Raman AmplificationLonger, Wider, Faster, Cheaper CIBC World Markets In an effort to keep pace with bandwidth demands that are nearly doubling each year, WDM system designers are moving to more cost-effective solutions utilizing higher channel counts, higher transmission speeds and longer distances between regeneration sites. All these features require more power, and optical amplification is as crucial as ever. Attracted by unique power and distance-boosting attributes, including superior gain flatness, flexible gain profile and low noise figures, designers are looking to Raman amplification as the next generation technology for boosting optical signals. For the most part complementary to erbium-doped fiber amps (EDFAs), Raman helps reduce overall network amplification costs by 40% or more, compared with traditional systems that are totally dependant on EDFAs. As a result, the Raman amplifier market could reach $4.8 billion by 2005, from $245 million in this year [see ]. For all its current popularity, diode-pumped Raman is made possible only through the recent commercial availability of high powered InP-based pump lasers operating in the 14001500nm wavelength range (hereafter referred to as 14XXnm). The market for 14XXnm pumps could explode over the next several years as Raman amp deployments aggressively ramp, jumping from just under $110 million in 2000 to $2.3 billion in 2005. Why Use Raman? As photons travel down an optical fiber, power and signal quality degrade. Power can be added to the signal with optical amplifiers, but at certain points in the network the signal must be regenerated. This is done by converting the optical signal into its electrical equivalent, restoring the signal back to its original quality, and then sending the signal back to a modulated laser for re-transmission along the fiber. Regenerators are typically placed every 400600 km, an expensive process sometimes accounting for up to 50% of the cost of an entire network. Optical-electrical-optical (O-E-O) regeneration sites can cost more than $50,000 per wavelength. In the case of a Nortel 160-channel WDM system, this equates to about $8 million per regeneration site at 2.5 Gb/s (and roughly twice as much at 10 Gb/s). Raman enables ultra-long-haul WDM transmission by greatly reducing a system s noise figure. By deploying Raman amps in their networks, long-haul operators can increase the distance between regeneration sites to 4,000 km or more, leading to lower costs and, most importantly, a more efficient optical network. Companies such as Corvis and Qtera developed ultra-long-haul WDM solutions that increased the distance between electrical regeneration sites to more than 3,000 km. Raman was one of two key elements of this accomplishment (the other was strong forward-error correction, or FEC). The potential savings are clear: for a 160-channel, 10 Gb/s (OC-192) network, the operator can replace more than six regeneration sites (cost $75 million) with roughly 40 Raman amps costing less than $1 million. As channel counts in WDM systems increase from 16 to 80 or more, an incremental increase in amplification is required for each additional wavelength. Similarly, higher transmission speeds also require more amplification. When the photon stream is modulated at a higher data rate, each bit of data contains fewer photons. For a given photodetector to generate the same electrical signal at the higher data rate, the signal needs to be more powerful. The key is to do this without introducing signal distortions and crosstalk, which is where Raman comes in. RamanThe Basics Stimulated Raman scattering occurs when light waves interact with atomic vibrations in a crystalline lattice (or an optical fiber). The atom absorbs the light and then quickly re-emits a photon with energy equal to the original photon energy plus/minus the atomic vibrational energy. Raman amplification occurs when higher energy (shorter wavelength) pump photons interact with the atomic vibrational modes in the optical fiber, thereby coherently "stimulating" or adding energy to the lower energy (longer wavelength) signal photons. One major advantage of Raman is the ability to amplify over a broad spectral range, i.e. beyond the standard 1550 nm long-haul transmission window [see ]. This makes Raman potentially attractive for short-haul amplification, where 1310 nm transmission is common. Amplification of 1550 nm signals requires pump lasers operating at around 1450 nm, while 1310 nm amplification requires a 1240 nm pump source. Importantly, Raman can be used in any type of optical fiber, including the large installed base that is not optimized for WDM. Because the power conversion efficiency of Raman amps averages 25%, compared to 80% for EDFAs, a high pump energy is required. Normally, Raman requires output (launch) power in the 1W range, although some newer systems get adequate gain from 500750 mW of output. Types of Raman Amplifiers There are two fundamental categories of Raman amplifiers: discrete Raman is similar to an EDFA, where gain occurs in a section of specialty fiber pumped by a 14XXnm laser, while distributed Raman amplifiers rely on the transmission fiber itself to achieve gain. In distributed Raman, the pump power is launched into the transmission fiber at the end of its span, in the opposite direction to the transmission signal: power is transferred from the pump photons to the transmission signal via the atomic vibrations in the optical fiber. The distributed category is further broken down into cascaded, where a fiber laser provides the pumping, or diode-based, which uses 14XXnm diode laser pumps. Diode based distributed Raman provides launch powers typically in the 500 mW1 W range, and will come to dominate the terrestrial market as ultra-long-haul, high-channel-count, high-speed system deployments explode in 2001. Sales of diode-based distributed Raman could jump from about $220 million in 2000 to roughly $4.5 billion in 2005, or 94% of the total Raman market. Cascaded Raman using fiber pump lasers is typically utilized for undersea and remote amplification given its relatively high launch powers of 1.52 W. Cascaded Raman solutions should grow from $25 million in 2000 to $243 million in 2005 as undersea WDM system deployments proliferate. In the long term, Raman amplification is likely to be a complement to, not a replacement for, traditional 980 nm/ 1480 nm EDFA amplification. However, in some new systems operating at 10 Gb/s speeds or less, where the amps are 50 km apart, it is both possible and desirable to obviate the need for EDFAs provided the Raman power is high enough (20 dB). As transmission speeds migrate to 40 Gb/s and higher, we believe Raman will likely be used in conjunction with EDFAs in all systems, including those with reduced hut spacings. Distributed Raman Distributed Raman amplification reduces the gain required of an EDFA while providing a dramatic improvement in the overall noise figure. This translates to a six-fold or more improvement in distance traveled between O-E-O regeneration sites. There are two major types of noise associated with fiber-optic transmission: amplifier noise, and noise due to signal attenuation in the fiber. In a system amplified exclusively by EDFAs, the signal attenuates after leaving the first amplification site, and reaches the next EDFA greatly weakened. The EDFA pounds this weakened signal with 30 dB-plus of pump power, introducing a disproportionate amount of noise into the signal as it is amplified. Thus, the signal-to-noise ratio (SNR) continues to fall span after span. After roughly 400600 km, the signal needs to be cleaned up i.e. regenerated. Distributed Raman helps to reduce the noise figure, resulting in a much-improved SNR. Simply stated, the noise figure is improved because the signal is never allowed to become as weak between spans as in an EDFA-only architecture (see ). By maintaining a stronger signal at amplification (where the majority of noise is introduced), the overall SNR stays higher longer. Raman is proving to be particularly attractive for 40 Gb/s systems. An increase in speed from 10 Gb/s to 40 Gb/s using normal EDFA amplification requires four times the amount of launched signal power, which can distort signals and induce channel crosstalk. Raman allows designers to reduce the launch power of the 1550 nm signal, therefore reducing non-linearities. In a typical one-one configuration, a Raman amp is used as a "pre-amp" for each EDFA in a given span, as shown in . Diode-Based Distributed Raman As many as eight 14XXnm pumps are used per diode-based distributed Raman amplifier, for several reasons. First, high-powered 14XXnm pumps are still relatively difficult to find in volume (and very expensive, once a supply can be secured). The second issue is reliabilitythe more pumps an amp has, the better it will be protected from the failure of any single laser. Third, the unique properties of InP-based 14XXnm pumps allow system architects to design Raman amps with pump lasers of varying wavelength, enabling a very flat gain shape without the need for lossy gain-flattening filters (GFFs). However, amps begin to get too complex beyond eight pumps, and higher-powered pumps will become more readily available in the future, keeping the average number of pumps per Raman amp at around five or six. shows a schematic of a typical 1 W diode-pumped distributed Raman amplifier containing eight pump lasers of varying wavelengths, each with an output power of 125 mW. This example contains two polarization beam combiners (PBCs), which are passive optical components that combine the output of two lasers with the same wavelength but with orthogonal polarization modes. The amplifier also contains a multiplexer to combine the different pump laser wavelengths. After each amplification, if gain flatness is not tightly controlled across the entire band (all 40 or 80 channels), the power of each channel can eventually vary widely introducing non-linear effects in higher power channels or signal degradation in lower powered channels. As WDM networks increasingly migrate to 10 Gb/s modulation rates and higher, power levels must be managed effectively. In the EDFA-only model, both gain flattening and dispersion compensation are needed immediately after signals are amplified. Aside from added equipment cost, gain flattening filters (GFFs) represent an additional cost in terms of efficiency by introducing another point of loss into the system. Some Raman amplifier designs do not require discrete GFFs, although others do employ these components. Cascaded (Fiber Laser Pumped) Raman Cascaded Raman amplification uses several fiber lasers with a combined output in excess of 1.5 W to pump the transmission fiber. In turn the fiber laser is pumped using high power, multimode diode lasers. SDL, the dominant player in this market, couples ten multi-mode one-watt 920 nm pump lasers together into a co-doped erbium ytterbium (Er/Yb) fiber. The combined output passes through a special FBG-based resonating cavity that shifts the final launched wavelength to the desired value (typically 1455 nm or 1480 nm) for Raman amplification [see CS 6(7), p.66]. The availability of more powerful (2 W) 920 nm lasers could reduce the laser count to five per fiber laser, bringing down costs to achieve mass commercialization. Cascaded Raman amps are well suited to remote pumping applications in the submarine (undersea) market. In a festooned WDM architecture, a fiber-optic cable is dropped just off the coast of a continent or island. Every 400600 km, the cable is looped back to the shore into huts containing network elements like optical add-drop muxes (OADMs) and amplifiers. The benefit of this design is that no active components are located underwater. This helps reduce system costs and improve network reliability. Discrete Raman and the Metro Opportunity Discrete Raman is simply a stand-alone amplifier, similar to an EDFA in that the signal is piped into a discrete box and travels through a length of specialty fiber, where gain is achieved. This technology, yet to catch on commercially, is somewhat limited in its potential application. System designers will most likely deploy discrete Raman to overcome localized loss, which could occur in switches, OADMs, or in dispersion compensating fiber. We expect the market for discrete Raman could ramp from less than $1 million this year to $60 million in 2005. One potential application for discrete Raman is to provide low-cost optical amplifiers, or "amplets," for the metro WDM market, where there are many potential sites for localized optical loss. Several other technologies, including scaled-down EDFA designs and semiconductor optical amplifiers (SOAs), will also compete. These amplets will sit on the outputs of switches and OADMs to rebalance powers, and the Raman and EDFA versions will be pumped by low-cost, low power 980 nm and 14XXnm pump lasers. Pump Laser Diodes InP-based 1480 nm and 14XXnm pumps met with early acceptance (for erbium-based and Raman amplification, respectively), because they offered a high degree of reliability relative to early generation GaAs-based 980 nm pumps. InGaAsP/InP 14XXnm lasers tend not to fail suddenly, but to degrade slowly and predictably over time, while 980 nm lasers failed frequently, suddenly, and unpredictably due to mirror degradation. However, the introduction of patented 980 nm mirror passivation processes at JDS Uniphase and SDL led to vastly improved 980 nm reliability, with both companies producing 980 s with failure in time (FIT) scores near 50, equating to a mean time to failure of 100-plus years. Because 980 nm pumps were inherently more powerful (and introduced less noise when used in the first stage of EDFAs), Raman and 14XXnm pumps were pushed to the back burner. Raman amplification appears ready to blossom as supply of 14XXnm pumps has improved greatly over the past 12 months, led by ramping production at leading manufacturer Furukawa as well as Alcatel, Sumitomo and JDS Uniphase. Furukawa currently holds up to 90% of the capacity-constrained market for 14XXnm pumps, followed by Sumitomo, Alcatel, JDS Uniphase, Lucent, SDL, Corning, and Anritsu. Manufacturers are introducing modules with output powers exceeding 200 mW, and the introduction of 200250 mW modules (containing 300 mW 14XXnm pump chips) will ensure high barriers to entry in this market. Packaging to dissipate heat is a big challenge; Furukawa has a novel solution to this challenge, using a "heat pipe" with exceptional thermal conductivity. As shown in , the market for 14XXnm pump lasers is expected to grow at a CAGR of around 85%, reaching $2.34 billion in 2005. This assumes an average of 6 pumps per diode-pumped Raman amplifier. With average selling prices expected to decline from $1800 in 2000 to $1100 in 2004/5, unit sales of 14XXnm pumps will actually increase at a CAGR of over 100%, from around 60,000 in 2000 to over 2.1 million in 2005.