The emerging optical layer: Photonic technology developments aim to go beyond congestion relief to enable more efficient network restoration, wavelength grooming and ultimately, end-to-end wavelength-based services

Data traffic in the public network is outpacing the familiar Moore's Law by doubling about every 12 months. The cause for this growth is fueled by cost and capacity.

In many respects, the demand is occurring because matching capacity exists and for a lower cost than ever before. Thanks to new technologies such as optical amplifiers and dense wavelength division multiplexing (DWDM), the network infrastructure's cost is dramatically cheaper than in the past.

For example, optical in-line amplifiers have substantially displaced Sonet and synchronous digital hierarchy (SDH) regenerators in the backbone networks. When used with DWDM, the cost of regeneration with optical amplifiers is reduced up to 40 times.

This lets service providers offer high-bandwidth applications such as Internet and videoconferencing to a broader market, which in turn drives the network to higher capacities. The principles of supply and demand through lower cost for a packet of data are at the heart of the growth.

Now, 16-channel unidirectional DWDM is being deployed widely in both 2.5 and 10 Gb/s applications, as well as eight-channel bidirectional systems. These systems continue to be used primarily in point-to-point applications overlaid with high-capacity Sonet rings.

Point-to-point applications can have both embedded and open architectures. Carriers use transponder technology in the open architecture to manage a multivendor Sonet environment, and they often deploy embedded systems where the same manufacturer provides Sonet and DWDM solutions.

Already, 40-channel DWDM systems are appearing on the market, with lab trials underway. Interestingly, vendors also are advertising upgrades of the systems to 80 channels and higher through one of two techniques-tighter channel spacing or wider band amplifiers.

But the goal for these systems is not simply more channels. The key is increasing the amount of usable bandwidth in the fiber through a combination of time division multiplexing, WDM and wider band amplifiers.

Systems with 16 channels use either 200 GHz spacing on a wideband (30 nm) amplifier or 100 GHz spacing on a traditional (20 nm) bandwidth amplifier. But building a 40-channel system requires using either 100 GHz spacing on a wideband amplifier or 50 GHz spacing on the traditional amplifier.

The narrowest spacing compatible with 10 Gb/s transmission techniques is 100 GHz spacing, which equals 0.8 nm (Figure 1). Because the true information bandwidth of an optical digital signal is about two times the bit rate, and optical filters are not perfectly square or stable, 50 GHz spacing does not supply ample margin for a 10 Gb/s signal.

However, for 40-channel systems that use 100 GHz spacing, up to 400 Gb/s per fiber strand is now possible, giving a ten-fold increase in capacity over systems being deployed just two years ago.

More than just DWDM Regardless of the quest for more bandwidth in long-haul networks, the appearance of wavelengths as building blocks in the optical network opens opportunities. Competitive interexchange carriers such as Qwest and IXC Communications already have started a lucrative business of leasing individual wavelengths to other carriers that need the bandwidth but cannot afford the time or expense of building a network.

By selling the wavelength rather than dark fiber, the carrier can benefit from the large capital investment it has made in its fiber plant without giving up the fiber assets to customers. As one carrier network planner noted, "If I sell a fiber, I may have a competitor, but if I lease a wavelength, I have a customer."

The concept of leased wavelength in the backbone network is only the beginning of what is known as the emerging optical layer. Instead of being focused strictly on issues of fiber congestion relief in the network infrastructure, photonic technology developments will be directed toward this emerging layer to enable more efficient network restoration, wavelength grooming and ultimately, end-to-end wavelength-based services (Figure 2).

The industry's short-term challenge is to offer ultimate flexibility in local networks via bit-rate and protocol transparency, while simultaneously maximizing efficiency in long-haul backbones through fixed 2.5 and 10 Gb/s wavelengths.

Concepts being applied to the electrical domain, including time slot assignment and interchange, survivable ring switching and mesh network restoration, will be applied through the optical layer. Network functions now provided by Sonet multiplexers are being moved into the optical layer, which ultimately may hasten the demise of the stand-alone Sonet device.

In-line optical amplifiers already replace Sonet regenerators, while digital cross-connects, asynchronous transfer mode switches, Internet protocol (IP) routers and next generation digital loop carriers provide direct Sonet interfaces.

Because the optical domain promises to provide the flexibility of bit-rate transparency, new carriers could avoid the deployment of stand-alone Sonet network elements that only provide voice-optimizing multiplexing.

The all-optical network The all-optical network of the future will enable a new class of high revenue, end-to-end wavelength services for customers that want complete flexibility for their high-bandwidth data needs. However, several issues need to be resolved before affordable end-to-end wavelength services can be achieved.

To offer wavelength-based services, the wavelength must appear at all end-user locations. This means that WDM must find its way from backbone applications into local access networks.

But where fiber congestion relief in long-haul networks drove an obvious business case for WDM, it is unlikely that this will be the vehicle by which WDM is proved in local, short-haul networks. First, fiber congestion is not yet as large a problem in most local networks, and where it is, additional fiber often can be pulled through an existing conduit at a lower cost than WDM systems.

Nevertheless, the vendor community is working to adapt its systems for lower-cost, short-haul applications by removing the amplifier function and using lower-cost lasers and simpler WDM devices.

Local exchange carriers are evaluating economic feasibility to determine where the crossover occurs between deploying more fiber and installing an access WDM system. Meanwhile, manufacturers will continue to extract cost from their systems to achieve these targets through optimized designs and volume efficiency.

Telecom reform may prove to be a WDM driver in the local network because incumbent LECs ultimately could be required to unbundle their network infrastructure for competitive access. However, a resolution in this area could be years away because it is tied up in court battles.

The real WDM catalyst The true catalyst behind WDM in local networks may be through the technology's promise of transparency in offering a new high-end wavelength-based service. Transparency refers to the bit-rate and protocol independence of the optical layer (Figure 3). With a wavelength service, users potentially would have the flexibility to transport any kind of data without regard for the restrictions of SDH.

A mixture of voice, data and video could be carried in any format, and the data rate could change as needed-all within the end user's control. WDM systems are being developed now that offer direct IP and ATM interfaces or purely asynchronous interfaces, demonstrating the potential of the technology to displace the Sonet terminal as a stand-alone multiplexer.

Transparency in the local environment is a realistic potential both technically and economically. Ultimately, the wavelength would have to travel over the backbone network for transcontinental or global networks.

Transparency is a significant issue in long-haul networks where the optical signal still must be regenerated occasionally. Technology may never exist for all-optical regeneration.

After dispersion and signal-to-noise limits have been reached through a series of cascaded amplifiers, the composite WDM signal must be demultiplexed and each channel electrically retimed, reshaped and retransmitted. Providing this kind of capability in the optical domain without demultiplexing the channels may be years away.

Yet it is unlikely that the backbone carrier, wishing to operate its wavelengths with the highest bit-rate possible, will be willing to consume an entire wavelength for a service terminating through a local network using only a few hundred megabits per second.

Not only will carriers need photonic optical cross-connects to switch and route wavelengths dynamically, but they will need to implement a method to provide the transition between lower-speed transparent networks and high-speed fixed bit-rate or opaque networks.

Devices known as optical gateways will emerge to perform this function by aggregating lower-speed wavelengths from access WDM devices and efficiently grooming them onto a high-speed, fixed-rate wavelength for the long-distance network (Figure 4).

Coming to a network near you Network management arguably may be the most severe barrier to the end-to-end wavelength service. After more than 12 years, Sonet interoperability and management finally are reaching a comfortable level of maturity.Few op erators have the expertise to operate WDM systems in local networks, and few standards exist on how best to manage a wavelength circuit through the network.

In the near term, vendors will supply proprietary management systems to solve these problems, but agreements within the industry must be reached regarding the type of information to be monitored, the technique used to create the optical service channels and the way these channels are monitored and controlled with large operator management systems such as Bellcore's NMA and OPS/INE standards.

Newer competitive LECs may be able to implement these services sooner because they are not encumbered by large legacy management systems.

Despite these hurdles, expect wavelength services to begin to appear in limited local high-end business enterprise applications during the next 12 to 18 months. But on the five-year horizon, metropolitan ring-based WDM products with direct data interfaces will be deployed to offer a full range of wavelength services. They will include optical gateways to manage the transition between lower-speed, highly transparent local networks and high-speed, fixed bit-rate or opaque backbone networks. They also will feature photonic-based optical cross-connects to switch and route wavelength services, backbone optical layer bidirectional line switched products and the necessary management systems.

The final goal As the optical layer emerges to support a data-oriented network's demands through wavelength-based service offerings, carriers will require new kinds of network elements. In the access networks, service providers will need low-cost enterprise multiplexers with direct wavelength interfaces to provide low-cost connectivity to the end user.

These elements will support variable rate optical interfaces capable of accepting a broad range of data rates from 100 Mb/s to 2.5 Gb/s, and each wavelength will be path-protected. Expect integration of Sonet functionality such as limited virtual tributary or synchronous transport signal multiplex capability, as well as Sonet path performance monitoring.

In the backbone network, new optical add/drop multiplexers (ADMs) will appear supporting wavelength level add/drop functionality and optical layer four-fiber bidirectional line switched rings. In addition, the number of wavelengths will increase from 40 to more than 100 and the bit-rate per channel will increase from 2.5 Gb/s to 10 Gb/s and beyond.

Optical ADMs capable of managing several terabits of bandwidth are viable within five years.

Where backbone and access networks interconnect, optical gateways and photonic cross-connects will emerge. Optical gateways will displace the broadband cross-connect to manage the transparent-to-opaque conversion between the access and long-haul networks, and to manage the broad range of payloads in wavelength services. They will simultaneously support cell-based routing and aggregation of ATM or IP payloads and legacy synchronous transfer mode services.

In this way, lower-rate wavelengths from the access will be aggregated to high-speed, ITU-compliant channels for the backbone network.

Finally, carriers will use optical cross-connects to provide bulk restoration and wavelength grooming, switching and routing to maximize the all-optical network's economy and efficiency.

The goal is to take full advantage of the emerging optical network to enable high-revenue, end-to-end wavelength-based services. Many other telecom market segments are converging to this end as ATM switches and IP routers push into the public network with direct optical interfaces, and as large core network products engulf Sonet interfaces and multiple switch fabrics to handle a range of services.

Regardless of which protocol camp wins the war in the service arena, a simpler network will emerge, streamlined to handle the new network where data is king.

The all-optical network is not just a marketing strategy dreamed up by manufacturers trying to sell more equipment, but a solution to meet the demands of tomorrow's end user. To achieve their goals, manufacturers will have to commit to extensive technology research and development; carriers will have to adopt new, more flexible management systems for network operation and service billing practices; and both will have to converge on a new class of network interface standards.