By Allen Wang
Data Over Cable Services Interface Specification (DOCSIS) has proven to be one of the most successful broadband access technologies. Since 2008, DOCSIS 3.0 solutions have been widely deployed to meet the even-increasing bandwidth demand of tens of millions of subscribers. To date, the scalability and consistently improving performance of DOCSIS solutions has given the cable industry a competitive edge. As a result, cable broadband subscriptions outweigh telco broadband subscriptions by 40%.
As is often the case, success breeds competition. In the case of DOCSIS, the ongoing transition toward gigabit access is providing a golden opportunity to telcos and other competing operators. This is the dawn of the age of the Internet of Things. Broadband access will be an integral part of general household utilities, and the Gbps access network needed to acquire and retain subscribers is destined to be one the most valuable assets. The current trend of Gbps service offerings isn’t just a way to counter the increasing number of Google Fiberhoods; rather, it is the new baseline requirement for service delivery in the coming decade.
The Gigabit era is reshaping the competitive landscape
For competing operators without access to legacy copper, the focus has been on lowering the cost of FTTH buildout. Through a mix of lower component costs, construction/design innovations and public/private partnerships, per-home FTTH connection cost in the U.S. now averages below USD1,300, compared to USD4,000 for Verizon FiOS just a few short years ago. Moving forward, the FTTH home connection cost is expected to decrease by 7% CAGR. With a slew of recent Gbps projects announced by the telcos, cable is in jeopardy of being outbuilt in the area of fiber. G.fast fiber-to-the-distribution-point (FTTdp) is rapidly becoming a top Gigabit solution, especially for high-density deployment scenarios such as MDU. From a distance of 50 meters, G.fast can now deliver 1Gbps bandwidth on a single twisted pair. A typical G.fast FTTdp node can support up to 16 subscribers with a single drop fiber. With the advent of TWDM-PON, 10G access pipes over GPON can readily provide the uplink capacity required to deliver Gbps services to every subscriber connected to these FTTdp nodes.
With increasingly affordable FTTH and FTTdp with G.fast, the bar is set quite high for the next phase of broadband access competition. While the MSOs can also deploy FTTH for head-to-head competition, which will likely be the case for greenfield projects, the real question is whether hybrid fibre-coaxial (HFC) solutions will be competitive enough in the Gigabit era
Do CCAP and DOCSIS 3.1 give MSOs a competitive edge?
Until recently, most people in the cable industry had a lot of faith in HFC. After all, CableLabs standards specify DOCSIS 3.1 will provide a minimum of 5Gbps of downstream bandwidth by bonding two D3.1 channels together with 24 D3.0 channels. Fully integrated converged cable access platforms (CCAPs) will deliver these high bandwidth pipes to thousands of fiber deep nodes, just like it was envisioned; sounds straight forward enough.
Indeed, after ramping up rapidly in 2013, CCAP has achieved phenomenal growth, essentially replacing the traditional cable modem termination system (CMTS) as the mainstream cable access solution. In Q1 2014, CCAP/CMTS channel shipments were up more than 100% as MSOs continue to build capacity to meet sharply rising data consumption, which increases 40-to-50% annually. However, much of the CCAP being deployed today is nothing more than denser CMTS modules and edge quadrature amplitude modulation (EQAM) platforms. In other words, growing bandwidth demand is being addressed, while space shortage is not. Hub space shortage was identified as one of the key driving factors for CCAP, and the severity of this issue cannot be overstated. By some estimates, most hubs in North America will run out of space if only 10% of the HFC plant is converted to fiber deep nodes, which occurs after about 150 homes passed. Conversion to fiber deep nodes is a critical HFC migration step anticipated to be widely implemented by 2017 in conjunction with the introduction of DOCSIS 3.1.
So, why is there so little progress made in saving hub space after more than one year of CCAP ramp-up? There is no simple answer, but industry sources provided many clues. One of the most basic problems is the sheer complexity of migrating from a modular platform to an integrated platform in a tightly confined space while maintaining normal operation – something said to be as difficult as swapping out a plane’s engine mid-flight.
In most hubs, installing a fully integrated I-CCAP means the existing CMTS must be swapped out and reallocated to another hub, which is typically located in a less populated part of the network where space is available. That way, the CMTS can be expanded to meet bandwidth consumption growth. This swapping-out and relocating exercise also requires warehouse space for temporary storage of equipment en route, making it both a logistic challenge and a time-consuming process.
Perhaps more importantly, the impact of CCAP migration on both MSOs’ and vendors’ business cases also plays a key role in the pace of the migration. After investing tens of billions upgrading the HFC network during the past decade, MSOs will likely stick to a pay-as-you-grow strategy in order to maintain a sound business case. With this strategy in place, it is unlikely we will see any large-scale replacements of existing CMTS with full-spec CCAP chasses. Instead, CCAP migration will be gradually carried out with a variable mix of adding high-density line cards in the existing M-CMTS platform and installing full-spec I-CCAP chasses over several years, in much the same fashion as previous HFC migration efforts. From CCAP vendors’ perspective, the explosive growth in demand coupled with the dynamic platform migration creates many opportunities. Market data suggests that vendors with full-spec I-CCAP solutions are gaining market shares from less progressive incumbents. The number of CCAP channels being deployed worldwide seems to indicate future growth in bandwidth consumption is being adequately addressed. However, the coming crunch in hub space, driven by the massive number of node splits, will require a new migration option that can effectively address the space issues in the hub and headend.
Time to consider distributed access architecture
The HFC outside plant is a highly scalable broadband access facility. Much of DOCSIS technology’s success can be attributed to the superior performance of the tree-and-branch shared coax plant, compared to the point-to-point twisted-pair copper plant. In the age of Gbps broadband access, a properly scaled HFC fiber node stands as one of a few remaining options that are comparable to the bandwidth capacity of FTTH.
The traditional fiber node is a physical layer O/E media converter. For Gbps service delivery with a hub-based CCAP, each fiber node is connected via analog fibers to a pair of upstream and downstream CCAP ports. A number of service groups, including a dedicated DOCSIS service group, are transmitted to a group of subscribers connected to the coax plant.
With a distributed access architecture (DAA) platform, an Ethernet node replaces the fiber node. Instead of analog fibers, digital fibers are used to connect a number of Ethernet nodes to a centralized aggregation point such as an OLT, located in the hub. The Ethernet node is a layer-2 element containing a complete DOCSIS media access control (MAC) and physical layer (PHY) and other components needed to deliver a full suite of service groups. Functionally, the Ethernet node is equivalent to a pair of upstream and downstream ports in the hub-based CCAP modules. The Ethernet nodes are centrally controlled by the OLT and managed as a single network element. Collectively, the OLT and its subordinate Ethernet nodes function as a distributed CCAP (D-CCAP) by combining the functionality of a CMTS with an EQAM, similar to a hub-based CCAP. More significantly, D-CCAP solves the hub space issue and advances the HFC network further down the migration path to reach a higher level of competitiveness, a level that is essential for the MSOs to lead the broadband access race in the Gigabit era.
Key benefits of DAA
The space savings realized by replacing CCAP with the OLT is significant. A next-generation OLT is designed to support a minimum of 32K subscribers connected though a minimum of 450 Ethernet nodes from a single 11 rack unit chassis. To cover 30,000 homes passed from a single hub, assuming a highly dense 13RU large I-CCAP chassis capable of supporting 64 fiber nodes, it would take four I-CCAP chasses.
With D-CCAP, all 200 Ethernet nodes can be covered with 7 GPON line cards, about half of the available slots in the OLT, with the remaining slots available for FTTH. Given the severity of space shortage in many hubs and headends, it is expected that many of these hubs will need to switch to the D-CCAP platform in order to support the coming conversion to fiber deep nodes.
For migration to DOCSIS 3.1, it is well-understood that replacing the analog optical transmission system with a digital optical system such as PON, between the hub and the node, provides a 3-7dB improvement in end-to-end SNR, increasing bandwidth by up to 25%. In the near-term, for many HFC networks, the top-end of the downstream spectrum is expected to be 1 GHz. Before a transition is made to full IP video, the availability of downstream RF spectrum for DOCSIS will be extremely limited. Consequently, the bandwidth gained by converting to a digital optical system adds significant value to DOCSIS 3.1, making it more competitive against FTTH.
D-CCAP is the mainstream DAA solution
Recognizing the market demand for maximizing DOCSIS 3.1 performance, some hub-based CCAP vendors are preparing to offer remote PHY (R-PHY) where a digital optical system is inserted between a new D3.1 MAC module and a remote digital node with a built-in D3.1 PHY. Although the proposed R-PHY solutions can potentially allow hub-based CCAP to hit the DOCSIS 3.1 performance target, the approach reverts a fully integrated CCAP back to a modular architecture as described in CableLabs’ Modular Headend Architecture version 2 (MHAv2). In doing so, it created a DAA option for the M-CMTS platform that deviates from the migration path set forth by CCAP and does little to solve any of the issues addressed by I-CCAP.
D-CCAP is based on the mature FTTN architecture that has been widely deployed for more than a decade. Many of the FTTN solutions available today are based on the highly evolved PON OLT platform that has been addressing challenges such as density, space and power; the same set of challenges addressed by CCAP. In the Gbps access era, most broadband access systems will be PON- based with FTTH making up an increasing share of the deployment base. For the cable industry, FTTH is already used in most greenfield projects, and there is a growing consensus FTTH would be needed to overbuild parts of the brownfield HFC network facing competitive FTTH entrants.
With D-CCAP, HFC remains competitive. It remains to be seen how competitive DOCSIS 3.1 and CCAP class of solutions will be against competitors’ FTTx solutions. In addition to the size and scalability of the access pipes, many other factors are just as important in determining the outcome. But as history has shown, the advantages of HFC network gave MSOs a strong competitive edge in the pre-Gigabit era. Holding on to this advantage in the access infrastructure should be a prime objective.
Against a backdrop of uncertainty, it would be prudent for the MSOs to consider a scenario where FTTx may be needed sooner, and in a larger part of the HFC network. An overlay PON optical distribution network (ODN) provides a flexible FTTx migration path. With D-CCAP, MSOs can select a lower-cost FTTN option capable of delivering multi-gigabit bandwidth to around 75 subscribers from fiber deep nodes. When an all fiber business case presents itself, FTTH can be built out to business and residential customers. The highly efficient OLT can easily support FTTx expansion from any hub or headend and the small footprint makes it possible to coexist with other hub-based CCAP platforms, which facilitates a smooth migration process.
In the coming few years, as we approach the target timeline (2016/2017) for both fiber deep node splits and DOCSIS 3.1 rollout, it will be increasingly evident that the broadband access competitive landscape has changed with accelerating build-out of fiber access networks. As a cable-exclusive FTTN solution, D-CCAP provides an all important competitive edge and should be part of the HFC migration planning for any cable operator.
