Article | November 6, 2000

Optical Access Network Architecture Focuses On Design Tradeoffs

Optical Access Network Architecture Focuses On Design Tradeoffs
Although all-optical access networks are still far in the future, tradeoffs between power splitting and WDM will open the door for robust optical networks in the access loop.

By Jocelyn Lauzon and Pierre-Yves Cortès

To bring fiber-to-the-home (FTTH) to reality, carriers must deploy optical access networks on a broad scale. The optical access network environment is very different from that of long-haul optical networks, however. In long haul networks, the focus is on optimizing performance, assuming a certain maximum price ceiling. In access networks, the focus is on minimizing cost, assuming some baseline link performance.

The design architecture of the physical layer of these access networks is key to minimizing cost. Among the primary questions the access network designer must ask are:

  • How will the end-user interface connect to the optical access network?
  • What is the end-user interface?
  • How do we use the optical bandwidth efficiently?
  • Is the all-optical approach the best?
  • Is all-optical really all-optical?
  • Do we choose passive or active routing?

Challenges to FTTH
A fundamental concern in designing an optical access network is how the end-user interface connects to this network. Will your phone, TV or computer have integrated optical sources and receivers? One disadvantage to optical components in the home is cost; another is safety for the end user. The security and safety standards associated with optical power are very severe, and thus quite limiting. Moreover, high-speed electronics will continue to evolve, becoming more highly integrated and more economical. It is thus doubtful that photonic components will find a place within the interface.

Wireless is becoming a popular alternative for obvious practical reasons. The authors believe that wireless interfaces will become an interesting choice and that the transition toward optical access networks will probably favor and accelerate this trend. In the future, free-space optical communications within single rooms inside buildings or short distance radio-frequency communication from the front yard to the end-user interface may be used widely (see Figure 1).

Figure 1. Wireless interfaces in the periphery of the optical access network.

The concentration of services over a single access network could also mean that a single end-user interface would be used. Although the look of the end-user interface of the future remains to be determined, it will probably be a computer, since it will ultimately include a processor. Already, through the Internet, the computer can provide most servicesvoice, video, and data. Even today's cellular phones have integrated processors and are in fact small mobile computers with particular functions.

The end-user interface might be called something else than a computer in order to ease its acceptance in everyone's household, fundamentally it will still be a computer (see Figure 2). Also, the unique end-user interface might in fact be of dual format; a very practical and mobile miniature interface, plus a high-quality much larger interface.

Figure 2. Tendency toward having a single communication interface per household.

Power splitting versus wavelength splitting
At the receiver level, the signal has to be demultiplexed from a central office all the way to each household. Passive optical networking (PON) should be used as much as possible in order to reduce costs and management requirements. This is also symmetrically true at the transmitter level of course. This task can be performed in two different ways either by optical power splitting or by wavelength splitting (see Figure 3).

Figure 3. Wavelength splitting versus optical power splitting in the optical access network architecture.

Wavelength splitting is best known as wavelength division multiplexing (WDM). In this case, dense wavelength division multiplexing (DWDM) is not appropriate since wavelength channels in the access loop can be more widely spaced to reduce the cost of selection filters. And also because the transmission window is, in most cases, not limited by optical amplifier bands and can be as large as many hundred of nanometers.

Optical power splitting is much cheaper and simpler to do; however, it means broadband transmission and selection by the receiver. The approach opens up the possibility of data interception, creating concerns for privacy and security in certain applications, such as bank transactions. Power splitting also means wasting optical power, which could in turn create the need for optical amplifiers in the access loop. Optical amplifiers should be avoided as much as possible in order to reduce costs.

In some particular cases, optical amplifiers might be needed, even for access networks. It could be for power splitting architectures where optical power is wasted, but also for wavelength splitting architectures since the more complex optical components are usually associated with larger insertion losses. When unavoidable, low performance cheap optical amplifiers will be used. In a wavelength splitting architecture, the number of channels going through each optical amplifier will be limited in order to take into account their limited performance.

WDM in the access loop
WDM solves part of the privacy and security issue in the access network, but such a system is more complex to reconfigure or to be made to offer the necessary protection. For an access network, reconfiguration depends on each end-user and is thus an even more important factor than for metro or long-haul networks. A WDM access network necessitates the use of much more complex optical components. Complexity means increased cost, a key factor for access networks. It should also be stated that WDM will almost never mean using one wavelength per user, for obvious cost and reconfigurability reasons. An efficient wavelength re-use architecture must then be implemented.

Thus, in order for WDM optical access networks to become a serious contender, robust and low-cost tunable optical add-drop multiplexer (OADM) technology will have to be developed. This technology will probably reach fruition in the not-so-distant future, because some of these components are already being proposed. Manufacturing automation and integration of these optical components should also drive their cost down.

A trade-off between power and wavelength splitting should thus be made when designing the physical layer of an optical access network. This trade-off is made considering each particular case; the priority given to different specifications, the size of the access network and how static or not it might be. In most cases, combining both power and wavelength splitting is the best choice. Usually wavelength splitting will be used more in periphery in order to insure more privacy and security over the network. It is clear that as time goes by, as complex optical components become cheaper, as privacy and security is becoming more and more the top priority, wavelength splitting will become more and more popular.

The tendency towards all–optical systems will continue, although the short-term implementation of true all-optical access networks is doubtful. High-speed electronics will continue to evolve and will be used in periphery of the communication systems. These electronic functions will be kept in the periphery of the systems in order to avoid bottlenecks.

Passive optics will be able to provide some functions as mentioned previously. Active optics will need to be used within some sections of the access network architecture in order to switch, to route, to reconfigure and to offer protection in the systems. The current optical switches offer part of the solution, but will need to offer faster response times for them to cover more ground. In digital communications, if time-division multiplexing (TDM) is not completely avoided (which would be doubtful), the peripheric routing of a signal mandates reading a header. This task cannot be performed optically at the present time, since it requires optical memories.

The only other way to go truly all-optical would be to change the protocols from using digital headers to using optical protocols such as color coding. This approach would also require time for development. Thus, truly all-optical access networks are not foreseen for the short term. What's foreseen are low-cost strong and robust optical access networks.

Optical access networks are becoming a necessity in order to meet the demand of end-users. The emergence of these networks should be the catalyst for many changes for the end-users such as the concentration of services and going toward a single communication interface per household. As for the architecture of these optical access networks, it will evolve as the optical components technology evolve in lowering production costs and providing new breakthroughs such as low-cost tunable OADM. The tendency will be towards WDM PON, however, power splitting and some electronic functions will continue to share the tasks of optical access networking for at least a few years to come.

About the authors…
Jocelyn Lauzon and Pierre-Yves Cortès, INO, 2740 Einstein street, Ste-Foy, Québec, Canada, G1P 4S4.