High fibre diet

However, as data rates climb and the standards-based rules of distances become more flexible to allow longer cable runs, the technology pendulum swings back in favour of fibre. If this is taken in conjunction with cost savings in fibre optics cabling and systems, then perhaps it is time to review and update our thinking about the role of fibre-to-the-desk (FTTD).

The case for fibre
There are many reasons for using fibre optics as a communications medium, apart from its mind-boggling ability to transport stupendous amounts of information over tremendous distances:

• no electromagnetic current issues – the data that travels along a fibre is not affected by electromagnetic interference of any sort. Similarly, it does not cause any interference to anything else around it;
  
• security – it is very difficult to tap into the signal that is travelling along a fibre as any intrusion is detectable;
  
• compact – fibres are incredibly skinny.

But it is bandwidth that is central to the case for fibre. Computers are getting more powerful, memory is getting cheaper, applications are getting more and more bandwidth-intensive. The ubiquitous Ethernet protocol has responded by increasing its data rate from the original 10 Mb/s to 100 Mb/s, which we used to call 'fast' Ethernet, through to 1 Gb/s and now the 10 Gb/s standard that was ratified in June.

The original 10 Mb/s Ethernet was a shared bus topology with thick, yellow, coaxial cables. A later common implementation used thinner, cheaper, coaxial cable. When the 10BaseT Ethernet standard was developed for untwisted pair (UTP), all that was needed (and the best UTP available at the time) was Category 3 cable.

Any old fibre would easily support 10 Mb/s over desktop-type distances. Some early FTTD installations in the City of London were carried out with 200PCS fibre. Plastic optical fibre is another possibility for use at these data rates.

At 100 Mb/s came the first protocol that had to use fibre – FDDI (Fibre Distributed Data Interface); initially there was no UTP cable capable of supporting this data rate. This was the first time that fibre bandwidth became an important issue and the bandwidth requirements forced operation at the (more expensive) 1300 nm wavelength window, where 500 MHz.km bandwidth was specified. Even so, the maximum length was bandwidth limited to 2 km on multimode fibre.

The massive growth in the structured cabling market was kicked off by the development of Category 5 UTP capable of supporting 100 Mb/s to the desktop, coupled with the arrival of generic cabling systems standards. This was a severe blow to the role of FTTD; although 100 Mb/s on fibre over 100 m is certainly no problem to almost every fibre, the cost of the fibre transceivers was prohibitive.

The emergence of Gigabit Ethernet meant radical changes in cabling technology for copper and fibre. For copper UTP it meant a change from simply using one pair for transmit and one pair for receive as had been the case up until then. 1000BaseT requires all four pairs in the cable to carry traffic in both directions simultaneously. This meant a whole new raft of tests and the definition of Enhanced Category 5 cable.

For fibre it meant that we could no longer use cheap LEDs as the light source for systems. LEDs cannot be switched on and off fast enough. Therefore, Gigabit Ethernet meant that we had to start using lasers as the light source. Fortunately a new type of laser emerged at this time to give laser-type performance at LED-type prices. This new laser was the VCSEL (rhymes with pixel), the Vertical Cavity Surface Emitting Laser. This operated at 850 nm initially – though 1300 nm technology is coming along.

However lasers launch light into multimode fibre in quite a different way to LEDs, so this has some important implications for the way we specify and test multimode fibres and has also led to a whole new type of laser-optimised multimode fibre. If we consider the pessimistic minimum bandwidth performance specified by the fibre standards, this may limit transmission distance to just 220 m. Of course, when talking about 10 Gigabit Ethernet we're not really talking about current implementation to the desktop – however, if a cabling system is being installed with an anticipated lifetime of many years it may be sensible to plan for the future.

The low transmission losses associated with optical fibres allow long cable runs before the power level needs to be boosted up again. However, in all high performance fibre optic systems today, transmission distances are more likely to be limited by bandwidth and distortion factors than by a lack of signal strength. Telecomms systems use optical amplifiers to boost signals without regeneration to achieve transoceanic distances.

So why are extended distances important for a desktop cabling scenario? The answer lies in the increased flexibility that fibre can give you in planning your network. If you are using copper for your desktop connections then you are constrained to a maximum copper cable run of 90 m. This usually means that in a large installation there is a floor distributor or wiring cabinet on every floor of a multi-storey office block or in every region of a large office area. This cabinet often occupies expensive floor space and needs power and fans because it has to house a lot of electronic equipment.

Using fibre allows you to extend the desktop connection back to a central location where floor space is cheaper (such as the basement) and where full control may be exerted over patching and reconfiguration. Efficient utilisation may be made of switch ports, leading to further cost savings. This topology is known as a collapsed backbone or centralised office architecture (see figure 1).

The case against
A number of objections are often raised when FTTD is proposed: "fibre is too fragile", "it's difficult to install", "it's expensive" – these are outdated or uninformed perceptions.

Fibre is surprisingly strong, for a thin glass rod an eighth of a millimetre in diameter, however the office can be a harsh environment, especially when people try moving computers and desks without first disconnecting them. Experience has shown that this can be one of the most challenging aspects of a FTTD installation. However, selection of the right products, such as cable management in office furniture, coupled with a little basic awareness training for end-users, can solve the problems.

I don't believe that fibre is difficult to install. The five-day City & Guilds 3466 certificate course will produce a competent, professional fibre optics installer. There is no need for a six month apprenticeship before being let loose on real installations.

However, there have been many developments over the years that have made fibre optics easier to install. Fitting connectors using traditional methods can be tricky – that's why many installers use the quick and easy method of fusion splicing preterminated pigtails. The cost of a pigtail is cheaper than many of the rapid termination connectors now on the market, although these connectors (3M's Volition, Corning Unicam MT-RJ and others) can cut installation time dramatically compared with traditional methods.

An old objection to fibre to the desk was that a copper cable was still needed to provide a telephony service. Now, Voice over Internet Protocol can make the telephone into just another software application. An IP telephone or a headset can be plugged into the computer.

The costs
If a straight comparison is done for the cost of replacing existing copper horizontal distribution cable with identical fibre runs the cost comparison will probably favour copper – though the actual installed cost of the cabling itself is not likely to be too different. Many factors have evened out:

• the cable installation and containment costs will be the same;
  
• the cost of fibre compared with good quality (Category 6) cable will not be too different;
  
• there is not likely to be much, if any, premium for fibre labour;
  
• fibre connector costs have come down, as have termination times.

The major cost differential comes not from the cost of the fibre, but from the cost of the equipment at the ends of the fibres. However, if the cost comparison is conducted from the perspective of what can be achieved with a fibre-specific design then the balance may shift in favour of fibre. For example, if a centralised office architecture is used, the floor distributors and all of the equipment that is inside them is eliminated completely, together with their requirements for expensive office space and electricity.

The development of VCSELs has reduced the cost of high performance optical transceivers. Some equipment vendors have implemented single fibre working where signals are carried in both directions along a single fibre simultaneously – halving the number of fibres and connectors needed in the cabling.

Optical equipment has much lower power requirements than equivalent copper equipment, it is less likely to generate heat and therefore puts less demand on an air conditioning system. Active equipment may be mounted away from the working environment anyway.

Some of the developments that are happening in fibre may further reduce the cost of FTTD or increase the bandwidth requirements, hence the demand for FTTD:

• new types of single mode fibre suitable for operation at the cheaper wavelength of 850nm – this means cheap fibre and cheap equipment;
  
• bi-directional traffic is becoming common in optical access networks. This may involve different wavelengths of operation in a wavelength division multiplexing (WDM) environment;
  
• there is a lot of development work on producing optical integrated circuits, where transmitters and receivers are incorporated onto a single device together with the required WDM devices. Automated manufacturing and increased integration are driving costs down;
  
• on the demand side there is an increasing trend for rich internet content including video and audio to be delivered everywhere;
  
• desktop video conferencing using web cameras is a low cost option that is becoming commonplace.

It pays to keep an open mind on the role of FTTD and assess installations on a case by case basis.