Dispersion-Shifted Fiber Applications in Long-Haul

Background: Why Chromatic Dispersion Matters

Chromatic dispersion is the phenomenon where different wavelengths of light travel through a fiber at slightly different speeds. A modulated optical signal contains a small spread of wavelengths around the carrier frequency. Over distance, those wavelengths separate in time, causing the originally sharp pulses to spread out and overlap with adjacent pulses. The result is intersymbol interference, which limits the maximum data rate and distance of a fiber link.

Standard G.652 singlemode fiber has zero chromatic dispersion at approximately 1310 nm (the original operating wavelength of singlemode fiber). At 1550 nm, where attenuation is lowest (about 0.20 dB/km vs 0.35 dB/km at 1310 nm), the chromatic dispersion is approximately 17 ps/nm/km. This dispersion limits 1550 nm transmission distance to about 80 km at 10 Gbps before the pulse spreading becomes unmanageable.

For long-haul telecom in the 1980s and 1990s, this was a problem. The 1550 nm window had the lowest attenuation (longest distance per amplifier) but the highest chromatic dispersion (shortest distance per pulse spreading). Engineers wanted both: lowest attenuation and lowest dispersion at the same wavelength. Dispersion-shifted fiber was the answer.

How Dispersion-Shifted Fiber Works

The chromatic dispersion of a singlemode fiber is the sum of two components: material dispersion (the natural variation of glass refractive index with wavelength) and waveguide dispersion (the variation caused by the fiber's geometric design). In standard G.652 fiber, these two components combine to give zero dispersion at 1310 nm.

By modifying the fiber's refractive index profile (specifically, adding a triangular or stepped index profile in the core), engineers can shift the waveguide dispersion contribution to cancel material dispersion at a different wavelength. Dispersion-shifted fiber (G.653) is engineered to have zero chromatic dispersion at 1550 nm instead of 1310 nm.

The result: a 1550 nm signal in DSF experiences both the lowest attenuation (0.20 dB/km) and zero chromatic dispersion. A 10 Gbps single-wavelength transmission can run 200+ km without dispersion compensation, dramatically simplifying long-haul system design. Throughout the 1980s and 1990s, DSF was the dominant long-haul backbone fiber for SDH/SONET STM-16 (2.5 Gbps) and STM-64 (10 Gbps) systems.

The Comparison: G.652 vs G.653 vs G.655

Why DSF Failed in the DWDM Era

In the late 1990s, DWDM (Dense Wavelength Division Multiplexing) emerged as the dominant technology for scaling long-haul capacity. DWDM packs 40-80 wavelengths spaced at 100 GHz (about 0.8 nm) intervals onto a single fiber pair, multiplying capacity by the number of wavelengths. Suddenly, the fact that DSF was optimized for a single 1550 nm wavelength was a liability rather than an asset.

The fatal problem with DSF and DWDM is four-wave mixing (FWM), a nonlinear optical effect. When multiple wavelengths travel through a fiber, they can interact through the fiber's nonlinear refractive index to generate spurious wavelengths at predictable frequencies. The strength of FWM depends on phase matching: it is strongest when the wavelengths travel at exactly the same speed, which happens at the zero-dispersion wavelength.

In DSF at 1550 nm, all DWDM wavelengths travel at essentially the same speed because dispersion is zero. FWM is at its maximum, generating noise that interferes with the actual data signals. The signal-to-noise ratio collapses, and DWDM transmission becomes essentially unusable beyond a few hundred kilometers.

Service providers who had invested heavily in DSF infrastructure during the 1990s found themselves unable to deploy DWDM upgrades on those routes. Some routes had to be recabled with G.652 or G.655 fiber. Others were limited to single-wavelength operation while neighboring routes built on G.652 supported full DWDM.

Non-Zero Dispersion-Shifted Fiber (G.655): The Fix

NZ-DSF (G.655) was developed in the late 1990s specifically to solve DSF's DWDM problem while retaining most of its advantages. The fiber's refractive index profile is engineered so that the zero-dispersion wavelength falls outside the 1530-1565 nm C-band window where DWDM operates. The dispersion at 1550 nm is small (typically 2-6 ps/nm/km) but non-zero.

That small amount of dispersion is the key. It is enough to make different DWDM wavelengths travel at slightly different speeds, breaking the phase matching that drives four-wave mixing. FWM is suppressed by orders of magnitude compared to G.653 DSF. At the same time, the dispersion is small enough that 10 Gbps transmission can run 100-200 km before requiring dispersion compensation.

NZ-DSF was widely deployed for long-haul DWDM in the 2000s and 2010s. Major variants include Corning LEAF (Large Effective Area Fiber), Lucent TrueWave RS, and OFS TrueWave RS Plus. Each manufacturer optimized the dispersion profile for slightly different goals (effective area, dispersion slope, polarization mode dispersion), and the G.655 standard accommodates this variation by defining several sub-categories (G.655.A, B, C, D, E).

The Coherent Revolution: Back to G.652.D

In the 2010s, the long-haul transmission landscape shifted again with the deployment of coherent optical transceivers. Coherent receivers use heterodyne detection and digital signal processing (DSP) to recover the optical signal's amplitude and phase. The DSP can compensate for chromatic dispersion electronically, regardless of how much dispersion the fiber introduces.

This changed the economic calculation. With coherent 100/200/400 Gbps transceivers, the fiber's dispersion characteristics no longer matter much. The signal can run thousands of kilometers through ordinary G.652.D fiber and the receiver compensates for the accumulated dispersion in real time. The fiber selection no longer needs to optimize for zero or near-zero dispersion at the operating wavelength.

The result: new long-haul deployments use standard G.652.D fiber with coherent optics. NZ-DSF is still used in some networks (existing deployments are kept, and some new builds use G.655 for niche applications), but G.652.D plus coherent optics dominates new long-haul builds. For more on standard singlemode see our OS1 vs OS2 guide.

Where DSF Still Lives

Legacy Telecom Backbone

Long-haul routes built in the early 1990s often have G.653 DSF in the ground. These routes still carry traffic but are limited to single-wavelength operation or to DWDM with significant FWM penalty. Some service providers have replaced or supplemented this fiber with G.652.D or G.655; others continue using it for low-rate single-wavelength applications.

Submarine Cable

Some submarine cables built in the 1990s used DSF or near-DSF designs. Modern submarine cables use specially designed singlemode fiber optimized for the unique requirements of submarine transmission (long distance, EDFA amplification, high power tolerance, sometimes Raman amplification). G.653 is not used in new submarine designs.

Niche Single-Wavelength Long-Haul

For specific point-to-point links where DWDM is not needed and 1550 nm single-wavelength transmission over very long distances is the goal, DSF still provides the lowest dispersion-distance product. These applications are rare but exist in some specialized military, research, and metro applications.

Identifying DSF in an Existing Plant

If you inherit a fiber plant that may contain DSF, you can identify it through testing:

• Dispersion measurement: A chromatic dispersion measurement at multiple wavelengths reveals the zero-dispersion wavelength. G.652 has zero at 1310 nm; G.653 DSF has zero at 1550 nm; G.655 NZ-DSF has zero outside the 1530-1565 nm window.
• Cable print legend: If the original cable jacket print is readable, the fiber type designation (G.652, G.653, G.655) appears in the manufacturer information.
• Vintage and route history: Long-haul routes installed between 1988 and 1998 are likely DSF candidates. Routes from 1998-2010 are likely G.655 NZ-DSF. Routes from 2010+ are likely G.652.D.
• FWM observation: If you put up a DWDM signal and see severe noise that disappears when you reduce the number of wavelengths, FWM is the cause and DSF is likely.

For dispersion testing equipment, modern OTDRs from EXFO, VIAVI, and AFL Globe-trotter offer chromatic dispersion measurement modules. The measurements are typically performed at multiple wavelengths over a known fiber span. For OTDR fundamentals see our OTDR basics guide.

Specifying Fiber for New Long-Haul Builds

Default: G.652.D with Coherent Optics

New long-haul builds in 2026 specify G.652.D singlemode fiber and pair it with coherent 100/200/400 Gbps transceivers. The DSP-based dispersion compensation eliminates the need to optimize the fiber's dispersion profile. Cost is minimized because G.652.D is the highest-volume singlemode fiber type with the lowest unit cost.

Alternative: G.655 NZ-DSF

For specific applications where coherent optics are not available or not cost-effective and DWDM is needed, G.655 NZ-DSF remains a viable choice. Some metro and regional networks continue specifying G.655 for new builds.

Avoid: G.653 DSF

G.653 DSF is essentially obsolete for new builds. Its incompatibility with DWDM rules it out for most long-haul applications. Even single-wavelength systems can use G.652.D with coherent optics for greater flexibility.

Related Reading

• OS1 vs OS2 Singlemode Fiber -- the modern singlemode standards.
• Single-Mode vs Multimode Fiber -- fundamental fiber types comparison.
• OTDR Basics -- testing long-haul singlemode fiber.