The origins of most recent technological innovation in networking can be traced back to the growing need for bandwidth and this in turn to the demand for streamed video, multiple IP devices in homes, etc. Starting almost a decade ago, MSOs began to consider Converged Cable Access Platforms (CCAP) as a way to replace edge QAM and CMTS units with a single platform that better handled and integrated the data and video traffic that was growing exponentially. From this beginning conventional CCAP evolved into distributed CCAP architecture, which itself became part of a bigger technology trend; Distributed Access Architecture (DAA).
DAA enables the MSO to relocate functionality that has traditionally lived in a headend or hub to nodes, network cabinets or even multi-dwelling units much closer to the subscriber. By putting remote PHY technology into an optical node in an HFC plant, fiber portion of the HFC plant become digital. Historically, the HFC forward path has used analog optics, and the reverse path has used both analog and digital optics.
This new architecture helps relieve crowded real estate and power/cooling constraints at head-ends and hubs. DAA architectures can also ultimately expand the DOCSIS spectrum up to 3 GHz, although this is several years away. However, the emerging 1.8 GHz technology can buy the cable television industry another 10 years or more in terms of enhanced capacity. Meanwhile, a scalability path is being built through the use of housings that can support 1.2GHz and 1.8GHz technology but can also be retrofitted with 3GHz component.
The other advantages of DAA too. The distances traversed by the network can be doubled or more, maintenance costs are reduced and so are hardware costs – which now will mostly be based on 10G Ethernet, providing a cost-effective way to add new services. At the same time, much of the complexity is kept centralized where it is easier to maintain , although it is harder to change with deploying an overlay network. DAA offers a huge advantage for MSOs by using digital fiber links between the headend and R-PHY node, resulting in better signal quality in the coax plant. Lastly, the use of digital fiber allows support for more wavelengths per fiber and a path for future evolution to FTTx.
The Remote PHY Approach
Various possibilities exist for implementing DAAs. The exact choice of DAA approach varies from MSO to MSO and depends on and what service groups need to be implemented in any particular geography. The most popular approach for now is to distribute the physical layer (modulation/demodulation/signal generation) to the node – this is usually referred to as Remote PHY (R-PHY). Here PHY circuitry is relocated to the access network in the form of a separate R-PHY device (RPD) which itself is located in a standalone chassis called a shelf, or inside a node. R-PHY is already well supported by vendors and is relatively easy to deploy.
In addition to the basic standalone RPD it is also possible to buy a node with RPU functionality already installed. In addition, some companies make special shelves designed to hold multiple RPUs. Functionally, the role of RPU is to (1) convert downstream MPEG video and out-of-band signals received from the CCAP over a digital medium to analog for transmission over RF and (2) also carry out the reverse process. With R-PHY, links become fully digital and hence more reliable, requiring less maintenance. There are also significant SNR gains enabling higher modulation orders for DOCSIS 3.1 downstreams.
The R-PHY DAA is an additional evolution in DOCSIS data and MPEG video delivery for cable operators. In addition to R-PHY, DAA thinking supports two other options for the future called remote MAC/PHY and split-MAC. In R-PHY, only the PHY moves, the MAC layer remains at the headend or hub site in the CCAP core.
A Boost for Optical Networking
A 10G Ethernet connection links the RPD to whatever remains at the headend. (Of course, the headend continues to house the usual servers, switches and routers at the headend and – assuming the MAC functionality is also not distributed – the MAC circuits as well.) This aspect of the cable television hardware market is often understated in the DAA literature in terms of the considerable demand it will bring to the optical components market. Most articles on DAA stress the abstract advantages of distribution, not the components required. But in practice, the components needed to build an RPD optical architecture are quite demanding and significantly change the technology required towards the optical.
A typical 500-100 HFC home node may be split into multiple N+0 nodes with 10- 20 RPDs serving an ideal number of 50-4 passed home Each RPD is served by a 10G optical connection providing sufficient bandwidth. Suddenly this takes the level of connectivity required to a new level. A 10G link is little more than a commodity, but a (20 x10) 200G link is much more leading edge. Additional bandwidth may also be may also be needed to support PON networks as well as high capacity services for businesses, all on the same trunk.
What is being proposed at the present time to meet this need is a bi-directional 10G DWDM trunk or a high capacity coherent optical link.10G DWDM links of the kind that are needed for DAA/RPDs are certainly the way to go right now. They represent mature technology and are relative low cost. Coherent links of the kind needed for DAA are not commercially available in cost-effective field-hardened format. There is also an issue of building networks that enable the coexistence of both 10G NRZ and coherent networking infrastructures.
Lowering the Cost of R-PHY
In any case, the new DAA thinking will stress CAPEX at the MSOs and they will be looking for ways to mitigate this extra expense, even though DAA is long-term OPEX reduction strategy. One way to do this is to innovate new networking options for DAA that haven’t been quite thought through yet. For example, would a link using the relatively new in inexpensive 25G technology have a role here?
The future of cable TV networks appears to be one with “passive in the field,” “fiber deep” architectures with more – likely PHY-enabled – nodes. The passive in the field aspect means reduced power consumption and increased network reliability.