Attenuation in fiber optics is the gradual loss of light signal strength as it travels through a fiber cable. It’s measured in decibels per kilometer (dB/km), and it determines how far a signal can travel before it becomes too weak to read. A standard single-mode fiber operating at 1550 nm loses about 0.22 dB/km under normal conditions, meaning even the best glass in the world slowly eats away at your signal over distance.
Understanding attenuation matters whether you’re planning a network, troubleshooting slow links, or just trying to figure out why fiber has distance limits. The causes range from the physics of glass itself to something as simple as a cable bent too tightly around a corner.
How Light Gets Lost in Glass
Two fundamental mechanisms cause attenuation inside the fiber itself: absorption and scattering. These are intrinsic to the glass, meaning they exist even in a perfectly manufactured, perfectly installed fiber.
Scattering is the bigger factor at the wavelengths most networks use. The silica glass in optical fiber contains tiny density variations, frozen in place when the glass solidified during manufacturing. When light hits these microscopic irregularities, some of it scatters in random directions and escapes the fiber core. This effect, called Rayleigh scattering, gets stronger at shorter wavelengths, which is one reason longer-wavelength signals (like 1550 nm) travel farther than shorter ones (like 850 nm).
Absorption works differently. The glass itself absorbs certain wavelengths of light and converts that energy into heat. At very short wavelengths, photons have enough energy to excite electrons in the glass structure. At very long wavelengths, the light interacts with the vibrations of the glass molecules themselves. Fiber optic systems operate in a sweet spot between these two absorption zones, where the glass is most transparent.
The Water Peak Problem
One of the most well-known sources of absorption loss comes from trace amounts of water trapped in the glass during manufacturing. Hydroxyl ions (essentially fragments of water molecules bonded to the silica) create a sharp absorption spike near 1380 nm. This “water peak” can reach staggering loss levels, over 150 dB/km in some measurements, making that wavelength essentially unusable in older fiber types.
Modern manufacturing has largely solved this problem. Fibers designated as “low water peak” or “zero water peak” reduce hydroxyl contamination enough to open up the full range of wavelengths between 1260 nm and 1625 nm for transmission. If you’re working with older fiber installed before the mid-2000s, though, the water peak may limit which wavelengths you can use effectively.
Bending Loss: Macro and Micro
Not all attenuation comes from the glass itself. The way fiber is handled and installed introduces additional loss, and bending is the most common culprit.
Macrobends are bends large enough to see with your eyes. Routing a patch cable too tightly around a corner in a rack, stuffing excess fiber into a small splice enclosure, or running cable around a sharp building corner can all cause macrobend loss. When fiber bends past a certain radius, light at the outer edge of the bend escapes the core instead of reflecting back inward. Bend-insensitive fiber, now standard in most installations, was specifically designed to tolerate tighter bends without significant signal loss.
Microbends are microscopic deformations along the fiber that you can’t see or easily detect. They happen when something presses against the fiber, like other cable elements squeezing it during manufacturing, or the fiber’s own coating contracting in extremely cold temperatures. There’s no direct test for microbending, making it one of the harder problems to diagnose. Careful attention to fiber coating design during manufacturing is the main defense.
Typical Attenuation Values by Fiber Type
Different fiber types have very different loss characteristics, and the wavelength of light you’re using makes a big difference.
For single-mode fiber (the type used in long-distance and high-speed networks), typical values under normal conditions are about 0.38 dB/km at 1310 nm and 0.22 dB/km at 1550 nm. Under ideal conditions, those numbers drop to around 0.3 and 0.17 dB/km respectively. This is why 1550 nm is the preferred wavelength for long-haul links: you lose roughly 40% less signal per kilometer compared to 1310 nm.
Multimode fiber, commonly used for shorter runs inside buildings and data centers, has significantly higher attenuation. At 850 nm, the standard maximum is 3.0 to 3.5 dB/km depending on the cable type. At 1300 nm, it drops to 1.0 to 1.5 dB/km. These higher loss numbers are one reason multimode fiber is limited to shorter distances, typically a few hundred meters at most for high-speed connections.
Connectors and Splices Add Up
Every point where two fibers join introduces some loss. The amount depends on the connection method.
A typical fiber connector (the plug-and-socket type you’d find on patch panels) adds around 0.5 dB of loss per connection. Higher-quality connectors under ideal conditions can get down to about 0.2 dB. On a short link with several patch panels, connector loss can actually exceed the loss from the fiber itself.
Fusion splices, where two fiber ends are permanently melted together, perform much better. A high-quality fusion splice adds only about 0.02 dB of loss. Mechanical splices, which hold fiber ends in alignment without melting them, fall somewhere in between. For long-distance links that may have dozens of splice points, the difference between 0.02 dB and 0.5 dB per connection becomes enormous.
How Attenuation Is Measured
The primary tool for measuring attenuation in installed fiber is an Optical Time Domain Reflectometer, or OTDR. It sends a pulse of light into one end of a fiber and analyzes what bounces back. The natural backscatter of light in glass (the same Rayleigh scattering that causes loss) actually becomes useful here: by measuring how the backscattered signal decreases over distance, the OTDR calculates the fiber’s attenuation rate in dB/km.
An OTDR trace looks like a downward-sloping line. The slope itself represents the fiber’s attenuation. Any sudden drop in the line indicates a localized loss event, like a splice or a tight bend. Spikes in the trace point to reflective events where light hits an abrupt change in material, such as a connector, a mechanical splice, or a fiber break where glass meets air. This makes the OTDR valuable not just for measuring total loss, but for pinpointing exactly where problems occur along a link.
Overcoming Attenuation Over Long Distances
Attenuation is the primary factor limiting how far a fiber optic signal can travel. For short links inside a building or campus, it’s rarely a concern. But for runs spanning tens or hundreds of kilometers, the signal eventually becomes too faint to detect reliably.
The original solution was electronic regeneration: converting the optical signal to electrical, cleaning it up, and converting it back to optical. This works but is expensive and adds delay, especially in networks carrying many wavelengths simultaneously on a single fiber.
Optical amplifiers replaced most regenerators starting in the 1990s. The most common type, the Erbium-Doped Fiber Amplifier (EDFA), boosts the light signal directly without converting it to electricity first. A short section of fiber doped with the element erbium is pumped with energy from a laser, and incoming signals passing through it get amplified. EDFAs work across a range of wavelengths simultaneously, making them ideal for modern networks that pack dozens of channels onto one fiber. By placing amplifiers every 80 to 100 kilometers, telecom networks can push signals across oceans.
For shorter networks, simply choosing the right fiber type, minimizing connectors, using fusion splices where possible, and operating at the lowest-loss wavelength your equipment supports are usually enough to keep attenuation well within budget.