Broadband networking of any sort, but especially broadband to the individual consumer, demands new access strategies. Traditional copper loops were designed for analog voice service. The way voice has been provided has evolved over the years to use more fiber-linked remote digital loop carriers (DLCs). And with the need to develop broadband connections to consumers, DLCs were upgraded to so-called "new-generation DLCs" (NGDLCs) that supported digital subscriber loop (DSL) connections over the same copper pairs. These NGDLCs are fed by a fiber connection, creating what is popularly called a "fiber-to-the-node" (FTTN) architecture.
This increase in the number of optical-to-electrical transformations increases the per-customer cost. Nowhere would that be more evident than the limiting case of fiber deployment, when it is run all the way to the home. Passive optical networks (PONs) were developed to solve the problem of electrical equipment cost multiplication. An industry group, the Full Service Access Network (FSAN) group has been the driver in PON standardization, though the IEEE and the ITU now play major roles.
A PON network is a "tree configuration" of fiber created by splicing fiber strands rather than using an electrical device. The optical splice in the fiber splits the downstream optical signal and combines the upstream optical signal.
All PON architectures have a mechanism (time, frequency, or wavelength-division) to keep the individual traffic for each branch/user separate. The maximum number of "splits" or branches in a PON tree is limited by the loss of optical signal quality generated by both the splicing process and the simple fanning out of the photons to a larger number of branches. The currently accepted upper limit is 64 splits, but in most PON installations the number of splits is held to 32 or even 16.
PON system capacity depends on electrical overlay
All PON systems have essentially the same theoretical capacity at the optical level. The limits on upstream and downstream bandwidth are set by the electrical overlay, the protocol used to allocate the capacity and manage the connection. The first PON systems that achieved significant commercial deployment had an electrical layer built on Asynchronous Transfer Mode (ATM, or "cell switching") and were called "APON." These are still being used today, although the term "broadband PON" or BPON is now applied. APON/BPON systems typically have downstream capacity of 155 Mbps or 622 Mbps, with the latter now the most common. Upstream transmission is in the form of cell bursts at 155 Mbps.
The successor to APON/BPON is GPON, which has a variety of speed options ranging from 622 Mbps symmetrical (the same upstream/downstream capacity) to 2.5 Gbps downstream and 1.25 Gbps upstream. GPON is also based on ATM transport, and this is the type of PON most widely deployed in today's fiber-to-the-home (FTTH) networks in new installations, and it is generally viewed as being suitable for consumer broadband services for the next five to 10 years. From GPON, the future could take two branches: 1) 10 GPON would increase the speed of a single electrical broadband feed to 10G; and 2) WDM-PON would use wavelength-division multiplexing (WDM) to split each signal into 32 branches.
A rival activity to GPON is Ethernet PON (EPON), which uses Ethernet packets instead of ATM cells. EPON should be cheaper to deploy, according to supporters, but it has not garnered the level of acceptance of GPON, so it is not clear how EPON will figure in the future of broadband access.
Broadcast TV signals over PON
One of the attributes of all PON architectures is the ability to carry broadcast television signals on a separate wavelength, creating what some call "CATV overlay" and others call "linear RF over fiber" delivery of multi-channel TV. This PON attribute can create a planner's dilemma when it is related to architectures for multicasting TV at the IP layer. Some operators believe that the question of CATV-overlay versus multicast is the most significant question in carrier television infrastructure planning today.
Video delivery today can be categorized as broadcast or video on-demand (VoD), where the former is sent to all qualified customers at the same time on a schedule, and the latter is selected on a per-customer basis when needed. Traditional television is broadcast-based and both cable and satellite TV viewing is dominated by broadcast channels. PON systems that reach the home can deliver broadcast channels outside the data pathway, in the same way that cable systems deliver channels today. In contrast, FTTN systems must either use some multicast IP mechanism to deliver broadcast channel programming (Microsoft TV is an example) or must rely on a parallel broadcast delivery system like satellite. Both Verizon and AT&T in the U.S. offer hybrid satellite/DSL services where VoD is delivered as DSL data and broadcast channels are received through a relationship between the carrier and a satellite TV company.
Deciding what approach is best for a given geography requires the careful balancing of a number of factors. There is no question that FTTH deployment is less likely to be rendered obsolete than any FTTN approach, but it is also more costly up front. Today's estimates are that FTTH will cost approximately four times as much as FTTN in "pass cost," but it may pay most of that back within 10 years on outside plant maintenance costs.
PON technologies support distribution of broadband services over a considerable distance, far more than could be supported using a combination of FTTN and DSL, and the advantage grows as the speed of the connection increases. DSL at 20 Mbps or so can be delivered to 10-to-15 kilofeet depending on the condition of the loop, but at 50 Mbps, most operators would try to keep loop length to 2 kilofeet or less, and many to 1 kilofoot.
As the loop length for DSL shortens, the number of FTTN remotes needed to support a given population of users increases, and the cost advantage of PON grows larger. Interestingly, where population densities are very high and many users can be reached with 50 Mbps VDSL technology at acceptable loop lengths, the difference in cost between running PON and FTTN may also be lower because the PON tree branches are short, and so PON may be more economical for both very thin and very thick populations of users, making alternatives to PON attractive in a middle zone.
PON deployment depends on needed Internet access speeds
Because of the effect of access speed on PON's benefits, an economic assessment of the total opportunity is essential in planning PON deployment, and the primary question facing planners is whether the future is likely to demand Internet access speeds of 50 Mbps or more. If network operators expect to offer Internet services, or a combination of Internet and IP video services, whose combined bandwidth exceeds 50 Mbps, it is very unlikely that anything but fiber/PON will be suitable.
The requirement for high-speed broadband could be generated by aggressive multicast IPTV plans, expected competition from cable operators that convert to DOCSIS 3.0, migration to HDTV, and increased consumer demand for Internet bandwidth—and in any combination. In areas of high demand density (suburban/urban areas with above-average household income levels), it is unlikely that many operators will be able to avoid FTTH and PON in the next decade, and so would likely benefit from at least selective deployment of PON in the near term.
Fiber is an asset that doesn't become obsolete, and PON technology shows every sign of offering those who deploy it a steady increase in available per-customer capacity not only for residential services but for many business sites as well. The question operators should likely answer is not "Why PON?" but "Why not?"
About the Author: Tom Nolle is president of CIMI Corporation, a strategic consulting firm specializing in telecommunications and data communications since 1982. He is a member of the IEEE, ACM, Telemanagement Forum, and the IPsphere Forum, and the publisher of Netwatcher, a journal in advanced telecommunications strategy issues. Tom is actively involved in LAN, MAN and WAN issues for both enterprises and service providers and also provides technical consultation to equipment vendors on standards, markets and emerging technologies. Check out his SearchTelecom networking blog Uncommon Wisdom.