Many cable operators have clearly telegraphed plans to move to DOCSIS 4.0 as competition from fiber players intensifies. But the question of which of the two versions of the standard they will roll out remains to be settled. In the end, Dell’Oro Group VP Jeff Heynen told Fierce it could end up being more a matter of timing than technology preference.
Operators chasing DOCSIS 4.0 are largely divided into three camps: those like Comcast, which are pursuing full duplex DOCSIS (FDX), another cohort including Charter Communications, which has lined up behind the extended spectrum variant (ESD, also known as frequency division duplex or FDD) and those who have yet to pick a path forward. Testing of both FDX and ESD is chugging along apace, with both Comcast and Charter touting recent breakthroughs with their respective partners.
FDX uses the same 1.2 GHz spectrum allocation called for in DOCSIS 3.1 and was originally envisioned as the node+0 option for operators running fiber deep networks. Meanwhile, ESD requires a 1.8 GHz allocation but allows for multiple amplifiers in between the node and the subscriber.
Heynen said if industry players like Comcast can prove out an FDX system that is compatible with a node+2 architecture relatively soon, that could change how operators view the different flavors of DOCSIS 4.0.
He explained most operators are upgrading their networks right now with amplifiers which support a 1.2 GHz allocation. “If you’re upgrading now, why would you in just a few years upgrade to support extended spectrum if you have an option available, another option, that gives you the type of bandwidth that you were looking for with 1.8 [GHz ESD] within that 1.2 [GHz] spectrum with equipment you already have deployed?” he asked.
Added pressure to deliver multi-gig speeds as competition from fiber overbuilders increases could also factor into operator calculations around DOCSIS 4.0, Heynen noted. New Street Research recently forecast fiber will pass around 80% of cable homes by 2030.
“If you could wait ‘til 2030 to get everything ready for your 1.8 [GHz] extended spectrum network then fine. But if you’re starting to be faced with competition from fiber overbuilders like an AT&T that’s offering a symmetric 2-gig service and a symmetric 5-gig service and you start to lose customers in droves to that kind of service, then that changes the timing I think,” Heynen said. “If Comcast can make that work within a timeframe that allows an operator to once again be competitive with that kind of service, then they’re going to do it.”
Cisco CTO of Broadband Technologies John Chapman told Fierce, “The future is fiber for everybody, including the cable operators.” So the question of whether to go with FDX, ESD or even just to ride out DOCSIS 3.1 until the end of its useful life comes down to “what’s the right investment profile into the plant to go from the present, which is DOCSIS, to the future, which is fiber.”
It’s worth noting there’s also a political element to decisions being made around FDX and ESD. As Chapman put it, it’s not just a technical choice between the two technologies because each comes “with a set of business and strategy choices.”
Are we there yet?
Comcast’s most recent FDX trial demonstrated a full end-to-end system, all the way from the virtual cable modem termination system (vCMTS) through the node and on to CPE equipped with a prototype DOCSIS 4.0 modem. But that trial relied on a node+0 architecture.
When the announcement was made in January, Comcast Cable’s SVP of next generation access networks Elad Nafshi told Fierce it was working on an amplifier “that’s going to be enabling us to deliver multiple gig symmetrical services over a cascade of amplifiers on our digital plant. And there’s more to come on that later this year.”
According to Chapman, though, it’ll likely be 2025 before FDX and ESD are ready for prime time. He noted it generally takes about a year or two from the time trials are conducted in the lab to get to full productization and deployment. That’s because once hardware testing is complete, software interfaces need to be written and revisions need to be made to the silicon for the final product, he explained.
In terms of the hurdles specific to each technology, Chapman said FDX in a node+2 setup is a bit more complicated because of the amplifier issue and likely a bit more expensive. While ESD is a bit simpler since it’s primarily a spectrum upgrade, it is also expected to be coupled with deployments of Remote MAC-PHY, which will require rebuilding an operator’s CMTS footprint to adopt a distributed rather than centralized model.
“Qualifying a new CMTS is a lot of work. So that’s going to slow down the ESD camp,” he said. “That will end up giving FDX a competitive advantage. So even though FDX is more technically complex and expensive than ESD, it’s less technically challenging than ESD plus Remote MAC-PHY.”
Chapman added the ESD camp was recently hit with a “pretty huge setback” when CommScope reportedly canceled a Remote MAC-PHY project that was based on a MaxLinear chipset. CommScope declined to comment on the status of its silicon collaboration with MaxLinear when questioned by Fierce last month.
Responding to questions about whether it might shift its strategy to pursue FDX should it become available in a node+2 configuration before ESD is ready for prime time, a Charter Communications representative referred Fierce to comments made by its CEO Tom Rutledge on a recent earnings call. These referenced the operator’s planned high-split upgrades as well as an ESD trial conducted alongside with Vecima Networks in January, which achieved speeds of 8.5 Gbps.
Answering the same query, a Cox representative told Fierce it remains focused on ESD and declined to speculate beyond that. Meanwhile, Mediacom said it is currently working with Node+2 designs that utilize DOCSIS 3.1, but is eyeing “some promising developments that could allow Full Duplex DOCSIS compatible active designs to work within Node+2 architecture.”
This story has been updated to clarify that CommScope allegedly canceled its Remote MAC-PHY project rather than MaxLinear canceling the chip.