Is GPON the access technology for the next decade?

GPON dwarfs most other access technologies in terms of bandwidth deliverable to the subscriber. And GPON economics, both capital and operational, are superior to those of other access technologies. In fact, within the cable industry it is conceivable that GPON will be a successor to DOCSIS 3.0.

Th 283098

GPON dwarfs most other access technologies in terms of bandwidth deliverable to the subscriber. And GPON economics, both capital and operational, are superior to those of other access technologies. In fact, within the cable industry it is conceivable that GPON will be a successor to DOCSIS 3.0.

By David Cleary, Calix

Technological innovation often presents an intractable dilemma: too many choices. When applied to networking technologies—especially within the access network—a multiplicity of choices often results in negative side effects such as higher cost and poorer interoperability. Costs tend to be higher because the components associated with an individual technology can't achieve the volumes required to deliver scale economies. And interoperability tends to be poorer because the set of suppliers pushing individual technologies isn't large enough to drive robust interoperability regimens.

The telecom access network has faced just such a dilemma over the past several years: Copper or fiber? ADSL2+ or VDSL2? Fiber-to-the-node or -curb or -premises (FTTN, FTTC, FTTP)? Passive optical networking (PON) or point-to-point Ethernet? GPON or EPON? Fortunately, this dilemma is being resolved. With increasing frequency, GPON is viewed as the optimal technology for building access networks.

The reasons that GPON is increasingly dominating the access network are manifold, especially, but not exclusively, within North America. First, GPON satisfies the two most critical components of the technical debate: high bandwidth and support for RF video. But more importantly, the economics of GPON are so compelling that they provide network planners (even cable network planners) with a more straightforward decision.

Technical attributes
Bandwidth. Many within the industry continue to engage in the how-much-bandwidth-is-enough debate. While the raw numbers themselves are instructive, this debate too often ignores what bandwidth really means to the subscribers: response time. Yes, the sheer number of bits being moved does indeed necessitate higher capacity. Throw in multiple high-definition IPTV streams, a handful of standard-definition streams, and a lightning-fast Internet access service, and you're pushing well beyond copper's reasonable capacity to support tomorrow's services. But humans are unable to "see" bandwidth; rather they perceive it based on how long it takes to download or upload something. And what they want to see is obvious: instant gratification. This is especially important with applications such as gaming, voice over IP (VoIP), and personal video conferencing, which are relatively modest in terms of bandwidth demands but extremely intolerant of network-induced delays.

What this means is that service providers will increasingly be driven to offer Internet access speeds that for one home may be dozens of megabits per second, but hundreds of megabits per second to the house next door. And these services will need to operate symmetrically. An Internet service that could spike to a gigabit per second might be highly desired not because subscribers can actually consume or generate a billion bits every second, but because they want (and are willing to pay for) the nanosecond-per-bit response time characteristic associated with such a high capacity service.

So what technologies are best suited to deliver the high-bandwidth quick response time demands that will only worsen over time?

Th 283098

RF video support. One of the major drivers behind the radical increase in access network bandwidth is video. And while IPTV and Internet video are being widely deployed, more than half of the traditional telcos delivering video over FTTP today are using an RF video overlay on the PON to do so. This approach leverages an existing RF headend and uses the 1,550-nm wavelength to deliver the entire channel lineup to the home over fiber, then leverages the existing in-home coaxial infrastructure to distribute throughout the home in a manner virtually indiscernible from the traditional cable service. For many service providers with existing cable headends, high market demand for basic analog cable services, or a skepticism regarding the maturity of IPTV technology, this architecture makes sense. Overbuilding a service area with fiber is not a trivial undertaking, and many service providers (including Verizon, for example) are mitigating risk by using cable TV-like RF video to deliver the linear channel lineup and sometimes using IP for video-on-demand (VoD) and various forms of internet video. This gives them a proven video delivery system while still enabling a shift to IPTV if/when circumstances warrant.

So support for RF video is requirement number two. How do our access alternatives stack up?

Th 283101

There are certainly a number of other technical attributes we could study (support for DS1s, flexibility, etc.), but bandwidth and RF support are the two biggest factors that service providers focus on when analyzing these technologies. And in this context, only GPON passes both tests.

But in the end it all comes down to fundamental economics.

Economics 101
Over the past few decades we've witnessed a number of game-changing technology debates: Ethernet versus Token Ring, VHS versus Beta, etc. While these debates concerned mostly technical issues, they were only settled when one technology secured sufficient critical mass to gain an advantage in terms of cost and interoperability. However slight that edge may have been, it typically signaled a tipping point where the advantage of one alternative began to widen at an accelerating pace. Once that tipping point is reached, it's "game over." All other technologies will be discarded or relegated to niches.

Cost is especially important. After all, unless a given technology meets certain cost targets, it cannot be profitably deployed in a service provider network. Some of the cost characteristics are inherent in the technologies themselves. PON architectures, for example, benefit by employing a single port in the central office (an optical line terminal [OLT] port) to support 32 to 64 downstream subscribers and also by employing a passive (i.e., non-powered) outside plant. But in the long term real cost advantages become almost entirely attributable to volume.

Increased volumes produce economies of scale as well as an ecosystem of competitors attempting to drive cost down. As Fig. 1 illustrates, GPON is precisely at that proverbial tipping point, at least within North America.

Th 283102

Figure 1. GPON is forecast to become the predominant FTTP technology in North America in 2008 and beyond. (Source: Infonetics Research)

Given this reality, network planners opting for technologies other than GPON for anything other than niche applications (for example, point-to-point Ethernet is well suited for large business applications) are depriving their organizations of the economic benefits afforded to users of the dominant technology. I'd hate to have been the last network planner to recommend the implementation of a Token Ring LAN as opposed to Ethernet, but someone was.

The cable factor
While telcos are in almost universal agreement that GPON is the optimal technology for their access networks because it will give them a strong foundation and competitive advantage for decades to come, cable operators face a tougher decision. Although DOCSIS 3.0 and a variety of node splitting and spectrum enhancement techniques will meet the near-term bandwidth demands using existing hybrid fiber coax (HFC) infrastructure, FTTP technologies provide an interesting alternative that balances their need to leverage existing infrastructure with more flexibility.

Although DOCSIS 3.0 will provide a significant increase in bandwidth capacity deliverable to the home, it still cannot deliver the bandwidth that telcos are able to marshal today with GPON (especially with newer 2.5-Gbit/sec GPON optical network terminals [ONTs] sporting Gigabit Ethernet subscriber-side interfaces). Figure 2 compares the average bandwidth deliverable over GPON and DOCSIS 3.0 at various split factors. In a bandwidth-hungry world sensitized to response times, average bit rate is much more meaningful than peak bandwidth. But even in looking at peak bandwidth, GPON wields a 6:1 advantage relative to DOCSIS 3.0. Although FTTP is unlikely to be widely deployed as an immediate replacement to the HFC infrastructure, it is clear why many cable operators are beginning to evaluate GPON for their long-term planning.

Th 283104

Figure 2. Although both are shared networks, GPON holds a significant downstream bandwidth advantage per home passed over HFC networks in typical deployment scenarios.

GPON: The optimal access technology
In summary, optical fiber represents the optimal waveguide for delivering digital information to business and residential subscribers. Arguably, PON technologies are the optimal form of FTTP, and GPON is the optimal form of PON. With the economic tipping point having been reached, telecommunications and cable service providers alike are increasingly electing to employ GPON in all greenfield builds. Increasingly, first with telcos, but eventually with cable operators as well, we expect to see all service providers moving toward an end-to-end fiber infrastructure. As this takes place, GPON, with its superior bandwidth and operational flexibility, is emerging as the technology that will dominate all access networks in the future.

David Clearyis vice president of advanced technology with Calix. He is responsible for identifying new and emerging technologies that will lower cost and improve performance of access networks. Cleary may be reached via the company's web site at

More in FTTX