ADVANCED OPTICAL FIBER FOR LONG DISTANCE TELECOMMUNICATION NETWORKS

Transcription

1 Presented at AMTC 2000 ADVANCED OPTICAL FIBER FOR LONG DISTANCE TELECOMMUNICATION NETWORKS Christopher Towery North American Market Development Manager & E. Alan Dowdell European Market Development Manager Corning Incorporated One Riverfront Plaza Corning, NY USA Page 1 of 10

2 I. Introduction Deregulation of the telecommunications industry throughout the world, combined with an astronomical increase in demand for data services, has lead to a dramatic growth in the telecommunications industry worldwide. This growth is manifest not only in the sheer number of telecommunications companies and the route kilometers of their networks, but also in the amount of data they can carry. This rapid growth has also lead to fierce competition. Carriers are now struggling to build and maintain networks that will provide a flexible platform from which to offer a myriad of services at the lowest cost possible. The cabled fiber infrastructure of any major long distance telecommunications network is the foundation upon which a carrier will build its business. Consequently, when faced with the decision of building or expanding a network, a network engineer must take into account several key decision factors. One of these decisions is choosing which optical fiber type to deploy. He must be careful to balance the requirements of today with the potential technologies that may be available tomorrow. Consequently, a network planner must not only consider the multiple optical fiber types currently available, but he or she must also have an eye on the natural evolution of network technologies and how such technologies may be impacted by the optical fiber which has been deployed. Although today s options offer a variety of performance and economic advantages, depending on the configuration of the network, newer technologies may promise even more efficient and lower cost transport. II. Types of Single-Mode Optical Fiber There are three major classifications of single-mode optical fiber in use in today s telecommunications networks. They are specified by the International Telecommunication Union (ITU) as G.652 conventional single-mode fiber (SMF), G.653 dispersion-shifted fiber (DSF) and G.655 non-zero dispersion-shifted fiber (NZ-DSF). The primary way to differentiate between fiber types is by their chromatic dispersion. Chromatic dispersion is the linear phenomenon that causes different wavelengths of light to travel at slightly different speeds, leading to pulse expansion. As Figure 1 depicts, SMF was designed to operate at 1310nm with zero dispersion, DSF was designed to operate at 1550nm with zero dispersion, and NZ-DSF was designed to operate across a window of wavelengths near 1550nm with a small amount of dispersion. Dispersion can limit transmission speeds and distance, but a small amount of dispersion is required to reduce some detrimental non-linear effects such as four-wave mixing and cross-phase modulation. Page 2 of 10

3 Dispersion in Picoseconds per Kilometer SMF S-band C-band L-band + NZ-DSF DSF Wavelength in Nanometers - NZ-DSF Figure 1. Chromatic dispersion curves for SMF, DSF and NZ-DSF. In addition to chromatic dispersion, other critical parameters of optical fiber that impact network performance are attenuation, polarization-mode dispersion (PMD) and effective area (A eff ). PMD occurs when the two orthogonal polarizations of an optical pulse travel with different velocities. The A eff of a fiber is loosely defined as the light-carrying region of the fiber. It is a critical characteristic for determining the amount of optical power that can be launched into a fiber before non-linear effects limit the transmission speed and the distance, and where optical-toelectrical-to-optical (OEO) regeneration must take place. III. Fiber Deployment Trends The fiber cables that were deployed throughout the world until the early 1990 s were primarily low fiber count SMF cables. At the time, most of these networks were designed to carry a single optical channel at 1310 nm. They operated at relatively low bit rates and extended up to a maximum of 100 kilometers before OEO regeneration was required. However, carriers that wanted to extend the distance between OEO regenerators looked to transmission at 1550 nm and its accompanying lower attenuation when compared to 1310 nm. The use of 1550 nm as a transmission wavelength resulted in some carriers choosing to deploy DSF fiber with its near zero dispersion at 1550 nm. By the mid-1990 s several trends began to emerge as network builders rushed to meet the demand for cost-effective bandwidth. First, long-haul carriers had begun wide deployment of dense wave division multiplexing (DWDM) equipment which, when coupled with optical amplifiers, allowed multiple signals to be carried over long distances before costly OEO regeneration. It is the widespread use of DWDM and optical amplifiers that has given rise to carriers using NZ-DSF. A quick scan of the many recent press releases for such large network builds as Level 3, i21, 360networks, COLT, Williams, Qwest, and AT&T reveals that NZ-DSF has become the fiber of choice for deployment in new long distance network builds and Page 3 of 10

4 expansions. Another interesting trend with these new network builds is the dramatic increase in fiber count. Whereas only a few years ago 48 fibers in a cable was considered sufficient, many of the new networks are deploying cables with fiber counts in excess of 288 fibers. Coupling these high fiber counts with the large number of new network builds and expansions, it becomes evident that the vast majority of fiber now installed in national long haul networks is NZ-DSF. IV. Comparisons between SMF and NZ-DSF The primary reasons why carriers are choosing to install NZ-DSF in their long haul backbone networks are both technical and economic. NZ-DSF, due to lower dispersion in the 1550 nm window than SMF, has only one quarter of the total dispersion compensation requirements of SMF for a network to operate at its optimal level. Considering that most long haul networks are operating at OC-192 (10Gbps) data rates and require more precise dispersion compensation than OC-48 (2.5 Gbps), the cost savings of using an NZ-DSF fiber can be dramatic when compared to utilizing SMF. This is true even when an expected higher cable cost for NZ-DSF is factored in. Figure 2 shows the cost difference between lighting two pairs of SMF and two pairs of NZ-DSF to be in excess of 57% in a typical point-to-point network with a total reach of 450 kilometers and 90 kilometers between amplifiers. The reason for this differential is the added cost of bulk dispersion compensation for SMF at every amplifier site. Dispersion compensation for NZ-DSF occurs only at the transmitter/receiver sites. NZ-DSF reduced dispersion has other potential advantages when compared to SMF. Since NZ-DSF does not require dispersion compensation at every amplifier site, the mid-stage access port on a dual stage amplifier becomes available for use by an optical networking device such as cross-connects or add/drop multiplexers. In most SMF networks the use of a conventional dispersion compensation module at an amplifier site will preclude the use of any additional devices without purchasing or upgrading the amplifier at considerable cost. Page 4 of 10

5 Cost per Gbps-km (Normalized to 40ch C band LEAF) DCM Cable ch C Band LEAF fiber 40ch C Band SMF -28 fiber Figure 2. Cable and dispersion compensation cost comparison for SMF and NZ-DSF in a typical OC-192 network. V. Increasing the Capacity of Fiber There are a variety of ways to increase the data carrying capacity of a single fiber: using tighter spaced channels, increased TDM rates or using additional operating windows such as the Long (L) or Short (S) Bands. A typical network today will use transmission equipment that operates at OC-192 with 16 to 40 channels. These channels are spaced 100 Ghz apart in the conventional or C-band (1530 nm to 1565 nm). The use of tighter spaced channels at 50 Ghz would double the capacity of the fiber, but would also requires the use of more costly filters and lasers. Moving to the L-band (1565 nm to 1625 nm ) adds more useable bandwidth, but at the cost of additional amplifiers and dispersion compensation modules. These added components are more costly than their C-band equivalent because both require more fiber in their construction; erbium doped fiber in the case of the amplifier and dispersion compensating fiber in the case of the dispersion compensating module. Increasing TDM rates from OC-192 to OC-768 (40 Gbps) adds nearly 4 times the capacity to a network, however the technological hurdles required before such systems are commercially available maybe over a year out. Page 5 of 10

6 1.2 Normalized Cost/(Gbps/Km) Gbps 100 GHz C-Band 10 Gbps 50 GHz C-Band 10 Gbps 100 GHz C/L-Band 40 Gbps 100 Ghz C-Band Figure 3. Estimated cost comparisons per Gb/km for various TDM rates, channel spacing and amplification windows. As depicted in Figure 3, 100 Ghz spacing in the C-band is currently the most cost-effective manner to carry data. As long as a carrier has access to installed fiber it would be a rational decision to light up the next fiber using 100 Ghz C-band technology prior to moving to a more costly alternatives, at least in the short-term. However, future price declines or increases in demand may cause carriers to increase capacity on a single fiber beyond the 400 Gb/s which is currently available utilizing today s 40 channel OC-192, 100 Ghz spaced, C-band technology. As a result, and with the above upgrade paths in mind, it would be worthwhile to take a look at some of the enabling technologies that will facilitate an increase in the cost-effective capacity of future networks. VI. Raman Amplification The basis of Raman amplification is the nonlinear Stimulated Raman Scattering (SRS). SRS is the scattering of light by molecules, in which the scattered light is shifted from the incoming light by a frequency characteristic of the molecules. In a Raman amplifier, an optical carrier signal interacts coherently with the silica molecules of an optical fiber that has been excited by a high-power laser pump. This results in an amplification of the optical carrier. Since Raman amplification can reduce the amplifier noise present in a network, there are several potential advantages for its use. These advantages include longer distances between optical amplifiers, longer distance between OEO regenerators, tighter channel spacing or an enabling of higher data rates. Page 6 of 10

7 Lower Launch Power Power (db) Signals without Raman Signals with Raman Raman Pump Higher signal power at EDFA Higher signal power during Raman amplification Distance Figure 4. Counter-propagating Raman pumps allow for the use of lower launch powers, while increasing the signal strength arriving at the EDFA. Although Raman amplification has been explored in laboratories for more than a decade, it has been slow to reach widespread commercialization. Raman amplifiers have seen only limited use in such applications as un-repeatered submarine systems. Because high-power laser pumps are required, and for a variety of other technical reasons, Raman amplifiers have not been able to compete with erbium-doped fiber amplifiers (EDFAs). However, the development of counterpropagating Raman pumps has made for a much more efficient geometry than earlier Raman amplifiers utilizing co-propagating Raman pumps. A counter-propagating configuration is shown in Figure 4. In a co-propagating geometry (pumping from the beginning of the span), the growing signal power will quickly result in gain saturation. Consequently, a large increase in pump power is needed to produce a small increase in Raman gain. By comparison, in a counterpropagating geometry where the Raman pump is fed from the end of the span, i.e. where the signal power is much lower, Raman gain can build up much higher without saturating the amplifier. This allows the use of much less pump power for a higher gain. Raman amplifiers can be quite versatile because Raman gain can be generated in all types of fiber. SMF, DSF, and NZ-DSF can all act as the gain medium, albeit with different pump efficiencies. In addition, Raman amplifiers can allow for gain across a wide spectrum of wavelengths by simply varying the associated Raman pump wavelength. VII. Forward Error Correction Another technique that submarine networks have pioneered is Forward Error Correction (FEC). For example, the demanding requirements for submarine systems have necessitated the use of FEC for this market. These error correction algorithms add system margin to overcome degradation and push longer distances. As terrestrial networks continue to challenge transmission technologies, FEC will become more and more prominent there as well. FEC techniques operate by adding redundancy to the transmitted signal. Basically, the incoming signal is manipulated by a very complex algorithm that adds this redundancy in the form of parity bits. In the case that the received signal takes a hit and several errors are detected, the original signal can be reconstructed. Page 7 of 10

8 Reed-Solomon codes are a common set of codes used in error correction schemes. The encoder can take data symbols of a certain number of bits each and add information to a codeword [1]. A Reed-Solomon encoder can work in either a single stage, or it may be strengthened by concatenating and interleaving multiple codes as shown in Figure 5 [2]. data source Reed- Solomon encoder transmission channel Reed- Solomon decoder data sink Single Encoding Technique data source data sink Reed- Solomon encoder Reed- Solomon decoder interleaver De - interleaver transmission channel Concatenated Encoding Technique Figure 5. Graphical representations for single and concatenated coding schemes VIII. Duobinary Modulation Another potential enhancement to optical transmission systems is the use of alternative modulation formats. One scheme that shows promise is duobinary modulation. A simple duobinary coding scheme is shown in Figure 6 [3]. 1 bit delay duobinary code Decision 1 2 0, 1 1 Decision 2 0 0, 1 1 Or binary code Xor 1 bit delay Encoder Receiver Decoder Figure 6 - Simple duobinary coder The binary input to the encoder is converted to a three level duobinary code in which there are no transitions between the zero and the two level. The receiver contains two decision circuits to detect the difference between the 0 s and 1 s, and the difference between the 1 s and 2 s. Finally a decoder is used to re-constitute the original binary signal. Page 8 of 10

9 Because the encoder prevents transitions between the zero-level and the two-level, the duobinary format requires less bandwidth than the standard non-return to zero (NRZ) format. The following power spectrum shows the relative bandwidth requirements of binary (NRZ) formats versus duobinary. binary duobinary -2Rb -Rb 0 Figure 7. Power spectrum of binary and duobinary codes (excluding DC component) Rb = bit rate One significant advantage of duobinary encoding is that since the signal uses approximately onehalf the bandwidth, the dispersion tolerance of a duobinary system is three to four times longer than when NRZ formats are used [4]. Also, the narrower spectrum will also allow twice as many channels in dense wavelength division multiplexing (DWDM) systems to be packed into the same bandwidth. Rb 2Rb IX. Advanced Fiber Designs As carriers look to new amplifier and transmission technologies to increase the capacity and distances that they can transmit information, optical fiber types will evolve to add further value to a network. The design of optical fibers is a multi-dimensional balancing act which requires precise tuning of a fiber s optical characteristics to achieve a design that will maximize value when installed a network. As an example, Corning LEAF fiber was designed to provide a balance between a large effective area to minimize non-linear effects and a low dispersion across the 1550 nm region to reduce the costs associated with dispersion compensation. Although, it has been the case for submarine cable systems for many years, it is possible that within a few years terrestrial networks may begin to employ dispersion managed cables. The use of dispersion managed cable in a terrestrial network may be to facilitate very high TDM rates such as 80 Gb/s. Such a terrestrial dispersion managed cable system would likely utilize two or more different fibers distinctly optimized for different performance features. These systems differ from those utilizing dispersion compensation modules in that the compensation is accomplished in a distributed fashion as opposed to discrete placement of modules. Of course there are numerous, primarily logistical, issues that have so far precluded the use of dispersion managed cables terrestrially. However, there have been recent demonstrations of TDM rates up to 80 Gb/s utilizing dispersion managed cable with very real implications for terrestrial use. X. Summary Page 9 of 10

10 This paper has taken a brief look at the evolution of optical fiber used in today s terrestrial long haul telecommunications networks. NZ-DSF has become the fiber of choice for new network builds or expansions. This is primarily the result of economic and technical advantages over SMF which allow a carrier more flexibility with respect to future optical networking technologies, and reduced dispersion compensation requirements. Additionally, this paper addressed emerging technologies for terrestrial networks such as Raman amplification, FEC, and duobinary modulation. These advancements have the potential to increase transmission distances and the data carrying capacity of fiber. References: [1] Lin and Costello, Error Control Coding: Fundamentals and Applications, Prentice-Hall 1983 [2] Sab and Fang, Concatenated Forward Error Correction Schemes for Long-Haul DWDM Optical Transmission Systems, ECOC 99 [3] Taylor, Neal, Internal Technical Report Duobinary Coding, BICC Cables Limited. April [4] Katsaros et al, Experimental Demonstration of the Reduction of FWM with Duobinary Modulation, ECOC 99 AN7079 Page 10 of 10