PON
PON
PON
PON
PON

PON Today & Tomorrow

Dec. 21, 2016
Next-Generation PON Architectures and Their Impact on the OSP by: Scott T. Wilkinson (This article originally ran in the December 2009 issue of OSP Magazine) Passive Optical Networks (PONs) have […]

Next-Generation PON Architectures and Their Impact on the OSP

(This article originally ran in the December 2009 issue of OSP Magazine)

Passive Optical Networks (PONs) have become arguably the most popular outside plant architecture for delivering Triple Play services of high-speed data, video, and voice. A passive outside plant (OSP) has no active components, which eliminates the need for powered cabinets, batteries, and real estate and has the potential to reduce operating expenses. As a purely passive OSP, a PON has nearly unlimited potential for future bandwidth expansion. PONs have been deployed by some of the largest telecommunications companies in the world (Verizon, AT&T, and NTT, among others) as well as some of the smallest independent operators, municipal networks, and utilities in the U.S.

The first modern PON networks deployed in North America, starting in the early 2000s, were based on the ITU standard known as Broadband PON (BPON). Newer deployments are typically based on either the ITU standard known as Gigabit PON (GPON) or the IEEE standard known as Ethernet PON (EPON). In general, all 3 standards allow for a passive split ratio of 1:16 or higher, with 1:32 being the most popular ratio deployed. The maximum distance from the head end to the subscribers is generally 20 km, although other options are possible. To eliminate the need to modify a deployed OSP, the standards bodies that are working on Next-Generation PON (NG PON) have made it a priority to ensure that NG PON technologies will work on a deployed OSP with a 1:32 split ratio and spans up to 20 km long.

Both the IEEE and the ITU (via recommendations of the FSAN working committee) have defined NG PON as a 10Gbps downstream, single wavelength PON. The upstream speeds differ somewhat between the standards, with IEEE defining rates of 1 Gbps and 10 Gbps and the ITU adding an additional 2.5 Gbps upstream option.

The IEEE has already defined the wavelengths to be used in NG PON, and the ITU has a goal of matching at least one of the designated IEEE wavelengths (a goal that is still under debate). Both standards bodies intend to define span budgets that will allow NG PON to operate on an OSP designed today, based on the current PON standards.

PON and Attenuation

In the OSP, there are many sources of signal attenuation. In a PON, the largest source of loss is the splitter. In an ideal splitter, the loss is 3 dB for each power of 2 (1:16 = 12 dB, 1:32 = 15 dB). However, in a real deployment the loss is never ideal. Typical loss for a 1:32 splitter used in OSP design, including margin for end of life and temperature variations, is 17.5 dB.

Other sources of loss are connectors, splices, fiber loss, and loss due to WDM couplers. While each design uses nominal values in designing the outside plant, the values in the tables shown are typical. Note that the fiber loss is dependent on the wavelength of the signal, where 1310 nm is the wavelength used for upstream transmission and 1490 nm is the wavelength used for downstream transmission. The losses shown in Figure 2 and Figure 3 assume standard single mode fiber (SMF), which is the predominant fiber deployed in metro and access networks in the U.S.

Figure 2 details the 3 span budgets for 10G-EPON as defined by the IEEE. These 3 spans are referred to as PRn and PRXn, where n=10, 20, or 30. The PR span budgets are defined for a 10 Gbps bidirectional (upstream and downstream). The PRX span budgets are defined for 10 Gbps downstream / 1 Gbps upstream asymmetrical. PR(X)-30 budgets are designed to operate over a span of at least 20 km. The PR(X)-20 and -10 are designed for shorter distances.

The ITU and the FSAN working group have not yet defined span budgets for 10GPON. However, the standards bodies have agreed that 10GPON must be able to operate on an OSP designed for GPON. The GPON Class B+ span budget is the most widely deployed, and has a downstream span budget of 28 dB (29.5 dB if deployed by Verizon).

The standards bodies working on NG PON have both used new wavelength plans for the new standards to allow current and NG PON to coexist on the same OSP. For example, the IEEE has defined downstream wavelengths of 1577 nm and 1590 nm (a recent straw poll recommended eliminating the 1590 nm option). Upstream, the IEEE NG PON standard defines 1270 nm for the 10 Gbps rate and reuses the 1310 nm wavelength from EPON for the 1.2 Gbps rate. The ITU has yet to settle on a wavelength plan, but among the options are schemes to match the IEEE wavelength plan or to define upstream and downstream wavelengths in the low-loss C-band near 1550 nm.

Using the numbers in Figure 3, an OSP with a 1:32 splitter, 20 km spans, and standard single mode fiber has a total attenuation of 20 dB + 20 km x (loss/km @ wavelength). At the EPON/GPON wavelengths of 1490 nm and 1310 nm, that corresponds to a total attenuation
of 27 dB upstream and 24 dB downstream, within the 28 dB span budget defined for Class B+ GPON optics. The question is whether or not NG PON will be able to run on the same infrastructure.

The NG PON proposals being standardized in both the IEEE and ITU use slightly different wavelengths than the current standards, so the impact of fiber on the loss will be minimal. In the upstream direction, the move from 1310 nm to 1270 nm is negligible over the 20 km PON distance. In the downstream direction, the move from 1410 nm to wavelengths around 1590 nm will actually improve the loss budget. The ITU standards are being designed to include a 28 dB or higher option to allow operation on the existing OSPs. The IEEE PR(X)-30 standard already exceeds the 28 dB span. Therefore, loss in the fiber should not be a barrier to an upgrade to NG PON. If fiber loss is not a barrier to a NG PON upgrade, what will be the impact of other changes, like the change in wavelengths?

New and Old Coexisting Together?

Both standards bodies have chosen new wavelengths for NG PON with the intention of allowing NG PON and current PON to co-exist on the same OSP. Theoretically, this would allow an operator to upgrade a few customers on an OSP to NG PON while leaving others on the older generation PON. Whether or not this type of deployment makes sense is a point of contention among service providers. Some providers claim that they would not want to deploy two sets of OLTs on the same PON at the same time. However, should a customer want to deploy both technologies at the same time, the standards are designed so that co-existence is possible.

A potential barrier to deploying NG PON and current PON on the same OSP is the existence or non-existence of wavelength filters at the ONTs. Most modern GPON ONTs have an integrated filter that would eliminate interference from NG PON wavelengths. However, most EPON ONTs, older GPON ONTs, and BPON ONTs do not have these filters installed and will not be able to co-exist with NG PON. Service providers with these ONTs deployed will have to either change the OSP and install filters at the older ONT locations or abandon the idea of current PON and NG PON co-existence.

One part of a PON deployment that may be affected by an upgrade to NG-PON is the RF overlay, if any. Because 10G PON requires higher transmit powers due to the higher speeds, the necessary higher power data signals will interfere with the RF overlay signal. The GPON standard has integrated mitigation techniques to reduce the interference from the data channel onto the RF overlay, but those techniques will not be sufficient at the higher powers required in NG PON.

One way of mitigating the effect is to separate the signals physically by running a separate fiber from the head end to the splitter and recombining signals at the split. This upgrade will require a change in the OSP (a new fiber to the splitter, an upgrade to the splitter).

However, the best way to avoid RF signal interference is to change from RF video to IP video when making the move from GPON or EPON to NG-PON. A wholesale move to IP video has been predicted by analysts since the advent of GPON, but has been limited by cost, complexity, and embedded base. Perhaps the enormous bandwidth offered by NG PON, coupled with the increasing maturity of IP video solutions and the difficulties of providing RF video with NG PON will finally provide the impetus for an industry-wide move to IP video.

Looking to the Future

PON based on wavelength division multiplexing (WDM) is an alternative form of NG-PON in which each end device is assigned a different color of light. WDM-PON is not currently supported in a PON standards body, but there are proprietary solutions available from several PON vendors.

WDM-PON will have difficulty being deployed on a current OSP without significant changes to OSP components. Upgrading from a GPON or EPON deployment to a WDM-PON deployment will, at a minimum, require changes in the field splitters. Fortunately, a WDM splitter has less loss per wavelength than a traditional power splitter, so more span budget is potentially available on a WDM-PON and the installed 20 km radius fiber infrastructure will not need to be changed. The overall impact of an upgrade to WDM-PON will not be trivial, but should not be on the order of the copper-to-fiber migrations ongoing today.

As mentioned before, perhaps the most important reason to deploy a passive fiber infrastructure today is that it will support many iterations of technology evolutions without major OSP changes.

There are some other non-standard interim PON strategies being proposed by PON vendors outside of the NG PON standards activities, including a hybrid GPON / WDM-PON approach. As with WDM-PON, those technologies will work over the deployed fiber span, but will probably require a change to the splitters and deployment of wavelength filters at the ONTs. Each of these interim technologies must be evaluated with respect to the operator’s deployed OSP, but the odds are very small that a change in the fiber will be required.

In short, operators should be assured that an OSP built today will be ready for an upgrade to NG PON in the future without major changes. An OSP built today to meet the requirements of a standards-based OSP (20 km distance, 1:32 or 1:64 splits, single mode fiber) will certainly be able to support the upcoming NG PON standards without changes to the installed fiber. In many cases, the other passive components (splitters, couplers, etc.) will not need to be changed either. The electronics at both ends will, of course, need to be upgraded to the higher speed, but a major infrastructure upgrade of the passive OSP should not be necessary for the foreseeable future.

About the Author

ISE Staff