brings to addressing the seemingly never-ending need for bandwidth. The combination of growing video traffic, the coming of 5G wireless and the rise of the Internet of Things are all expected to drive bandwidth, ubiquity and availability requirements up.
"I think from a standardization perspective, the industry recognized early on that it needed an ultimate flexible high-bandwidth scalable solution that can go up to 40 or 80 gig, that gives you a lot more runway, many more years of runway to your access investment," he says.
Interim steps such as EPON or XGS-PON may work for some operators near-term, but the longer-term play is one that allows the network to be upgraded without changing out the distributed pieces such as optical network terminals, he said.
"NG-PON 2 uses TWDM [time and wavelength division multiplexing] to offer multiple wavelengths and then uses tunable optical technology that has traditionally been used in the core network and applies that not only in the access network but at the end of the access network at the ONT, which allows the ONT to be placed with longevity or sustainability," Wilson explains. "So you can initially deploy a gigabit service to that home or business location or wireless cell site but over time can upgrade to 10-gig symmetric capacity per wavelength."
That means starting with a 40-gig standard for access, and having the ability to add another 40-gig later.
The other argument for the move to NG-PON 2, Wilson argues, is the need for bandwidth to support backhaul for 5G wireless, whose millimeter waves will travel shorter distances and thus require more towers and much more backhaul. The standard today would allow point-to-point WDM connections over the same fiber infrastructure to support that backhaul.
That gets NG-PON 2 closer to being a universal fiber access system that is affordable sooner rather than later, he says, and doesn't look to justifying its cost based on single market economics. Of course, Adtran is one of the companies vying for Verizon's NG-PON 2 business.
Verizon remade the local loop fiber market with its original FiOS build, helping drive down prices and push standards that had an impact on the broader industry. It may be positioned to do the same thing with NG-PON 2 in what is obviously a very different market today.
— Carol Wilson, Editor-at-Large, Light Reading