An optical fiber consists of a cylindrical core of silicon dioxide (fused silica glass) surrounded by a cladding. The core and cladding comprise the optical transmission media (or optical waveguide) and are usually coated for protection.
Latency in optical transmission media is a consequence of limited velocity in the optical media. To determine the latency contribution of an optical fiber, it is necessary to know the refractive index of the glass used in the core, as well as the length of the optical fiber.
Refractive index is a measure of the degree to which the light rays will be bent (or refracted) when light enters the media. The refractive index also determines how much slower light will travel in the optical fiber compared to a vacuum.
The refractive index for the core and cladding is denoted as n1 and n2. The use of n for the refractive index is the standard nomenclature used in optical fiber data sheets.
Because light travels approximately 1.5 times slower through optical fiber than in a vacuum, the latency is 5 μsec per kilometer. The number 1.5 is referred to as the Index of Refraction and will vary slightly based on the wavelength of light being propagated and composition of the optical fiber.
As shown above, a straight section of a single fiber that is 10 kilometers long will contribute approximately 50 μsec of latency as compared to 33 μsec in a vacuum.
However, fiber between two locations is not always routed along the most direct path, so the latency of an optical fiber path must be measured after installation.
Latency cannot be eliminated—only managed. In optical network applications where latency can be detrimental, the contribution from active network equipment, optical transceivers, optical cable, and even optical cable routing must be carefully considered.
Selecting an optical fiber that has the lowest index of refraction for the wavelength of interest will mitigate latency. Using the shortest possible optical fiber during routing can also mitigate latency. The graphic below illustrates why an 11-kilometer link contributes five μsec more latency than a 10-kilometer link.
Both links have a data rate of 10 Gbps. Due to its shorter distance, the 10-kilometer lower latency link will receive 50,000 bits before the 11-kilometer link has received any.
The effects of latency are cumulative. Each message cycle (msg, ack) on the 10-kilometer link requires 100 μsec, while the 11-kilometer link, which is 10 percent longer, requires 110 μsec. So, in the time it takes to transmit 10 message cycles on the 11-kilometer fiber, the 10-kilometer link would have transmitted 11.
Active elements that contribute to latency—such as active network equipment and optical transceivers—usually have enough data provided in their technical documentation so the contribution of each element can be determined.
This information, in combination with the use of good optical fibers, will enable an experienced network designer to develop the best implementation strategy for a latency-sensitive application.
In today’s world, bandwidth demand is growing at an exponential rate, so latency in telecommunication networks is constantly being evaluated and methods are being developed to monitor, and, if possible, minimize latency. In optical fibers, latency is dependent upon the refractive index of an optical fiber and is relatively constant at a specific optical wavelength.
This enables data center operators—especially those that provide co-location services to the financial sector—to “calibrate” optical links to ensure uniform latency among all customers.
There is a direct correlation between latency and maximum bandwidth in optical fiber systems. Latency limits the maximum rate information can be transmitted because all systems have limits on the amount of information that can be “in-flight” at any one moment. Excessive latency can have a detrimental effect on high-bandwidth applications.
Content sourced from paper by Joseph Coffey, senior principal engineer, CommScope