From point-to-point cable connections to "self-healing" networks
There is a general trend to integrate modern undersea cable systems intimately into the overall world-wide SDH infrastructure. Cables initially were operated as point-to-point devices. Then they acquired branches, were formed into loop configurations for redundancy, and were designed with multiple drop and insert points. The use of WDM helps in these more complex system designs, since Bragg filters (which are passive optical devices, therefore reliable for undersea operation) can be used to selectively extract and insert one wavelength to or from a side branch onto the main cable. In the industry jargon these devices are OADMs, for Optical Add/Drop Multiplexors (or Branching Units).
Modern cable systems are now almost always designed as long loops, connecting two landing stations in one country, which are separated by some hundreds of km, from two similar landing stations at the other side of the ocean. The signals are typically transmitted at 2.5 Gbps (STM-16) on the primary path. If there is any physical break or electronic mal¬function in the primary path the signals are switched to the back-up path, which is on the other (parallel) cable, within a few tens of milliseconds. Since the two main lengths of cable are typically several hundreds of km apart under the ocean the chance of both being affected by the same marine "event" are very small, and the overall availability of such self-healing SDH rings seems to be very high.
Power on repeatered cables
When repeaters are needed they must be powered. The standard approach is to send a constant current of about 1A from one end of the cable to the other, along a copper sheath which lies outside the fibres and inside the armour (if present). Each km of cable offers a resistance of close to 1 ohm, and there is a drop of about 20 V across each repeater, leading to a requirement of close to 10 KV across a typical 7500 km oceanic crossing with 100 repeaters. In branched cable systems the power management becomes somewhat more complex, and the branching units incorporate very high reliability relays to cope with the power reconfiguration needed in case of repairs.
Some of the engineering limits
Many papers presented at the SubOptic’97 conference dealt with the topic of the engineering limits which are currently being faced by submarine cable design teams. These address almost all elements of cable systems, including the fibres themselves, where there is a search for a broader effective area and for features, including dispersion management, which will allow better handling of WDM and/or soliton transmission; the laser transmitters; the optical amplifiers; the detection systems; cable and repeater design for more fibre pairs; and signalling formats including improvements in the forward error correction techniques.
Marine factors
The main source of problems with submarine cables comes from external sources such as fishing and anchors, and the damage is either via compression, when a heavy weight hits a cable, or via dragging, and, in the limit, breakage. Anchors from the largest vessels can penetrate the sea-floor by several meters, but present no problem at depths of more than 150 m. Fishing nets, on the other hand, penetrate less deeply, but are operational at depths of up to 1500 m. Since the early 1980s submarine cable has been buried to a depth of typically 1 meter in areas where threats from anchors or fishing can be anticipated, and, indeed they avoid anchorages used by deep-sea vessels as much as possible. The cables are also armoured, typically with one or two layers of high tensile steel strands, when they have to be laid in those areas. Burial, and the use of armoured cable, makes the installation slow and expensive, but seems to pay off well in terms of reliability.
There is a fleet of some thirty more or less specialised vessels used world-wide for submarine cable installation and repair. The largest vessels are not cheap (close to 100 M$ by the time they are ready for sea), but can install many thousands of km of cable (with the repeaters already spliced in) in a single operation. They have several very deep holds which can take a few weeks to fill with the cable - the filling is a manual operation. They can then lay the cable at speeds of about 7 knots (15 km/hour) in the deep ocean, at depths of up to 7000 m.
While the position of the ships as they lay the cable is well understood, via the Global Positioning System, the position of the cable on the ocean floor is less well known, owing to a variety of dynamic effects, including motion induced by currents at different depths, as up to 20 km of cable drop from the ship. In a recent press release Cable and Wireless Marine announced that the accuracy of its laying of the Gemini cable at the bottom of deep ocean was consistently better than 200m. It is obviously important to lay the cable with a small (but positive!) amount of slack, since loops or suspensions on ridges are very likely to cause subsequent problems via chafing or by being snagged.
Closer inshore, ploughing-in of cable can only proceed at speeds of less than about 1 knot, and special care may have to be taken with very rocky marine bottoms, or in the presence of strong currents, which can uncover cables buried deep in sand, and near anchorages or fishing grounds.
Experience indicates that less than 20% of all repairs are in the deep ocean, where the intrinsic faults of the cable system are the major factor, while repairs at depths of less than 1000 m. are mainly due to “external factors”. One captain with experience of repair work indicated that one repair every three years might be typical for any given North Atlantic cable, whereas one repair every five weeks would be more normal for a cable crossing the North Sea, where there is very heavy fishing activity and the tides are continually moving the sandy sea floor. TSSL’s accumulated experience shows that modern cable systems should only need of the order of one repair for internal failures (i.e. not counting external events) during their 25-year planned lifetime.