Submarine cable

A submarine cable is an optical transmission line laid on the seafloor to connect landing stations on different coasts. Its core is a small number of fiber pairs, one fiber for each direction, protected by a steel strength member, a copper power conductor and polyethylene insulation. In deep water the cable is about the diameter of a garden hose; near shore, where fishing gear and anchors are the main hazards, it is armored with layers of steel wire and usually buried. Long-haul cables contain optical amplifiers, called repeaters, spaced typically 50–100 km apart, and are designed for a service life of about 25 years. The first transatlantic fiber cable, TAT-8, entered service in 1988; the global network now comprises several hundred cable systems.

Repeaters and power feed

Each repeater is a pressure-resistant housing containing erbium-doped fiber amplifiers, one per fiber, with pump laser diodes and supervisory circuits. There is no local power on the seabed, so the cable conductor carries a constant direct current, on the order of 1 A, fed from power equipment in the landing stations at voltages up to about 15 kV on the longest systems. The current passes through every repeater in series; the return path is the sea. The electrical power available per repeater, and the number of pump lasers it can run, has become a design limit, which is why newer cables increase fiber-pair count and run each pair at lower optical power and spectral efficiency, which gives more total capacity for the same electrical power.

Fiber and span design

Submarine fibers are optimized for low loss and low nonlinearity. Pure-silica-core fibers with attenuation near 0.15–0.16 dB/km at 1550 nm are common, with effective areas of about 110–150 µm², compared with about 80 µm² in standard single-mode fiber; a larger area lowers the intensity and so the nonlinear distortion for a given launch power. The loss mechanisms are described under fiber attenuation.

A span of 60 km at 0.16 dB/km has a loss of 9.6 dB, which each repeater restores. Every amplifier adds amplified spontaneous emission, and the noise accumulates over the whole link. The optical signal-to-noise ratio in a 0.1 nm reference bandwidth at the end of $N$ identical spans is approximately

$$\begin{aligned} \mathrm{OSNR} \approx{}& 58 + P_{ch} - NF \\ &- L_{span} - 10\log_{10} N \end{aligned}$$

with $P_{ch}$ in dBm per channel and the noise figure $NF$ and span loss $L_{span}$ in dB. For 0 dBm per channel, a 5 dB noise figure, 9.6 dB span loss and 110 spans (6,600 km), the OSNR is about 23 dB, before nonlinear penalties. Shorter spans raise the OSNR but require more repeaters and more power.

Transmission

Modern systems use coherent detection with digital signal processing, which compensates chromatic dispersion electronically, so the fiber is no longer dispersion-managed. Each fiber pair carries wavelength-division-multiplexed channels across the C band, sometimes the C and L bands, with a capacity per pair on the order of 20 Tb/s on transoceanic routes. Cables with 12 to 24 fiber pairs, an application of space-division multiplexing, reach total design capacities of a few hundred Tb/s. The submarine line terminal equipment at each end can be upgraded during the cable's life, so the capacity of a cable often rises well beyond its original specification.

Branching units split fiber pairs to intermediate landings, and in newer systems reconfigurable wavelength switching allows wavelengths to be routed between branches. Short routes, up to a few hundred kilometers, are built unrepeatered, relying on high launch power, Raman amplification and remotely pumped erbium amplifiers.

Latency

Light in silica fiber travels about 4.90 µs per kilometer (speed of light in fiber), so a 6,600 km cable adds about 32 ms of one-way delay. Route length, set by landing points and seabed geography, matters as much as the fiber itself for latency-sensitive traffic.

Faults and monitoring

Most faults are caused by fishing and anchors in shallow water; the rest by seabed movements, abrasion and component failure. Faults are located from the landing stations using coherent OTDR and repeater loopback paths, and repaired by cable ships that recover both cut ends and splice in a new section. The fibers of in-service cables are increasingly used for environmental sensing, through distributed acoustic sensing and through changes in the polarization of transmitted light.

Common questions

How far apart are the repeaters on a submarine cable?

Typically 50–100 km, with 60–80 km common on transoceanic systems. Spacing is chosen jointly with fiber loss, amplifier noise figure and available electrical power.

How much data does a submarine cable carry?

Recent transoceanic systems have design capacities of a few hundred terabits per second across all fiber pairs, on the order of 20 Tb/s per pair. The figure depends on the terminal equipment installed and usually increases after the cable is laid.

How are submarine cables powered?

By direct current from the landing stations, through a copper conductor in the cable, at constant current of about 1 A and voltages up to roughly 15 kV. The seawater serves as the return path.

References: J. Chesnoy (ed.), Undersea Fiber Communication Systems, 2nd ed. (Academic Press, 2016); G. P. Agrawal, Fiber-Optic Communication Systems, 4th ed. (Wiley, 2010); E. Desurvire, Erbium-Doped Fiber Amplifiers: Principles and Applications (Wiley, 1994).