"A small number of premium tier and (small) businesses might need to get pulled off for fiber-to-the-premises, but for the large majority of subscribers, HFC should be plenty," said John Ulm, fellow of the technical staff within the ARRIS (NADSAQ:ARRS) Network Solutions CTO group.
The cable industry's focus, however, will shift from splitting nodes to enhancing the plant in order to get a higher burst rate or service tier, said Ulm, who recently presented a webinar, "Migration Paths to Full CCAP Functionality," for the SCTE Canadian Summit, 2014.
"If service groups get smaller, bandwidth usage gets cut in half, but the service tier doesn't (increase)," Ulm said. "The amount of time you can buy from a split gets less and less. At the same time, the cost of node splits is going up. Operators will hit the point where the economics of making node splits starts to go away."
Moving to DOCSIS 3.1 will give operators the tools to compete with a 10 Gig EPON system and a 50% increase of capacity within the existing spectrum. "This is really huge. Add new spectrum on top (of this) and it is an incremental change," Ulm said.
Integrated CCAP along with analog optics have "plenty of life" and provide a clear roadmap to get to a tenfold increase in service group growth by the end of the decade. Specifically, Ulm said that by 2020, operators should be able to get 200 service groups per CCAP, fit into a single rack.
"There still may be scenarios where you need a distributed system ... but (integrated) will serve many of your needs," Ulm said. "As you look at analog optics, the main takeaway should be that they continue to improve." This means integrated, centralized CCAP might be appropriate for a remote site because analog optics can now handle the distance.
However, Ulm also provided a roadmap for moving to a distributed access architecture, beginning with broadband compressed forward (BCF) in the downstream. This involves replacing the logic in the optics shelf, but keeping the headend intact. The spectrum is digitized and sent over the fiber link in a way that is transparent so that analog video can be sent down to the node.
The second step is Remote PHY, which keeps the MAC core in the headend but pushes the PHY layer to the node. Timing needs to be resolved, Ulm said, but overall, remote PHY has an advantage over BCF in that it makes more efficient use of the 10-Gig Ethernet pipe. "This is where the current technology sweet spot is. It is less complex than remote CCAP and provides better link efficiencies than (BCF)."
Remote CCAP, on the other hand, gives a fourfold space gain over integrated CCAP. By putting the MAC as well as the PHY closer to the subscriber, the MAC "runs better" with lower latencies. "We are hoping to get to the point where remote PHY and remote CCAP is the same implementation in the node," Ulm said.
There is an additional option called a remote shelf, which can be either remote PHY or remote CCAP. "This hybrid is a distributed system, but it is distributing from a centralized headend out to remote hubs and still running analog optics," Ulm said. "You don't have to touch the plant, but you have eliminated a large piece of the analog link. It is the best of both worlds."