The Passive Optical Network (PON) has become the most widely deployed FTTH technology worldwide, and is evolving to meet customer video bandwidth demands. From 622/155 Megabit per second (Mbps) B-PON and 1 Gigabit per second (Gbps) Ethernet PONs, most carriers have upgraded to deploying 2.5 /1.25 Gbps G-PON systems beginning in 2007.

Carriers have a shared vision that the next jump in PON speeds will be to 10 Gbps, and in response IEEE and ITU/FSAN are creating 10 Gbps PON standards that will be finalized in 2010. In addition MSOs are beginning to get serious about deploying FTTH, with the SCTE writing its first FTTH standard known as RF over Glass or RFoG.

The optical fiber distribution network (ODN) installed today will need to support many upgrades to higher data rates during its expected 25-year or longer lifetime given the expected continued 40-50% annual growth in video driven bandwidth demand.1 (See Figure 1.) How can ODNs being built now be optimized for both today’s PON systems and a likely future upgrade to 10 Gigabit speeds? The 10 Gigabit PON is here: is your network ready?

Figure 1. Optical Distribution Network (ODN). The optical path between the CO/HE and the subscriber.

Improved ODNs Can Support Lower Cost Architectures
While the IEEE 10GE-PON standard supports 20 KM distances with a 1:32 split ratio, some carriers see value from increasing the split ratio and/or extending the maximum PON reach. Longer reach support can help reduce the number of central offices, resulting in lower costs for facilities and labor. Higher split ratios can increase the number of subscribers supported by each Optical Line Terminal (OLT), thus reducing OLT costs. The IEEE draft 10GE-PON standard allows for compliant systems to exceed the maximum distances stated in the document, and/or to support greater split ratios than specified. Using components with lower optical loss throughout the ODN can extend the reach and/or enable increased split ratios, as shown in an example for 10GE-PON in Table 1. Since 10G-PON will likely utilize similar PMDs to the IEEE 10GE-PON, similar improvements in 10G-PON performance should be possible using lower loss ODN components.

Table 1. 10GE-PON performance comparison.
Notes: General: penalty and loss values calculated using IEEE 802.3av Link Model v2.3.
a. 1577 nm wavelength is +3 nm /-2 nm, 1270 nm is +/-10 nm, 1310 nm is +/-50 nm.
b. Fiber attenuation values for the standard loss network are those assumed in the IEEE Link Model.
c. Dispersion penalty assumes expected use of DFB lasers. SRS penalty assumed to be zero for O-band. Wavelengths due to lack of applications having high power wavelengths in or near the O-band.

From the example in Table 1 we see that significant savings are possible through the use of a low loss ODN. The over 3 dB of additional optical power margin from the low loss ODN can be used to increase the split ratio from 1:32 to 1:64 in this example, resulting in a $75 savings per home passed by halving the number of required OLT. Or the lower loss ODN could enable a split ratio increase from 1:16 to 1:32 for distances between 20 and 30 KM, saving $150 per home passed.2 While loss is the most important and manageable constraint for ODN reach and split ratio, there are also other optical impediments that the standards writers have analyzed to determine their effects on optical system performance.

Fitting Two PONs Onto One ODN
Co-Existence Consumes Wavelengths: Are We Running Out?
Carriers want the flexibility to run 10 Gigabit and lower PON speeds on the same fiber, which is resulting in the specification of three new PON wavelengths: 1577 nm, 1610 nm, and 1270 nm. As a result PON ODNs will need to support the configuration shown in Figure 2.

Figure 2. Each band represents the center wavelength and wavelength tolerance for a single transmitter.
Notes: Each band represents the center wavelength and wavelength tolerance for a single transmitter
• u - upstream from subscriber (ONU) to network (OLT)
• d - downstream from subscriber (ONU) to network (OLT)
• 10 Gigabit PON denotes 10GE-PON and 10G-PON

Optical Penalties Discussion
Optical Loss
The optical signal loss (technically known as channel insertion loss caused by the fiber, splitters, connections, and splices has the greatest impact on the supportable reach and split ratio achievable on most PONs. For every dB of loss eliminated in the ODN, the reach can typically be extended by about 3 KM at 1310 nm, and about 4 KM at 1550 nm. A 3 dB reduction in loss can enable a doubling of the split ratio, lowering the cost and space required for CO electronics.

Lower loss can liberate power that can be allocated for additional connections to enable lower labor plug and play MDU systems. Lower loss is potentially valuable at any wavelength since the wavelengths that might be utilized over the lifetime of the ODN could range from 1260 nm to 1620 nm based on the ITU-T 694.2 specified CWDM grid. Bend Insensitive ITU-T G.657 fiber for the first time is specified in a PON standard, IEEE 802.3av 10G-EPON, based on concerns about the bending loss that may occur with the bend sensitive 1577 nm wavelength used for 10 Gbps downstream services.

Other Optical Penalties
Other optical penalties are managed by improved electronics and lasers on modern PON systems such that PON reach is primarily loss limited. Stimulated Brillouin Scattering (SBS) is easily electronically managed on standard G.652D fibers (such as Zero Water Peak fiber), using an existing “free” feature on 10 Gigabit transmission chips, so that no special high power fibers are needed. Even high power (20 dBm) RF analog video transmitted over long fiber lengths at 1555 nm has no SBS issues with quality G.652D or G.657A1 fibers based on the transmitter wavelength dithering employed as a standard feature by vendors of transmission equipment.

Stimulated Raman Scattering (SRS) is a very small penalty that might affect longer wavelengths, and need not be a consideration for ODN design if high quality and IEEE/ITU compliant passive components are utilized.

Multipath Interference (MPI) should not create any detrimental effect in PON systems using ITU G.652 or G.657 fibers that meet 1260 nm cable cutoff as measured by the test procedure in TIA/EIA 455-80.

Chromatic dispersion (CD) can cause spreading of the laser pulse over the length of the fiber. Fortunately, modern lasers used with PON systems have narrow spectral widths, resulting in chromatic dispersion penalties of well under 1 dB even for 20 to 30 KM PON lengths. For systems designed to support very long reach PONs, a more tightly specified narrow spectral width laser can enable support to 60 KM.

In summary, minimizing optical loss across the full spectrum enables more flexibility to lower system cost and/or install ODNs faster and easier using plug-and-play MDU systems.

Preparing the ODN for Cost-Effective 10G, RFoG, and Beyond
Full Spectrum Lower Loss ODN Provides Tangible Benefits
An optical distribution network (ODN) being installed today for GE-PON, RFoG, or G-PON will likely need to support a 10 Gigabit PON and later a next-generation PON over its lifetime. Minimizing optical loss across the full spectrum (1260 nm to 1620 nm) of wavelengths offers the greatest opportunity to those specifying ODN architectures and components to improve the reach, split ratio, and multispeed upgradability of a PON. Optical loss can be minimized across the full spectrum (1260 nm to 1620 nm) of wavelengths by utilizing zero water peak fiber, lower loss splitters, lower loss connectors, and bend optimized fiber.

Keeping Small Bends from Causing Big Trouble
Tighter fiber bends with longer more bend sensitive wavelengths is putting a double squeeze on the ODN. Bending loss increases with wavelength in optical fibers. The long wavelengths (1490 nm -- 1550 nm) used in current PON systems create enough of a challenge to service providers contending with the tight bends which can happen in the last mile. Even tighter bends are needed to support fast and easy to install indoor drop cables. Now we have three new standards that will likely be widely deployed which use even longer wavelengths that will further increase bending loss, as shown in Table 2.

Table 2. Bending Loss Increase from current to fast emerging next-generation PONs

Benefits of Bend Optimized Fiber
Bend Optimized fiber provides performance optimized to benefit the application. There are two types of bend optimized fiber designed to benefit the two levels of bending performance needed in the ODN. The first, generically known as zero water peak bend insensitive fiber (ZWP BIF), is optimized to support most bending challenges. The second, known as Resonance Assisted Fiber (RAF) is an Ultra Bend Insensitive fiber (UBIF) optimized for the bendable optical drop cable needed for fast and easy indoor installations.

Most Segments of the ODN
The ODN for PON networks can be bend-challenged given the many points of fiber management throughout the Last Mile. A view of some of the challenges is shown in Figure 3.

Figure 3. Bending Challenges in most segments of Last Mile ODN.

Most segments of the ODN are optimally served by a ZWP BIF which offers easy low loss splicing and low loss connections between itself and other fiber types. Most segments of the ODN have sufficient bend radius management such that bending radius rarely drops below 30 mm, but on occasion can accidentally drop down to about 10 mm (0.4") in CO/Head End patch panels, Fiber Distribution Hub (FDH) splitter cabinets, outside plant (OSP) cable fibers in closure splice trays, and MDU building backbones.

Standard single-mode G.652D fiber has excessive bending loss at the longer wavelengths that are used in today's PONs, and to a greater degree in the fast emerging next-generation PONs. For one turn at 10 mm radius, the bending loss shown in Table 3 would result. Under this condition the G.652 std SMF produces high bending loss, G.657A1 ITU compliant fiber exhibits high loss for the next-generation applications, while the ZWP BIF (improved G.657A1) fiber exhibits bending loss of only 0.1-0.4 dB depending on the wavelength.

Table 3. Bending loss comparison relevant for most bend-challenged applications.

The Optical Drop Cable Inside the MDU and Home
There is a growing trend toward installing the ONT inside each subscribers MDU residence, and even inside single family (SFU) residences. Verizon and others see value in using a more compact, non-hardened indoor ONT which can more easily attach to other indoor wiring (COAX, computer cables), and connect to power supplies.3  In addition, deploying indoor ONTs to subscribers on a success basis can lower the cost per unit passed compared to a more expensive multi-unit ONU than must be deployed upon the initial service offering.4  But there are challenges to running optical fiber drop cables quickly and easily to and inside apartment units and inside homes, as can be seen in the photos in Figure 4.

Figure 4. Indoor Optical Drop Challenges: MDU and SFU.

Ultra-bend insensitive fiber (U-BIF) has been developed for this special application to enable the fast and easy in-residence drop cable installation. A new U-BIF, Resonance Assisted Fiber (RAF) provides 1550 nm bending loss < 0.1 dB for a 360 degree turn at 5 mm fiber radius. RAF is constructed of solid glass, and enables fusion splicing to standard and conventional bend insensitive fibers using existing equipment, connector mounting using standard procedures, and has uniform optical properties along its length. RAF cables can be stapled and routed around corners with negligible signal loss. The industry is moving toward creating a new standard addressing this application which requires a tight 5 mm optical fiber bend radius.

It is clear from Table 4 that the bending loss performance of fibers compliant to the existing standards for conventional BIF can produce very high bending losses which might severely reduce the reach of a PON or shut down service.

Table 4. Bending loss comparison for Indoor Optical Drop Application (Benchmark 5 mm radius).

The single 360-degree turn at 5 mm fiber radius bending loss performance has become a benchmark in the industry with respect to MDU and in residence drop cable bending loss performance. However, 20 or more 90-degree corner bends with many staples are possible in some installations.

This drove Verizon to create an MDU Simulation Test, described in TPR 9424, which is intended to emulate the potential bending of MDU drop cables installed with fast and easy techniques, without the expense or labor of bend radius management hardware or conduit. Figure 5 shows that in this simulation test, G.657A2 compliant optical fiber can add 4 dB of bending loss in the test shown at 1550 nm. At 1577 nm in the same test the loss would likely be 5-6 dB, while 1610 nm would result in >8 dB. The new RAF U-BIF by contrast shows only 0.16 dB in the test at 1550 nm, and under the same condition at 1577 and 1610 nm would be expected to exhibit 0.24 dB and 0.32 dB of bending loss respectively.

Figure 5. MDU Simulation (Verizon TPR-9424) Comparison at 1550 nm -- Relevant for Indoor Drop Application.

Recent experiments with modern synthetic silica optical fibers such as RAF show that if every drop cable installation bent and stressed the 4.8 mm cable as described in the above MDU simulation, the mechanical failure rate would be less than 1 in 5 million drop cables per year.5

E-Band: For the Next NG-PON?
What will be next after 10 Gigabit PONs? The wavelength range of 1360-1460 nm, known as the E-Band, has not yet been utilized by FTTH standards, but is used today for CWDM applications in metropolitan area networks. The E-Band could be the next frontier for NG-PONs, making Low Water Peak and Zero Water Peak (12% lower maximum E-Band loss than LWP) valued assets in PON ODNs.

Conclusion
The ODNs being installed today will likely need to support higher speed and longer wavelength applications being standardized today. Building a bend optimized, low loss, full spectrum ODN can enable significant benefits:

- Fewer CO/Head-end locations through extended reach support.

- Lower OLT/Head-end equipment cost by increasing homes passed per port through greater split ratios, saving $75 to $150 in OLT cost per home in the example shown.

- Faster MDU and in home optical drop installations through greater use of plug and play MDU systems by having additional loss budget to accommodate additional connectors.

- Smaller space required for fiber management and smaller diameter cables.

- Easier and lower cost upgrades to next-generation PONs using longer, more bend-sensitive wavelengths and wavelengths in the 1360-1460 nm band designated by the ITU "for future use".

The creation of a low loss full spectrum ODN is possible using solutions available today. Loss can be minimized across the full spectrum (1260 nm to 1620 nm) of wavelengths by utilizing bend-optimized fiber, lower loss splitters, lower loss connectors, and in the OSP cable zero water peak fiber. Using ZWP Bend Insensitive Fiber in the ODN up to the drop demarcation, and RAF Ultra Bend Insensitive Fiber for the drop into the residence can greatly reduce macro-bending loss to enable reliable services at longer reach with greater split ratios.

In summary, the solution exists today for building cost-optimized scalable ODNs for today's GPON, RFoG, and GE-PON applications, and emerging future applications to 10 Gigabit PONs and beyond.

Endnotes
1. "FTTH Design with the Future in Mind", John George, FTTH Conference 2005.

2. Assumes a $4,800 cost per OLT port and would vary proportionally with the cost per OLT port.

3. “Verizon Begins Deploying Breakthrough Optical Technology That Saves Space and Makes It Easier to Install FiOS Services in
Apartment Buildings”, New, Smaller Optical Network Terminals Support Company's Fiber-to-the-Home Initiative - http://newscenter.verizon.com/press-releases/verizon/2009/verizon-begins...

4. "The Trend Toward Desktop ONTs", Bhavani Rao, FTTH Conference 2009.

5. "Reliability Considerations for Next-Generation Bend Optimized Fibers", D. Mazzarese, et al., Proceedings of the 57th International Wire and Cable Symposium (11/2008).

About the Author
John George is Director, Systems and Applications Engineering for OFS. John has served with AT&T, Lucent Technologies, and OFS for 25 years, and directs the OFS systems and applications engineering group. He has published and presented more than 35 papers on fiber optics and FTTH. John has been an active member of the FTTH Council since its inception in 2001. With the heritage of Lucent Technologies and Bell Labs, and strong backing of Furukawa Electric, OFS provides innovative optical fiber based solutions to communications and specialty customers worldwide. For more information, email johngeorge@ofsoptics.com or
visit www.ofsoptics.com

What’s your take on this subject? Leave a comment and get the conversation going.