Meeting metro capacity with cost-reducing fibre infrastructure

Sept. 1, 2003

Though difficult to quantify, signs indicate that some level of infrastructure growth is likely the latter half of this year and into next year. Networks are running at historically high capacity utilisation levels.1 Worldwide Internet traffic, which is continuing to double every year, is forecast to reach 5,175 petabits per day in 2007 from 180 petabits per day last year.2 In addition, the Federal Communications Commission's initial rulings in February freeing the incumbent local-exchange carriers from unbundling fibre to the home (FTTH) loops and the Defense Information Systems Agency-led GIG BE project point to increased telecommunications spending in the access and government sectors.

Metro and access infrastructure represents the fastest-growing segment and the most complex in terms of the variety of protocols and architectures used. WDM is emerging as a powerful technique for the metro segment, given its cost-competitiveness for broadband IP (Fast Ethernet, Gigabit Ethernet) and other data applications (Fibre Channel, DV6000). WDM-based networks are bit-rate- and protocol-independent, so these infrastructures can carry various types of traffic at different speeds concurrently—a crucial requirement for metro systems. The technology is also synergistic with next-generation SONET/SDH and resilient packet ring.

Both CWDM and DWDM technology have their place in current and emerging metro-network infrastructure. When these technologies are used in combination with appropriate optical fibres, the economic benefits, which help to lower system costs, are significant. Metro application-specific fibres such as zero water peak fibre (ZWPF) and nonzero dispersion-shifted fibre (NZDSF) can enhance the benefits of WDM. Capital expenditures by service providers remain very weak, however, so any network-capacity increases will face scrutiny, placing even greater emphasis on the lowest possible system cost.

Initially, DWDM solutions were considered for the metro market because this technology was already available in long-haul (LH). Since LH required the use of erbium-doped fibre amplifiers (EDFAs) to overcome loss, the objective was to squeeze as many channels as possible into the amplified portions of the EDFA spectrum. That necessitates precise optical multiplexing/demultiplexing filters to provide channel spacings of 200 GHz or less and very well controlled lasers with wavelengths maintained at a constant temperature to prohibit drift outside the given channel. The added expense of this precise wavelength control is prohibitive for short metro distances.

In the metro, optical amplifiers are not required for short distances, so WDM channels are not limited to a single spectral band. Such freedom of bandwidth use allows for widely spaced channels, which are found in CWDM systems with 20-nm channel spacing, based on ITU-T Recommendation G.694.2. Widely spaced channels help eliminate the added costs associated with wavelength control. When CWDM is coupled with ZWPF, channel count and distance are not sacrificed either, as evidenced by the demonstration of a full 16-channel CWDM system over 75 km of ZWPF without amplification.3

The cost-effectiveness of CWDM is directly related to channel spacing. With DWDM systems, narrow channel spacing requires more expensive components. Since laser wavelengths drift in proportion to temperature increases, DWDM systems need to employ cooled lasers to keep the wavelength within the narrow optical windows recognised by the multiplexer/demultiplexer filters. Cost savings between 30% and 65% have been documented with CWDM.4

In addition to the transmission equipment, it is important to consider the optical fibre's impact on system performance. It is less costly to deploy fibre today that can also support future demand. Planning costs associated with obtaining the permission to dig and install cables are significant in the metro environment and dense population areas, driving up the outlay for network builds. Most analysis indicates that the cabled optical fibre only represents a small percentage (about 2–5%) of the total system cost once lit.

Fibre in the local and access domains of MANs must be extremely versatile. A single fibre type will link central offices that are tens of kilometers apart and connect businesses that may be within 100 m of a central office. Some of the key requirements for metro fibre include:

  • Supporting legacy 1310-nm SONET/SDH systems as well as low-cost Ethernet transport from 10 Mbits/sec to 10 Gbits/sec.
  • Supporting 1550-nm DWDM systems.
  • Supporting 16-channel CWDM—conventional singlemode fibre (SMF) typically carries eight channels in the S-, C- and L-bands.
  • Accommodating a wide variety of access solutions such as the FTTH-passive-optical-network enhancement band.5
ZWPF is emerging as a standard because it meets all these criteria at a competitive price point versus conventional SMF. From a technical perspective, the most challenging of these criteria is the requirement to support 16-channel CWDM systems developed to use the entire optical spectrum. Higher loss of the shortest wavelength channels is one reason why full-spectrum CWDM systems typically use the 16 channels from 1310 to 1610 nm. Full-spectrum CWDM is difficult to use with conventional SMF, where the high-loss peak associated with the hydroxyl ion (OH-), which is from water incorporated during manufacturing, virtually eliminates the 1400-nm region for use in most metro applications (see Figure 1).

Most important, the ion has demonstrated a tendency to creep back into conventional SMFs, which renders conventional SMFs selected for low water peak loss very risky for use with full-spectrum CWDM. ZWPFs, also referred to as ITU G.652.C/D fibres, meet International Electrotechnical Commission standards for H2-aged water peak loss performance, ensuring the means employed to eliminate the water peak also prevents its return over the lifetime of the fibre.

If capacity beyond 16 channels is needed, a DWDM upgrade in place of one or two C-band CWDM channels is an attractive option. These types of hybrid platforms are now commercially available from leading system houses. It is important to note that G.652.C/D fibres provide the same flexibility but at a much lower cost threshold—about 33% less on a lifecycle cost basis—than conventional SMFs when upgrading from CWDM to DWDM.

While ZWPF combined with full-spectrum CWDM enables the lowest-cost solution for short-distance metro access and edge applications, other application-specific fibres offer additional savings when combined with the greater traffic needs of the metro backbone. Most metro networks today run at data rates up to 2.5 Gbits/sec (OC-48); however, service providers anticipate the need for migration to 10 Gbits/sec (OC-192). Many carriers expect DWDM infrastructure to meet this increased capacity need at longer distances.

Typically, only a few DWDM channels are lit at startup; therefore, system costs are not only driven by end terminals, but also by the dispersion compensation modules (DCMs) and amplifiers required to meet the longer distances in the metro backbone. Although ZWPFs will perform better than conventional SMFs due to improved attenuation and geometric properties, the relatively high dispersion (~17 psec/nm·kmat 1550 nm) still limits transmission distance without regeneration for lower channel counts or dispersion compensation to accommodate larger channel counts. Either regeneration or dispersion compensation adds considerable cost. In addition, the use of dispersion compensation adds loss, which typically results in greater EDFA complexity and system polarisation-mode dispersion.

The lower-cost metro backbone fibre is the same type deployed in LH networks, NZDSF, particularly those fibres with low dispersion and low dispersion slope. NZDSF offers a dispersion level three to four times lower than that of conventional SMF, thereby allowing a three- to four-fold increase in the uncompensated reach at 2.5 and 10 Gbits/sec—a result verified with commercial systems from vendors such as Cisco Systems, Lucent Technologies, and ADVA Optical Networking.

The moderate dispersion characteristic of NZDSF provides not only longer reach with existing equipment, but also cost savings for metro and regional backbone distances (see Figure 2). In fact, Farr Farhan, chief development officer at Movaz Networks, reports Movaz has found in its deployments that a DWDM network deployed over OFS's NZDSF can lower systems costs by well over 20% at startup versus conventional SMF for regional metro backbone networks. The absence of dispersion compensation allows for simpler and cost-reduced amplifiers, which also reduces power and space requirements.

With these cost reductions, it is always important to remember that compatibility with existing networks and applications is a precursor to new fibre deployment. In the metro space, service providers strongly desire the flexibility of using the 1310-nm band. NZDSF has been demonstrated to transmit 10 Gbits/sec at 1310 nm for 70 km at a bit-error rate of 10-11 without forward error correction.6

CWDM is no longer a niche technology. It is emerging as a robust option for current and future metro/access bandwidth needs. Its largest benefit is when used over G.652.C ZWPF, which enables the use of the full optical spectrum, doubling the wavelength channel capacity from eight to 16.

DWDM will also play an increasing role in the metro backbone and regional applications. With the use of lower-dispersion, low-dispersion-slope NZDSF, DWDM is more economical and versatile, especially as bit rates migrate to 10 Gbits/sec and beyond.

Both ZWPF and NZDSF can be incorporated into a single cable for greater flexibility, as advocated in hybrid-cable designs.7 That gives the best-in-class metro performance—covering short access spans and laterals to regional networks several-hundred kilometers long interconnecting buildings to points of presence as well as simplifying design rules even for high-bit-rate traffic. Alternately, mixed fibre plant performance has also been characterised and shown to be a viable means of network architecting.8 Either way, the benefits of both fibres for metro networks are here to stay.

Dr. Ray Boncek is systems engineering research manager, Dr. Paul Dickinson is customer systems engineering manager, and Dr. Santanu Das is metro systems engineering director at OFS (Norcross, GA). They can be reached via the company's Website, www.ofsoptics.com.

  1. Telecommunications Industry Association, "TIA White Paper: Fiber Optic Network Capacity and Utilization, Part II," September 2002.
  2. D. Legard, "Internet traffic to double each year, IDC predicts main driver will be broadband," March 6, 2003.
  3. S.K. Das et al., "40 Gb/s (16×2.5 Gb/s) Full Spectrum Coarse WDM Transmission over 75 km Low Water Peak Fiber for Low-Cost Metro and Cable TV Applications," NFOEC 2002 technical proceedings.
  4. K. Kincade, "CWDM breathes life into metro, access, and enterprise applications," Laser Focus World, March 2003.
  5. B.R. Eichenbaum et al., "Economics of coarse WDM compared with dense WDM for wavelength-addressable PON access architectures," NFOEC 2002 technical proceedings.
  6. H. Thile et al., "Metro express transparency in excess of 250 km using dispersion optimized fiber," NFOEC 2002 technical proceedings.
  7. S.K. Das et al., "Building metro fiber networks that last," Lightwave, September 2001.
  8. H. Thile et al., "Economics and performance of 10-Gb/s metro transport over mixed fiber plant of G.655 NZDF and G652.C zero water peak fibers," 2003.