The wide-scale deployment of high-power dense wavelength-division multiplexing (DWDM) systems has forced standards bodies, telecoms operators and equipment makers to rethink their approach to laser safety over the past few years. For the operators, the priority is simple: to ensure that personnel - whether cable installers, equipment technicians, field engineers or even the general public - are protected from laser hazards under all reasonably foreseeable conditions.
Lasers provide a source of intense light output, emitting a high-energy, collimated and concentrated beam. Safety becomes an issue when that beam is concentrated in a small spot size, creating the risk of eye damage and tissue burns. While the eye is most efficient at focusing visible light, it is still fairly efficient at the near- and mid-infrared wavelengths used in telecoms transmission.
Of course, the laser hazard is more complex in an optical fibre communication system (OFCS) because the light is distributed over links that extend tens, if not hundreds, of kilometres from the source. In this environment, the only way to avoid laser-inflicted injury is to avoid exposure or, failing that, to reduce any exposure to safe levels. The question is: how can telecoms operators manage that exposure in an efficient and infallible manner?
Classification and control
Lasers are classified under the extensive International Electrotechnical Commission optical safety standard IEC60825, with the technically equivalent European standard known as EN60825. The standard is divided into several parts, with the first acting as the "core" document, describing the classification scheme and dealing with discrete lasers (equipment such as laser pointers, carbon-dioxide lasers used in welding applications and so on). Part 2, which provides the basis of this article, deals with the safety of OFCSs.
The classification scheme is based on a laser's power and potential to cause harm (see "Top class: lasers measure up" in Further information for a description of the categories). Of course, individual laser components can be controlled locally, which means that exposure to laser radiation is only possible over a limited distance from the equipment. This contrasts markedly with the situation in an OFCS, which is why OFCSs merit their own section in the optical safety standards.
The head-end transmission unit in an OFCS can and should be classified by the manufacturer or supplier. However, things get more complicated when this unit is connected to an optical fibre and becomes part of an OFCS. Why? Chiefly because the engineer working at any point along the fibre will more than likely have no knowledge or control of the head-end equipment, its power and capabilities. Also, the manufacturer will rightly point out it has no control over the use to which its equipment is put in an OFCS.
For this reason, the concept of "hazard level" is introduced in the second part of the optical safety standard. While the laser class depends (among other considerations) on the power emitted by the unit, the hazard level is defined as "the maximum class or power to which a person could be exposed at any point on an OFCS under a fault condition". A fault condition includes such mishaps as a fibre-cable break or a connector failure.
To give an example: if the head-end transmitter is class 1 (excluding considerations of any fault conditions, and assuming an unamplified system), the maximum power anywhere along the OFCS is class 1. In other words, the OFCS is hazard level 1. Similarly, if the transmitter is class 1M, the OFCS is hazard level 1M. To extend the concept, it is legitimate to use higher powers in the fibre (exceeding 1 W in certain circumstances) if the system incorporates automatic power reduction (APR). APR shuts off the system quickly enough to ensure it is not possible to receive a hazardous "dose" of radiation, so that OFCS as a whole can still be hazard level 1M or even 1.
It's important to remember that class applies to the stand-alone transmission unit, while hazard level applies to the whole OFCS. In fact, hazard level only has a meaning when applied to an OFCS. What's more, while it is the responsibility of the supplier to classify the stand-alone transmission unit, and supply performance and power data, it is up to the network operator to assess the hazard level of the end-to-end OFCS.
This is only fair, since manufacturers cannot be expected to know how their equipment will be put into service. Instead, the operator has the obligation to assess the hazard level of the OFCS, but also has the right to relabel equipment. For instance, although a piece of stand-alone kit might be class 3B, the way the equipment is used in the field may dictate a "safe for accidental eye viewing" hazard level of 1M.
Power management
Today's optical transport systems - particularly networks incorporating erbium-doped fibre amplifiers and DWDM gear - may exceed class 1M power levels (i.e. those safe for accidental eye viewing) at some point along the transmission path. In this scenario, an APR or automatic power shut-down (APSD) system can ensure that potential eye (or even skin) exposure does not exceed the maximum limit as set in the 1M standard.
So, as long as the APR or APSD kick in rapidly (within 1 or 3 s depending on the fault location), it is permissible under certain circumstances for powers in the fibre to exceed 1 W. In this way, the safety of the whole system relies on the APR/APSD operation, and to a much lesser extent on the reliability of the other OFCS units.
Studies here at British Telecom (BT) indicate that a sufficiently safe measure of reliability is 500 FITs (failures in 109 hours). What this means is that the probability of optical systems exceeding their given hazard level must not be greater than 500 FITs. Unfortunately, as OFCSs become ever more complex, particularly with the addition of DWDM capability, it becomes difficult to comply with this specification.
With this in mind, the area of APR/APSD and their reliable implementation is under discussion in the optical standards bodies, with input from manufacturers, network operators, telcos, government bodies and other interested parties: every contributor brings their own view on optical safety.
Controlled or 'dead' working
One way to control laser exposure is to ensure that engineers and technicians are only allowed to work on "depowered" fibres. Such an approach requires efficient access control and fibre identification (not always easy on a 96-fibre cable). Engineers in the field have to communicate with personnel at the transmission end of the fibre, who will shut down the appropriate equipment, apply key control (in which the equipment is locked in the off state and the key held by a suitable nominated person) and label it appropriately to ensure it is not inadvertently switched on again until the work is done.
Key control and labelling carry significant overheads and involve extra, probably unnecessary, work. Key control requires the writing of management procedures to ensure equipment is powered down and reactivated appropriately. It also needs additional equipment to lock the transmission gear. Moreover, the extra person involved will most likely be unable to do anything other than switch off equipment, twiddle their thumbs and then switch on the equipment again.
An alternative to depowering is to label all accessible locations (and this implies that fibre cables themselves are labelled at appropriate intervals on the outer sheath) and allow personnel to work with fibres that are powered up - though only if the hazard level is 1 or 1M. Again there are drawbacks. For a large service provider like BT, for example, a massive effort would be needed to label all manholes, street cabinets, distribution points and the like. On top of this, the optical safety standards are continually under review, and a change in standard might require, say, every 3A label to be changed to 1M.
In short, the disadvantages of dead working or label-controlled working are significant. So what about live working?