Tunable lasers enable novel network architectures with dynamic functionality.
TIMOTHY BODENHAMER and JOSEPH K. LEE, New Focus Inc.
WDM technology, featuring high-channel-count capability, is a reliable and economical means to scale telecommunications network capacity to meet ever-increasing demands. However, the sole use of fixed-wavelength lasers results in logistical and inventory burdens as channel counts increase. With fixed-wavelength lasers, each channel requires a unique line card and part number. From optical systems manufacturers to carriers and service providers, these logistical issues make internal planning and procurement efforts increasingly difficult to manage.
Continuity of service standards requires that a spare line card for each channel be kept onsite as a backup in the event of failure, resulting in hundreds of idle spare line cards in inventory. Considering the cost of a line card together with the backup or "spare" requirement at system nodes and regeneration stations across the network, service-provider inventory costs potentially run into the tens of millions of dollars, if using fixed-wavelength lasers only.A simple economic analysis shows how expensive this inventory can be. The cost of a single line card varies from $10,000 to $30,000 today, depending on data rate. In a fully deployed 80-channel system with as few as five regeneration points, total line-card deployment easily exceeds 500 lasers. With one fixed-wavelength backup card for each one deployed, spare inventory cost totals between $10 million and $30 million at today's line-card prices. In addition, the operational cost of maintaining physical inventory (i.e., facilities, insurance, time value of money, etc.) inflates this number up to an additional 30%.
In comparison, deploying widely tunable lasers on line cards can readily reduce inventory levels by an order of magnitude or more, from hundreds of spares to dozens or less. The development of compact, high-power, widely tunable lasers can dramatically reduce network operating costs, because a single, widely tunable "universal" line card is capable of being tuned to as many as 100 channels (see Figure 1). Part-number counts (SKUs) can be reduced by a factor equivalent to the number of channels a particular transmitter is capable of tuning to-the wider a tuning range, the greater the savings.
Service providers deploying tunable line cards maximize purchasing economies, assuring themselves a low-cost strategy. In an 80-channel network system, use of tunable lasers can reduce the number of distinct line-card parts from 80 down to one. That amounts to significant savings, since the cost of creating and maintaining a single part number can range from $10,000 to as much as $100,000 annually.
Tunable lasers provide an advantage over fixed sources, even when service providers use an alternative approach to sparing-that is, a "hot-backup" strategy of maintaining idle channels and activating them as required in a backup role. In this hot-backup application, up to 50% of the system bandwidth can be rendered unusable with fixed-wavelength lasers, because network carriers must maintain a spare channel for each channel deployed using a fixed-wavelength laser. With tunable lasers, however, only a small number of line cards are held in reserve slots. Thus, tunable lasers can restore usable system bandwidth to 90% or better.
Moreover, in the event of a channel failure, a backup card can be quickly and remotely configured to resume communications. Deploying tunable lasers provides near-seamless restoration in the event of a malfunction, allowing SONET protection to be implemented entirely in the optical domain. As a further guarantee of service continuity, tunable lasers therefore present additional opportunities for revenue generation.
Wide tunability also allows carriers to allocate bandwidth flexibly on an as-needed and where-needed basis-properties increasingly required by both network planners and end users. The ability to deliver "flexible provisioning" will simplify the process of planning and forecasting network growth while allowing the optimal allocation of available resources. For example, tunable lasers simplify channel-management issues. If a needed channel or wavelength is already in use along a particular route or is simply not available, widely tunable lasers can perform wavelength conversion, providing a nonblocking pathway and continuity of service.
Similarly, if network traffic requirements do not mandate that system capacity be fully populated along a particular route, the use of tunable lasers allows all available lasers to be used in times of peak capacity demand, regardless of individual channel demand. In this case, tunable line cards provide a simplified expansion strategy. Rather than continuing to populate additional channels with fixed-wavelength transmitters, tunable transmitters can be configured to any needed channel, thereby assuring the availability of additional capacity under all circumstances. With tunable lasers, capacity planning is simplified to forecasting overall demand and allocating the appropriate number of channels to provide required capacity.
In the same way, service providers can "provision" new services or capacity as usage and demand conditions change. For example, the current service or capacity provisioning model requires that network operators "roll trucks" carrying technicians to manually reconfigure the network to provision service for a new customer. Each truck roll costs in excess of $500 per field call. Tunable lasers reduce this practice to a remote point-and-click operation performed at the central office, thereby reducing activation times to minutes or even less. For service providers, the operational cost savings are not only enormous, but also rapid provisioning is a value-added service enhancement that provides significant competitive advantage, affording carriers additional opportunity to in crease customer satisfaction, loyalty, and top-line returns.Next-generation capabilitiesNext-generation services and functionality will likely be implemented initially in a new generation of optical add/drop modules (OADMs) and optical crossconnects (OXCs) (see Figure 2). Current-generation OADMs are limited in their ability to add channels to the network by the nature of fixed-wavelength transmitters deployed. Changing traffic patterns, customer requirements, and new revenue opportunities require greater flexibility than static OADMs provide, complicating network operations and planning. Tunable lasers remove this constraint by allowing any channel to be added at any time by the OADM. With the deployment of tunable transmitters at OADM sites, sparing and restoration capabilities follow economically, as well.
OXCs represent another opportunity for tunable lasers to improve network system efficiency. Widely tunable transmitters covering the full C-band enable the desirable nonblocking property, allowing an any-channel-to-any-channel connection at network node or crossing points. Widely tunable lasers simplify OXC planning, since all wavelength channels are available from any tunable line card. Even if a particular channel is already in use, tunable lasers allow a flexible, potentially automated determination of wavelength conversion, further simplifying system planning.
While today's networks are large and complex, future purely optical networks are expected to be greatly simplified and passive in nature, significantly reducing both capital and operating costs. One scenario is a "mesh" architecture in which nodal points on the network are designed to route signals on the basis of wavelength (see Figure 3). Such a network has the potential for great savings in capital outlay for service providers, while offering much lower operating costs. In this example, tunable lasers can be deployed to route signals to their destination on the basis of wavelength. This type of network architecture design is unlikely to become reality with fixed-wavelength or even narrowly tunable lasers and line cards.Performance requirementsWide tunability is not the only advantage of currently available tunable lasers. To successfully replace fixed-wavelength distributed-feedback (DFB) lasers-the standard technology employed in current network systems-tunable lasers must offer a range of improved specifications, including higher output power, superior spectral purity, and excellent wavelength stability.
These unique properties of tunable lasers will enable next-generation network enhancements, including higher data rates, higher channel counts (in creased channel density), longer span lengths for long-haul (LH) and ultra-long-haul (ULH) networks, greater functionality in the optical domain (thereby fewer expensive optical-electrical-optical conversions), and increasing complexity for MANs as population density and demand for bandwidth continues to grow.High-power tunable lasers not only allow LH and ULH systems makers to deploy longer span distances, but also mitigate an increasingly complex and therefore "lossy" environment in metro applications. Higher output powers (>10 dBm) enable network architects to in crease the number of system elements by increasing the total available power budget. Higher output powers also support higher data rates (OC-196, OC-768) by making more power available into the data stream. In combination with high power, enhanced spectral purity can reduce system costs by better enabling ULH systems and more complex architectures while reducing the amount of costly regeneration required in the transmission process.
Tunable lasers also deliver clean, narrow linewidths that remain stable over time and temperature. To increase capacity economically, some systems developers are beginning to use coherent modulation techniques to pack an increasing amount of information on the fiber. It is important that the laser deliver narrow linewidths with high wavelength accuracy and stability for increased channel densities, excellent spectral purity, side-mode suppression ratio (SMSR), and high polarization extinction ratio. High SMSRs significantly reduce crosstalk in high-channel-count systems, especially in networks based on periodic multiplexing architectures that lack the input selectivity of fiber Bragg grating, thin-film, and other narrowband technology. Indeed, the improved optical performance specifications of newly available tunable lasers provide many distinct advantages for next-generation systems developers.Approaches to tunabilityWith so much interest in laser tunability, many different technologies are vying to become the choice of future optical networks. Currently there are four primary approaches to providing tunability with semiconductor lasers1-each with distinct advantages and disadvantages-as well as some new developments such as quantum dots and two-laser pumping. Approaches include DFB lasers, distributed Bragg reflector (DBR) lasers, ver tical-cavity surface-emitting lasers (VCSELs) employing micro-electro mechanical systems (MEMS) technology, and external-cavity diode lasers (ECLs). The ultimate tunable-laser solution will supply high output powers over wide tuning ranges in a compact and reliable package, with a proven and scalable manufacturing plan.
DFB lasers. Tunable DFB lasers use temperature to tune the wavelength and are a variant of the widely deployed fixed-wavelength DFB lasers. Manufac turers of DFB lasers have achieved a limited tuning range of 2 nm with output power of 13 dBm.2 Ad ditional gain sections on the same semiconductor extend the tuning range to 15 nm at the expense of output power.3 Although the highly reliable fixed-wavelength DFB laser is a well-proven design, the higher diode temperatures of its narrowly tunable cousin, necessary for temperature tuning, raise well-placed concerns about expected lifetime.
DBR lasers. Tunable DBR lasers, on the other hand, extend the available tuning range beyond that of a tunable DFB laser by using additional functional elements on the same semiconductor. Several varieties of DBR lasers have been demonstrated with wide tuning, such as sampled grating distributed Bragg reflector (SG-DBR) lasers, super-structure grating distributed Bragg reflector (SSG-DBR) lasers, and grating-assisted co-directional with sampled reflector (GCSR) lasers. SG-DBR and SSG-DBR lasers use multiple Bragg reflectors with a phase section to achieve tuning ranges of 8-40 nm.4 However, output power of the SG-DBR has been reported only in the 3- to 8-dBm range.5 The SSG-DBR design offers uniform power output in excess of 35 nm, but power output is reported to be only 0 dBm. GCSR-DBR lasers consist of a gain block followed by coupler, phase, and reflector sections and have been demonstrated with a 40-nm tuning range, but with only 3 dBm of output power.6
VCSELs. The third tunable-laser approach uses a VCSEL with MEMS technology to manipulate a movable top mirror. Tuning ranges up to 35 nm have been demonstrated with tunable VCSELs.7 Optically pumped tunable VCSELs have been demonstrated with decent 6-dBm output power,8 while electrically pumped versions have been shown with nearly -3 dBm.9 Although this technology is fairly recent, tunable VCSEL manufacturers hope to capitalize on its inherent design potential for low-cost, high-volume manufacturability.
ECLs. The fourth approach to tunability also uses an external cavity to tune the laser. A new approach to ECLs, however, promises to provide both wider tuning ranges (~100 nm) with very high output powers.10 These ECL designs provide very narrow line widths and high-frequency stability-and in particular, improved immunity to temperature sensitivity compared with DFB lasers. Moreover, the utility and reliability of ECLs are proven in more than 1 million hours in test and measurement applications.
In a new version of ECL tunable-laser design, some portion of the laser cavity resides off the laser chip, and the diode acts as a gain medium only. In some ECL source lasers, one facet of the laser diode is coated to act as a high reflector, while the laser output and cavity tuning are provided on the other side of the chip. Littman-Metcalf and Littrow cavity configurations are examples of such single-sided ECL sources, in which gratings are used to provide optical feedback and to tune wavelength. This particular laser design is used extensively in test and measurement lasers, is widely used to manufacture fiber-optic components, and has been proven time and again in demanding operational applications. The lasers offer 24/7 reliability and are routinely tested to withstand harsh conditions, including shock and humidity.Fourth-generation ECL designAlthough this ECL design used in test and measurement tunable lasers has proven extremely robust in actual-use conditions, next-generation design used in network tunable lasers offers even more flexibility, because it is a two-sided ECL configuration (see Figure 4). In this implementation, one facet of the laser diode, uncoated or coated to be a partial reflector, acts as an output coupler. Light transmitted from the output coupling facet is collimated, passed through a Faraday isolator, and focused into an output fiber. The other side of the laser diode is antireflection-coated to minimize optical feedback, and a lens is positioned to couple the light into the external cavity. Various optical tuning elements may be used in the external cavity, including gratings and etalons, and by proper choice of cavity configuration and parameters, the laser can be made to emit light at discrete International Telecommunication Union wavelengths.
The resulting tunable lasers from this innovative design approach provide excellent optical output for DWDM transmission in optical networks. For example, with such an ECL laser, high output power (>13 dBm) is achieved over the entire C-band. Other advantages of ECLs include an SMSR substantially better than -50 dB, excellent wavelength accuracy, and narrow intrinsic linewidth. In addition, this ECL tunable laser employs inexpensive Fabry-Perot laser diodes that exhibit long life and stable, proven performance characteristics.
Yet another advantage of ECLs is that while most semiconductor tunable lasers face significant challenges in reaching wide tunability beyond the C-band to the L-band because of material and device limitations, the ECL design allows for a greatly simplified transition to L-band operation. Because the laser gain medium is a common Fabry-Perot device, only minor wavelength range adjustments to optical coatings and tuning elements is required for its use in L-band operation.
Indeed, the ECL design provides tunability, high output power, superior spectral purity, and excellent wavelength stability, while offering the proven reliability of existing ECL tunable lasers in laboratories and manufacturing plants worldwide. Yet, the network-deployable ECL must meet the more stringent requirements laid out in the Telcordia GR-468-CORE; the European BS EN: 61751:1998/IEC, CEI IEC 749, and CEI IEC 68; and the Japanese JIS C 5944 and JIS C 7021. Thus, early during the development process, reliability and testing efforts are critical, designing robustness into the product from the start.
Without question, widely tunable lasers are substantially more complex than their fixed-wavelength counterparts, increasing the range of components within a hermetic package. To ensure that the laser will meet the 20-year-plus lifetime requirements of telecommunications customers, reliability engineering efforts must begin with material selection where robustness, quality, performance, and outgassing characteristics are integral criteria in the selection process and thoroughly evaluated. Component and subassembly testing follows, selectively applying Telcordia GR-468 tests. Finally, after testing at all levels throughout the development cycle, network tunable lasers must undergo the standard Telcordia GR-468 testing as a full laser subsystem.
Recent advances in tunable-laser technology have brought the promise of tunable networks into reality. Widespread adoption of such tunable lasers will not only eliminate the logistical and inventory problems resulting from fixed-wavelength line cards-reducing operational costs for system manufacturers and telecom carriers-it will also enable novel network architectures with dynamic functionality. The result will be new value-added services and new sources of top-line revenues to system providers.
Tunable lasers will ultimately bring next-generation network functions such as dynamic add/drop closer to fruition, while they have proven their usage in routing optical signals through passive wavelength-selective mesh networks. Tunable lasers will play a central role in next-generation transport architectures. It is widely expected that network intelligence and functionality will eventually migrate to the optical layer in next-generation systems so that crossconnect, dynamic add/drop multiplexing, switching, and routing functions become increasingly performed in the optical domain.
Regardless of the underlying technology, a compelling tunable-laser solution must be able to provide high output powers over the full C-band. In addition, tunable lasers must meet the stringent performance and reliability requirements as outlined in Telcordia GR-468-CORE, BS EN: 61751:1998/
IEC, CEI IEC 749, CEI IEC 68, and other reliability documents. The challenge for component manufacturers is to develop a winning solution that not only meets the specifications of telecom systems companies, but that can also be produced in high volume with economic yields.
The new generation of ECL tunable laser combines a robust design, ad vanced product specifications, and proven manufacturing expertise to provide carriers and service providers with the confidence to operationalize a new generation of tunable line cards and networks that reduce operating costs and increase customer satisfaction, loyalty, and revenues.
Joseph K. Lee is vice president of product marketing-actives and Timothy Bodenhamer is product manager at New Focus Inc. (San Jose, CA). They can be reached via the company's Website, www.newfocus.com.
- "Tune In!" April 19, 2000, www.Lightreading.com.
- "Alcatel Optronics Division Delivers First Tunable, Wavelength-locked Laser", Aug. 9, 2000, Alcatel Optronics, www.alcatel.com.
- "Nortel Networks Announces World Leading Technology with 15nm Tunable Laser", Feb. 22, 1999, Nortel Networks, www.nortelnetworks.com.
- "Directly modulated sampled grating DBR lasers for long-haul WDM communications systems," B. Mason, S.L. Lee, M.E. Heimbuch, and L.A. Coldren, IEEE Photonics Technology Letters, Vol. 9, pp. 377-379, 1997.
- "Design of Sampled Grating DBR Lasers with Integrated Semiconductor Optical Amplifiers," B. Mason, J. Barton, G.A. Fish, L.A. Coldren, Fellow, IEEE, S.P. DenBaars, IEEE Photonics Technology Letters, Vol. 12, No. 7, July 2000.
- Altitun NYW-60, October 2000, www.altitun.com.
- "Error-free 2.5 Gb/s transmission of a C-band tunable 8-dBm vertical-cavity surface-emitting laser over 815 km of conventional fiber," M. Jiang, C-C Lu, P. Chen, J.H. Zhou, J. Cai, K. McCallion, K.J. Knopp, P.D. Wang, M. Azimi, P. Tayebati, D. Vakshoori, presented at NFOEC 2000, Aug. 27-31, 2000.
- "Top-emitting micromechanical VCSEL with a 31.6-nm tuning range," M.Y. Li, W. Yuen, G.S. Li, and C.J. Chang-Hasnain, IEEE Photonics Technology Letters, Vol. 10, pp. 18-20, 1998.
- "High-Performance 1.6-