Tasked with answering the question of "Why is a Remote PHY Architecture is better than a Remote MAC/PHY architecture" it would be prudent to articulate first what advantages Remote PHY has to offer over other proprietary solutions and let the readers draw their own conclusion. Let's first call out immediately that theonlyratified specification for DAA (Digital Access Architecture) is that of "Remote PHY", whereas Remote MAC/PHY is classed as a "Technical Recommendation." That immediately should raise some concerns about how a "technical recommendation" can be interpreted differently by vendors and silicon manufactures alike. The whole remit around standardisation is to drive interoperability and cost effectiveness with a standard, however, that advantage is diminished for the MSO with no more than a suite of guidelines and recommendations.

So what exactly is Remote PHY trying to solve for the MSO that its contemporaries cannot.

Let's break some of the key advantages out for discussion.

1. The point about Standards vs Proprietary is already made, but the first advantage of R-PHY would be the independent scaling between the MAC and the PHY layer. Scaling these individually immediately adds a cost benefit to the MSO. In contrast a combined MAC/PHY concept with some future forecast bandwidth requirement would necessitate the need to provision from day one "full capacity" for the MAC. A MSO having a view to potentially scale to 10G per SG would actually mean provisioning that initial 10G, even though the immediate aggregate requirement would be far less at time of deployment. The question of stranded capex is brought in to question with a MC/PHY approach.
2.  Centralizing the Software to the CCAP Core has the effect of amortising the costs and efficiencies across multiple RPD's. Any ECN's or updates to the DOCSIS MAC can be effected immediately in a more effective manner. Yet it is noted that even in a MAC/PHY with the aid of a "pseudo" Virtualisation of the nodes, that automation for upgrades can be achieved, however I see this as counterintuitive to the fact that one of the key drivers for Remote Phy DAA was to simplify the network and have as few as complex elements as possible external to the CCAP core. Centralizing the SW also adds for a more effective way of adding not only additional MAC capacity mentioned above but also scaling out the DOCSIS Scheduler. In a DOCSIS 3.0 world the CPU impact on the DOCSIS scheduler is proportional to the data rate, in a full spectrum DOCSIS 3.1 world the full the absolute impact on the scheduler is not known thus like above, to minimize unnecessary field upgrades to MAC/PHY nodes, a centralized SW construct offers a greater advantage.
3.  A consistent architecture for all services is something that a MSO take for granted to drive for. Remote PHY offers exactly that. It uses the same protocols for DOCSIS, Video and OOB; Think of the Remote PHY protocol as a pseudo wire manager able to support multiple services over a common architecture. Makes sense right? Then what would be the compelling reason to shift away from that level of simplicity to add further complexity to an already "siloed" solution for Video, DOCSIS and OOB and use R-PHY for Video and R-CCAP for a DOCSIS.
4.  Maybe contrary to popular belief, but Remote PHY should be considered as a natural evolution to the existing CMTS architectures for Cable. DMPI (DOCSIS MAC -PHY interface) gave us the Silicon to silicon interface on a line card, M-CMTS took that a further stage by remoting the DS element on an Edge QAM whilst Remote PHY completed the evolution by providing both DS and US remote PHY elements. So in my mind, given the fact that the intellectual property has been developed over the past decade (2000 for DMPI and 2005 for DEPI) I am inclined to favour an established suite of protocols and interfaces rather than going through revolution of architecting the MSO to adapt a "split R-MAC/PHY software model" and the pain that goes with the initial deployments.
5. It's stated that a MAC/PHY approach is common with other technologies and not a new concept in the Networking and Telecoms industry. Yet opinions would suggest that it is a rear view perspective of where Networks have been previously architected yet with "Virtualisation of Networks" a new approach of Centralised Software and a simple Remote Hardware is required. To support the above comments I subscribe to the notion that it's not just CCAP/CMTS that will adopt a "centralised Software" with Remote Hardware to facilitate "True Virtualisation", technologies used in the Mobile , PTT and OTT space will also adopted  the same approach and subscribing to the same approach. I fear the case made for Remote MAC/PHY may meet some of the immediate objectives but has little life expectancy for true NFV if both MAC and PHY are remoted.
6. With the simplicity requirements of the (RPD) Remote PHY Device, it will be possible to reduce the implementation down to a single piece of silicon. This silicon could be located deep in the network at a tap location and convert from a FTTC plant back to an HFC plant in order to maintain backwards compatibility with existing CM and STB but in the same time adds a platform for support of new and emerging developments within DOCSIS. A simple end point to facilitate Fiber to the Curb and DOCSIS to the home with potential for DOCSIS EFR (Extended Frequency Range) DOCSIS FDX (Full Duplex DOCSIS) R-MACPHY and R-CCAP have too complex of an end point design that will not provide them the evolution velocity, and same reduction on power and cost. Do you really want an entire CMTS at the curb?
7.  Last, but certainly not exhaustive of items to compare but industry acceptance is always key. There is an abundant of silicon talent in the industry that can design a PHY chip. It however requires a different skill set and more investment to get the MAC right. A number of prominent Silicon, manufactures, MSO's and vendors have endorsed the Remote PHY architecture and are building towards a common DAA goal. This in turn will lead to a greater level of innovation and ultimately cost effectiveness which as I mentioned above is one key attribute of driving a common standard.

With Vendors moving ahead with expediency to bring to market RPD's (Remote Phy devices) in multiple guises and support from silicon manufacturers engaged in the standards and product initiatives, the momentum for Remote PHY as a DAA architecture of choice is overwhelming.

In parallel MSO's are adapting strategies to leverage the deep fibre Remote Phy deployment to augment current capacity needs as well as a springboard to future "Virtualisation of the Core CCAP elements.

To learn more, visit the Cisco CCAP page.

Save