Click for next page ( 65


The National Academies | 500 Fifth St. N.W. | Washington, D.C. 20001
Copyright © National Academy of Sciences. All rights reserved.
Terms of Use and Privacy Statement



Below are the first 10 and last 10 pages of uncorrected machine-read text (when available) of this chapter, followed by the top 30 algorithmically extracted key phrases from the chapter as a whole.
Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.

OCR for page 64
3 Communications, Information Processing, and Data Storage INTRODUCTION Optics has become the way by which most information is sent over nearly all the distance that it travels. The remarkable growth of networks and the Internet over the past decade has been enabled by previous generations of optical technol- ogy. Optics is, furthermore, the only technology with the physical headroom to keep up with this exponentially growing demand for communicating information. The exceptionally high carrier frequency allows transmission of high bandwidths. Satisfying that future demand will, however, require the continued research and development (R&D) of new technology for optics and continued innovation in its interface with electronics. The use of optics will not be restricted to the traditional market of long-distance telecommunication. Increasingly optics will be used for ever-shorter distances, possibly providing few-millimeter or shorter links between the silicon chips themselves. One key driver for such shorter distances will be to attain sufficient density of communications. A second key driver will be to control the growing power dissipation that is due to switching/transmission elements and the environmental impact of information processing. This chapter summarizes the developments in optics over the last 15 years (i.e., since the publication of the National Research Council’s [NRC’s] Harnessing Light1) in information commu- 1  National Research Council. 1998. Harnessing Light: Optical Science and Engineering for the 21st Century. Washington, D.C.: National Academy Press. 64

OCR for page 64
C o m m u n i c at i o n s , I n f o r m at i o n P ro c e s s i n g , and D ata S to r ag e 65 nications, processing, and storage and indicates key directions for the future for optics in these fields. Communications Optical communications networks provide the underlying high-capacity, ubiq- uitous connectivity that underpins the global Internet. Figure 3.1 characterizes the growth of communication and computing between 1986 and 2007, based on a broad collection of data.2 Approximately around the year 2000, Internet traffic took over from voice telephone as the single largest communication format for information. Now Internet traffic dominates completely. All of the long-distance communications on the Internet are over optical fiber. Major advances in transmission techniques and technologies have allowed network providers to provide extremely cost-effective network upgrades that have kept pace with the extraordinary appetite for broadband Internet services. That growth, as exemplified in Figure 3.1, has driven network bandwidth demands by a factor of 100 over the last 10 years. That increase has been enabled by realizing the full potential of wavelength division multiplexing (WDM) that has resulted in fibers carrying as many as 100 separate wavelengths. In addition, the capacity per wavelength in commercially deployed terrestrial networks has increased from a maximum of 10 gigabits per second (Gb/s) per wavelength when the first edition of Harnessing Light was published in 1998, to 100 Gb/s today. As a result, per fiber transmission capacities in terrestrial systems today as high as 5-10 terabits per second (Tb/s) are possible.3 Transoceanic capacities have lagged somewhat behind terrestrial values because the long amplifier-only distances and the desire to extend the amplifier spacing have made upgrading to per wavelength capacities above 10 Gb/s problematic. Nevertheless, transoceanic per fiber capacities of approximately 1 Tb/s are typical. For the future there are expectations that this growth will con- tinue as more video content calls for bandwidth and that there is a need for another factor-of-100 growth in the coming 10 years as well. Major advances have also been achieved in both cost-effectively managing the large capacity in today’s WDM optical networks and in leveraging the value propo- sition of optical amplifiers to provide multi-wavelength amplification over network mesh and ring architectures. Reconfigurable, wavelength-routed networks—in which wavelength-defined units of capacity can be added, dropped, or switched 2  Hilbert, M., and P. Lopez. 2011. The world’s technological capacity to store, communicate, and compute information. Science 332(6025):60-65. 3  Alferness, R.C. 2008. “Optical Communications—A View to the Future.” 24th European Confer- ence on Optical Communications, Brussels, Belgium, September 22. Available at http://ieeexplore. ieee.org/xpls/abs_all.jsp?arnumber=4729111&tag=1. Accessed June 26, 2012.

OCR for page 64
66 Optics and Photonics: Essential Technologies for O u r N at i o n FIGURE 3.1  Traffic In bytes per second (B/s) (1 byte = 8 bits) on the Internet, on voice telephone, and overall, 1986-2007. Also shown is the capacity to compute in general-purpose machines, expressed in millions of instructions per second (MIPS). SOURCE: MIPS graph based on data extracted from Hilbert, M., and P. Lopez. 2011. The world’s technological capacity to store, communicate, and com- pute information. Science 332(6025):60-65. from one fiber route to another fiber route directly in the optical domain without the need for conversion to electronics—are now heavily deployed in long-haul ter- restrial networks as well as metropolitan networks. Wavelength-routed networks provide cost-effective solutions because they allow data on wavelengths passing through a node at a multi-route network node to remain in the optical domain and benefit from the cost-effective multi-wavelength amplification enabled by optical amplifiers, rather than needing to be individually electronically regenerated. The large increase in capacity demand has ensured that a prerequisite for the economic viability of such networks—namely, that the capacity demand between any two node pairs on the network be at least as large as that which can be carried by a single wavelength—is met. WDM optical networks require reconfigurable optical add/drop multiplexers (ROADMs) to, under network electrical control, drop or add wavelength chan- nels at a node and to switch wavelength channels from one fiber route to another. ROADMs are key enablers that have evolved significantly in their functionality, providing increasing levels of flexibility, and in their capacity, or number of fiber ports and wavelengths per fiber, over the last decade. Further progress in these network elements and their enabling technologies will be essential to addressing the growing demand for capacity.

OCR for page 64
C o m m u n i c at i o n s , I n f o r m at i o n P ro c e s s i n g , and D ata S to r ag e 67 Ultimately, networks are no better than the access capacity that they provide to the end user, whether that customer is a business or a residence. Increasingly that access is through an optical link. The last decade has seen significant increase in the deployment of fiber in the access network, initially to the curb, but increasingly also directly to the business or home. The United States is not a world leader in fiber to the home or business. It ranks roughly at number 11 among the countries in the world, with approximately 7 percent penetration. The penetration of fiber to the home in the United States is roughly 5 percent.4 Passive optical networks (PONs) are the primary broadband optical delivery architecture, providing the shared bandwidth of a fiber to multiple users (16 to 64 users). Initially, systems provided shared bandwidth of 2.5 Gb/s. New systems operating at a total bandwidth of 10 Gb/s are becoming available. Just as wavelength multiplexing has provided cost-effective bandwidth enhancement in long-haul and metropolitan networks, as capacity demand in the access network increases to enable new broadband services, it is expected that WDM will be employed for capacity expansion. A critical requirement will be robust, low-cost WDM optical components to operate in the outside plant. Research in this area at a somewhat modest level is ongoing; an example of U.S.-funded research in this area is the National Science Foundation (NSF) Engineering Research Center for Integrated Access Networks.5 In spite of the dramatic achievements that optics has brought to communica- tion networks over the last decade (or, perhaps, because of them), the demand for higher bandwidth, both in the fixed and the mobile domains, continues to grow rapidly. Comparing projections from a number of sources, it seems conservative to suggest that network capacity demand will grow at the rate of at least a factor of 100 over the next 10 years, approximately following the recent historical trend shown in Figure 3.1.6 Ubiquitous video is the key driver. Increasingly that video is two-way as more end users upload video to sharing websites. Mobile video is also growing at an extremely rapid rate—92 percent compound annual growth rate according to a Cisco report7—that puts large bandwidth demand on the backhaul 4  Montagne, R. 2010. “Understanding the Digital World.” Presented at the FTTH Council Eu- rope Conference. Available at http://www.ftthcouncil.eu/documents/Reports/Market_Data_Decem ber_2010.pdf. Accessed June 26, 2012. 5  More information is available at the website of the Center for Integrated Access Networks at www. cian-erc.org. Accessed June 26, 2012. 6  Korotky, S.K., R.-J., Essiambre, and R.W. Tkach. 2010. Expectations of optical network traffic gain afforded by bit rate adaptive transmission. Bell Labs Technical Journal 1(4):285-296. 7  CISCO. 2012. “Cisco Visual Networking Index: Forecast and Methodology, 2010-2015.” White pa- per. Available at http://www.cisco.com/en/US/solutions/collateral/ns341/ns525/ns537/ns705/ns827/ white_paper_c11-481360.pdf. Accessed June 26, 2012.

OCR for page 64
68 Optics and Photonics: Essential Technologies for O u r N at i o n network, which is increasingly requiring optical links. One can expect also that increasingly mobile backhaul will not be done by point-to-point links as is done today, but rather will be part of an extended-access network which serves end points that include wireless base stations, small enterprises, curb drops, and homes. Broadband mobility will require ubiquitous, cost-effective backhaul, resulting in the convergence of wire-line and wireless networks that are currently often sepa- rate. The nature of traffic in an Internet- and video-driven network could suggest different overall network architecture, with merging between the metropolitan and access networks. Integration of data centers into the network is key today and will be even more so in the future, especially as cloud services become more pervasive. Already in 2003, the NRC’s Committee on Coping with Increasing Demands on Government Data Centers stated: Recommendation: Data centers should aggressively adopt newer, more “bleeding edge” technical approaches where there might be significant return on investment. This should be done carefully to minimize the inevitable failures that will occur along the way. Even with the failures, the committee believes the cost savings and improvements for end users will be substantial when compared to the methods practiced today.8 Just as capacity demand has required the advantages of optics to be pushed further to the edges of the network, data centers will increasingly depend on optics for interconnection, and eventually for reconfigurability and switching. A future of “cloud services” in which most if not all digital services are performed by shared resources in the network is looking increasingly attractive. This new paradigm, which would depend on ubiquitous, instant, and highly reliable access to the net- work, could place demands on the network equivalent to those of the transition from voice to data in the late 1990s. The advent of WDM, made cost-effective with the optical amplifier, has enabled the cost-effective scaling of optical point-to-point transmission systems to meet the exponential demand growth. ROADMs and opti- cal cross-connects have made it possible to leverage that advantage to multi-node ring and mesh networks. As a result, optical networks that underpin the Internet have been able to keep pace with exploding demand over the last 10 years. How- ever, realistically, the several-orders-of-magnitude capacity increase resulting from many wavelengths will not be duplicated by simply adding even more wavelengths, because of limited optical amplifier bandwidth and fiber power limits in the fiber to mitigate transmission impairment.9 8  National Research Council. 2003. Government Data Centers: Meeting Increasing Demands. Wash- ington, D.C.: The National Academies Press. 9  Chraplyvy, A. 2009. “The Coming Capacity Crunch.” Plenary paper. 35th European Conference on Optical Communications, Vienna, Austria, September 21. Available at http://ieeexplore.ieee.org/ xpls/abs_all.jsp?arnumber=5287305&tag=1. Accessed June 26, 2012.

OCR for page 64
C o m m u n i c at i o n s , I n f o r m at i o n P ro c e s s i n g , and D ata S to r ag e 69 The optical communication industry is highly challenged to find the key new technologies, architectures, and techniques that will enable cost-effective growth in capacity to keep up with the capacity demand that video, including mobile video, will make on the network. Achieving that goal by simply increasing the number of wavelengths, or by increasing the bit rate per wavelength, appears unlikely with known or incrementally improved technology. Therefore, some fundamentally new optical solutions that build on top of today’s WDM optical networks will be urgently required in order to ensure that the global Internet is able to continue to serve as the engine for economic growth. It is estimated that without new tech- nology developments, Internet growth will be constrained starting between 2013 and 2014. Information Processing The main, and growing, use of optics in information processing is to connect information within and between information switching and processing machines. There are substantial physical reasons—specifically, reducing power dissipation and increasing the density of information communications—why optics is preferable and ultimately possibly essential for such connections.10 The idea of using optics for performing the logical switching in information processing—as in some kind of optical transistor—is one that has continuing research interest, although the criteria for success there are challenging,11 and no such use appears imminent. The last decade has seen sustained research interest in very advanced ideas such as quantum computing,12 although such ideas remain very much in fundamental re- search. Optics, however, is likely to have very important roles in any such quantum computing or quantum communications—for example, as the best means of com- municating quantum states over any meaningful distance—and optics might also be important in future quantum gates. Quantum encryption, a means of sending key information with immunity to any eavesdropping, has seen first commercial optical systems. In the meantime, efforts are needed to create systems that utilize the best combination of optics and electronics to enable integrated systems for efficient information processing. In the practical connection of information, optics has increasingly taken over the role of data communications within local networks for information process- ing systems such as data centers, supercomputers, and storage area networks, for 10  Miller,D.A.B. 2009. Device requirements for optical interconnects to silicon chips. Proceedings of the IEEE 97:1166-1185. 11  Miller, D.A.B. 2010. Are optical transistors the next logical step? Nature Photonics 4:3-5. 12  Nielsen, M.A., and I.L. Chuang. 2000. Quantum Computation and Quantum Information. Cam- bridge, U.K.: Cambridge University Press.

OCR for page 64
70 Optics and Photonics: Essential Technologies for O u r N at i o n distances from approximately 100 meters to 10s of kilometers. Commercial ap- proaches include optics in networking architectures such as Ethernet, Fiber Chan- nel, and Infiniband, especially as data rates and data densities rise, and in active optical cables that are internally optical but externally have electrical inputs and outputs. Low cost is particularly critical for shorter distances. Consequently, such data communications have often used technologies different from the long-haul telecommunications approaches, such as vertical cavity surface-emitting lasers and laser arrays, and multi-mode (rather than single-mode) optical fibers for the shorter distances. The use of optics in such networks is likely to steadily increase because of continuing increases in network traffic and information processing. As can be seen in Figure 3.1, the ability to compute information, as indicated by the growth of the computational power (expressed in millions of instructions per second [MIPS] that can be executed by the general-purpose silicon chips that are sold),13 has approximately the same growth rate as that of Internet traffic. The ability to connect information within information processing systems is, however, hitting increasingly severe limits. The ability of wires to interconnect does not scale well to higher densities, especially for information communications outside the sili- con chips themselves (“off-chip” communications). This point is well understood by the semiconductor industry, as shown, for example, in its projections of an off-chip interconnect bottleneck.14 The interconnect capability of electrical wires between chips, or processor cores, is not going to scale to keep up with the ability to compute—sometimes called the bytes per flops (floating point operations per second) gap.15 Thus, although Internet bandwidth and raw computational power in MIPS or flops might continue to scale, the systems performance in between will not, unless a new interconnect technology is introduced that can scale in capacity. High-performance silicon chips today, as characterized, for example, by the International Technology Roadmap for Semiconductors (ITRS) numbers,16 already have interconnect capabilities well into the range of multiple terabits per second per chip. To scale to keep up with processing power, future chips in the 2020 time frame would need 100s of terabits per second of interconnect capability, an amount of interconnect for one chip that is comparable to the entire Internet traffic today. It is worth noting that generally there is likely much more information sent inside in- formation processing systems than is sent between them. One order-of-magnitude 13  Hilbert M., and P. Lopez. 2011. The world’s technological capacity to store, communicate, and compute information. Science 332(6025):60-65. 14  Miller, D.A.B. 2009. Device requirements for optical interconnects to silicon chips. 15  Miller, D.A.B. 2009. Device requirements for optical interconnects to silicon chips. 16  More information is available through the International Technology Roadmap for Semiconduc- tors at http://www.itrs.net/Links/2010ITRS/Home2010.htm. Accessed June 26, 2012.

OCR for page 64
C o m m u n i c at i o n s , I n f o r m at i o n P ro c e s s i n g , and D ata S to r ag e 71 estimate17 is that each byte traversing the Internet causes approximately 1 million bytes of data communications within data centers. The large and increasing interconnected bandwidth inside information pro- cessing systems creates problems both for density of interconnects and for power sourcing and dissipation. Interconnecting one bit of information off a chip cur- rently involves several to 10s of picojoules (pJ) of energy. A substantial portion (e.g., 50 to 80 percent) of all power dissipation on silicon chips is used to interconnect the information rather than to perform logic.18 Power dissipation in general is a severe limitation on information processing; for example, the clock rate on main- stream silicon chips has not changed substantially in recent years because higher clock rates would lead to too much power dissipation. A recent NRC study19 on supercomputing also emphasizes this problem with power dissipation and the need, in particular, to reduce interconnection power while maintaining or even increasing the byte-per-flops ratio. Power dissipation in information processing is already environmentally sig- nificant. Electricity used in data centers in 2010 likely accounted for between 1.1 and 1.5 percent of total electricity use globally, and between 1.7 and 2.2 percent of electricity use in the United States.20 The power dissipation in interconnects in serv- ers alone has been estimated to exceed solar power generation capacity.21 Of course, information processing and communication can have substantial environmental benefits in improving the efficiency of many activities and in reducing travel.22 But, if current growth trends in Internet traffic and in information processing continue, reducing the energy per bit processed and/or communicated will be crucial. Even though sending a bit over the Internet might take approximately 10 nanojoules (nJ)/bit23 compared to 10s of pJ/bit for off-chip interconnects, the vastly greater number of bits being sent inside the machine (approximately 106 17  G. Astfalk. 2009. Why optical data communications and why now? Applied Physics A: Materials Science and Processing 95(4): 933-940. 18  Miller, D.A.B. 2009. Device requirements for optical interconnects to silicon chips. 19  National Research Council. 2010. The Future of Computing Performance: Game Over or Next Level? Washington, D.C.: The National Academies Press. 20  Koomey, Jonathan. 2011. Growth in Data Center Electricity Use 2005 to 2010. Oakland, Calif.: Analytics Press. Available at http://www.analyticspress.com/datacenters.html. Accessed October 27, 2011. 21  Miller, D.A.B. 2009. Device requirements for optical interconnects to silicon chips. 22  The Climate Group. 2008. SMART 2020: Enabling the Low Carbon Economy in the Information Age. A report by the Climate Group on behalf of the Global eSustainability Initiative. Available at http://www.smart2020.org/_assets/files/02_Smart2020Report.pdf. Accessed June 26, 2012. 23  Tucker, R.S., R. Parthiban, J. Baliga, K. Hinton, R.W.A. Ayre, and W.V. Sorin. 2009. Evolu- tion of WDM optical IP networks: A cost and energy perspective. Journal of Lightwave Technology 27:243-252.

OCR for page 64
72 Optics and Photonics: Essential Technologies for O u r N at i o n times24) means that the power dissipation inside information processing machines may substantially dominate over telecommunications power. Hence reducing in- terconnect power inside machines is crucial for controlling power dissipation in information processing. The research understanding of the reasons for, and the requirements on, op- tics and optoelectronics for such shorter connections is now relatively complete and consistent.25,26 Optics helps in two ways. First, optics can carry information at much higher density than is possible in electrical wires, as is essential for future scaling of interconnect capacity. Second, optics can fundamentally save energy in interconnects because it completely avoids the need for charging wires (the charging of electrical wires to the signal voltage is the dominant source of energy dissipation in electrical interconnects). The key challenges in short-distance, high-density, low-energy optical inter- connect are technological: optoelectronic devices must operate with very low ener- gies—for example, 10 femtojoules (fJ)/bit—much lower than have been required for telecommunications and data communications until now. For cost and perfor- mance reasons, the optical and optoelectronic technologies have to be integrated with electronics in a mass-manufacturable process. New technologies have been advancing in research and early product intro- ductions over the last decade, especially in terms of devising ways of combining optics, optoelectronics, and electronics in silicon chips and platforms (silicon photonics).27,28 These technological opportunities are discussed below. Data Storage The past decade has seen the maturation of technologies such as the DVD (digital versatile disk) for the distribution of consumer video, and the emergence of third-generation disk technologies such as Blu-ray for higher density. Research continues on possible fourth-generation optical disks, including multi-layer and possible holographic recording. Optics may also have crucial roles to play in other potential emerging technologies such as HAMR (heat-assisted magnetic 24  Astfalk, G. 2009. Why optical data communications and why now? Applied Physics A: Materials Science and Processing 95:933-940. 25  Miller, D.A.B. 2009. Device requirements for optical interconnects to silicon chips. 26  Astfalk, G. 2009. Why optical data communications and why now? Applied Physics A: Materials Science and Processing 95:933-940. 27  Lipson, M. 2005. Guiding, modulating, and emitting light on silicon—Challenges and opportu- nities. Journal of Lightwave Technolology 23(12):4222-4238. 28  Jalali, B., and S. Fathpour. 2006. Silicon photonics. Journal of Lightwave Technolology 24(12): 4600-4615.

OCR for page 64
C o m m u n i c at i o n s , I n f o r m at i o n P ro c e s s i n g , and D ata S to r ag e 73 recording)29 for next-generation magnetic hard drives, for which near-field optics using nanophotonic technologies may provide the precise heating required. The picture for the future of optical disks like DVDs is less clear because of the competition from other forms of storage, such as flash memory (memory sticks), and from networks. Whether optical disks will have the role in the distribution of data and entertainment, such as movies and games, which they have had in the past is debatable because of the growth of broadband networks as an alternative distribution channel—as, for example, in the growing use of video on demand over the Internet. IMPACT EXAMPLE: THE INTERNET As is now readily self-evident, the Internet has transformed the way in which society operates. Previous conceptualizations of distance and geography have dis- appeared in a maze of ubiquitous cellular telephones, e-mail, web browsing, and social networking. Indeed, the productivity of people coming of age after the mid- 1990s is intertwined with the easily available communication and information found through the Internet. Initially the Internet was not reliant on optical communications. However, data transfer rates during those formative times were painfully slow. Without optical technologies, conventional methods were utilized: users depended on slow, dial-up modems, e-mails occasionally took hours before arriving at their destination, and making intercontinental phone calls involved an annoying delay between speaking and listening. Optics changed the Internet and data transfer by increasing the capacity of the system by nearly 10,000-fold over the past two decades. The nearly instantaneous nature of current optical-based communication allows for real-time video chats. Telepresence, telemedicine, and tele-education would not be possible without op- tics. In short, without optics, the Internet as we know it would not exist. As it was stated in 2006 by the NRC Committee on Telecommunications Research and De- velopment: “Telecommunications has expanded greatly over the past few decades from primarily landline telephone service to the use of fiber optic, cable, and wire- less connections offering a wide range of voice, image, video, and data services.”30 The same report continues: Yet it is not a mature industry, and major innovation and change—driven by re- search—can be expected for many years to come. 29  Kryder, M.H., E.C. Gage, T.W. McDaniel, W.A. Challener, R.E. Rottmayer, G. Ju, Y.-T. Hsia, and M.R. Erden. 2008. Heat assisted magnetic recording. Proceedings of the IEEE 96:1810-1835. 30  National Research Council. 2006. Renewing U.S. Telecommunications Research. Washington, D.C.: The National Academies Press, pp. 1-20.

OCR for page 64
74 Optics and Photonics: Essential Technologies for O u r N at i o n Without an expanded investment in research, however, the nation’s position as a leader is at risk. Strong competition is emerging from Asian and European countries that are making substantial investments in telecommunications R&D. For many telecommunications products and services that are now commodities, the United States is at a competitive disadvantage compared with countries where the cost of doing business is lower. Continued U.S. strength in telecommunications, therefore, will require a focus on high-value innovation that is made possible only by a greater emphasis on research. Expansion of telecommunications research is also necessary to attract, train, and retain research talent. Telecommunications research has yielded major benefits such as the Internet, radio frequency wireless communications, optical networks, and voice over Inter- net Protocol. Promising opportunities for future research include enhanced Internet architectures, more trustworthy networks, and adaptive and cognitive wireless net- works.31 The 2002 NRC study Atoms, Molecules, and Light: AMO Science Enabling the Future stated: Internet optical backbone link capacity increased a hundredfold between 1995 and 1998, to 20,000 trillion bits per second. While the communications industry has recently suffered a downturn, the potential for increasing demand for capacity in the 21st century remains, with features and services such as HDTV, broadband communications, and advanced home security systems becoming more available through the use of optical fiber, wireless, satellite, and cable connections.32 The use of optics, however, is not always obvious to the casual observer. Often someone can be seen using a cell phone to perform an Internet search. Where is the optics? The cell phone uses a wireless radio connection to a local cell tower, but that radio signal is converted to an optical data stream and sent along the fiber optic network across the planet. The data search itself, such as through Google, relies on data centers in which clusters of co-located computers talk to each other through high-capacity optical cables. In fact, there can be as many as a million lasers in a given data center. During the introduction to the 2009 Nobel Prize Lecture by Dr. Charles Kao on the innovation of optical fiber communications, the Physics Committee Chair said: “The work has fundamentally transformed the way we live our daily lives.”33 31  National Research Council. 2006. Renewing U.S. Telecommunications Research. pp. 1-20. 32  National Research Council. 2002. Atoms, Molecules, and Light: AMO Science Enabling the Future. Washington, D.C.: The National Academies Press, pp. 7-8. 33  More information on the Nobel Prize is available at http://www.nobelprize.org/nobel_prizes/ physics/laureates/2009/. Accessed June 26, 2012.

OCR for page 64
C o m m u n i c at i o n s , I n f o r m at i o n P ro c e s s i n g , and D ata S to r ag e 91 materials.102 Novel nanophotonics approaches to optical components such as very compact wavelength splitters have emerged, including the concept of nonperiodic design for function.103 Extremely sensitive optical and optoelectronic device struc- tures have been demonstrated, including lasers, photodetectors, modulators, and switches, with some down to single-photon sensitivity.104 With such nanophoton- ics approaches, there is demonstrated potential for very low threshold lasers,105 very low energy modulators, and very low capacitance photodetectors. Novel ap- proaches to optoelectronic devices exploiting nanoscale growth techniques may circumvent some of the difficulties (such as crystal lattice constant matching) that often limit more conventional approaches to device fabrication.106 As the size of many of these new optics and photonics elements shrinks, it certainly opens up for the possibility to produce systems that utilize the best of optics and electronics to enable integrated systems to seamlessly provide solutions in many of today’s fields. Data Storage The advent of nanometallic structures to concentrate light to deeply sub- wavelength volumes has opened up new opportunities for optics in data storage, both for sub-wavelength optical reading and writing and for the use of optics to concentrate light for other storage approaches, as in heat-assisted magnetic recording, whereby the light provides the very localized heating above the Curie temperature to enable correspondingly localized changes in magnetic state. This emerging technology is apparently a serious contender for future mainstream magnetic hard-drive technologies.107 102  Chen, H., C.T. Chan, and P. Sheng. 2010. Transformation optics and metamaterials. Nature Materials 9:387-396. 103  Liu, V., Y. Jiao, D.A.B. Miller, and S. Fan. 2011. Design methodology for compact photonic- crystal-based wavelength division multiplexers. Optics Letters 36:591-593. 104  Fushman, I., D. Englund, A. Faraon, N. Stolz, P. Petroff, and J. Vuckovic. 2008. Controlled phase shifts with a single quantum dot. Science 320(5877):769-772. 105  Ellis, B., M.A. Mayer, G. Shambat, T. Sarmiento, J. Harris, E.E. Haller, and J. Vučković. 2011. Ultralow-threshold electrically pumped quantum-dot photonic-crystal nanocavity laser. Nature Photonics 5:297-300. 106  Chen et al. 2011. Nanolasers grown on silicon. 107  Kryder, M.H., E.C. Gage, T.W. McDaniel, W.A. Challener, R.E. Rottmayer, G. Ju, Y.-T. Hsia, and M.R. Erden. 2008. Heat assisted magnetic recording. Proceedings of the IEEE 96:1810-1835.

OCR for page 64
92 Optics and Photonics: Essential Technologies for O u r N at i o n MANUFACTURING Communications Optical communications network equipment includes functional elements such as transmissions systems; optical networking elements, including add/drop multiplexers; crossconnects; and network-management software systems. Substan- tial numbers of circuit boards for electrical multiplexing, electrical cross-connects, and control functions are also required. These optical systems and network ele- ments are assembled from optical modules (for example, optical transmitters and receivers) that in turn are built from optical components, including lasers, optical modulators, photodetectors, optical fiber amplifiers, and others. At the systems’ vendor level, which is a roughly $15 billion annual business, there have been significant changes in the industry since Harnessing Light108 was written. Nortel, a major Canadian vendor, has gone out of business. Lucent Tech- nologies has been acquired by Alcatel to form Alcatel-Lucent, with headquarters in Paris, France. Of the top 10 optical network systems vendors, two—Cisco and Tellabs—are headquartered in the United States. Alcatel-Lucent, with significant R&D operations in the United States, and Huawei, headquartered in Shensen, China, have vied for the market-leader position over the last several years. ZTE, also from China, is also in the top 5.109 In the optical module and component busi- ness, nearly half of the market share is owned by U.S.-based companies: Finisar and JDSU are the market number one and number two leaders, respectively.110 The growth and fabrication of the III-V semiconductor chips to build the basic laser and detector discrete devices are done in the United States, Japan, and Taiwan and are increasingly in areas like Singapore and Southeast Asia. There is little evidence that companies in China have mastered this highly specialized technology at this time, but work in the research laboratory is ongoing there. At the circuit board and optical module level, manufacturing assembly is increasingly being done in Asia. However, some U.S. companies continue to do assembly of leading-edge boards—such as 100 Gb/s transmitters and receivers—in the United States. Infinera, a metropolitan and long-haul optical networking com- pany based in the United States and a leading proponent of photonic integrated circuits (PICs) to provide transmitter and receiver arrays for WDM transmission 108  NationalResearch Council. 1998. Harnessing Light. 109  Ovum. 2011. “Ovum: ZTE Ranked Third in Growing Global Optical Networking Market.” Ovum. Available at http://www.tele.net.in/news-releases/item/7010-ovum-zte-ranked-third-in- growing-global-optical-networking-market. Accessed July 26, 2012. 110  Ovum. 2012. “Oclaro Combines with Opnext to Challenge Finisar for No. 1.” Available at http:// ovum.com/2012/03/27/oclaro-combines-with-opnext-to-challenge-finisar-for-no-1/. Accessed July 26, 2012.

OCR for page 64
C o m m u n i c at i o n s , I n f o r m at i o n P ro c e s s i n g , and D ata S to r ag e 93 systems, produces the PIC chip in the United States. Because PICs provide on- chip integration of basic components that otherwise would be done by hand, this technology eliminates much of the manually intensive activity of assembling and connecting individual components that is currently done more cost-effectively in low-labor-cost regions. More highly integrated optical modules will be needed for next-generation systems from a functional point of view. By being the technology leader in this area, the United States could potentially increase the value added in the United States that, with today’s modules made up of discrete devices, is mov- ing offshore. Optical components for communications networks still involve several t ­ echnologies—indium phosphide (InP), CMOS, lithium niobate, and silica. At the same time, volumes addressed by each technology are rather limited. The company and/or nation that finds a consolidation technology (if possible) could have sub- stantial advantage. Some might suggest that silicon photonics has the potential to displace other passive technologies as well as non-lasing electro-optic technologies such as lithium niobate. Also worth noting is anecdotal evidence which suggests that as the manu- facturing of optical systems equipment or optical modules moves to lower-cost manufacturing regions, the development function tends eventually to move to that region as well. Information Processing Current optoelectronic systems are dominated by III-V semiconductors in systems with generally small integration levels. To make the transition to a new technology such as silicon-based photonics with high integration levels is difficult, and it appears to need high volumes to justify it.111 There is, therefore, a “chicken and egg” issue with this major potential technology shift. There is substantial medium-to-long-term potential for a very high volume market for optical inter- connections, one that could help retain a U.S. technology and market leadership in information processing equipment, but it may need a shift to a low-cost, pos- sibly silicon-based technology to satisfy that market. The same silicon photonics platform, if successfully adopted, also has the capability to create a broad range of nanophotonic structures. 111  Fuchs, E., R. Kirchain, and S. Liu. 2011. The future of silicon photonics—Not so fast?: The case of 100G Ethernet LAN transceivers. Journal of Lightwave Technology 29(15):2319-2326.

OCR for page 64
94 Optics and Photonics: Essential Technologies for O u r N at i o n Data Storage At present, the substantial majority of manufacturing for optical data storage products, such as DVDs, is likely in Asia.112 Given that the impact of a next genera- tion of optical disk storage faces strong competition from networks as a distribu- tion medium for video, it is not clear that a major push to manufacture such a technology in the United States is a strong choice. There is significant research in the United States in future areas such as the HAMR optically assisted hard-drive technology, which may influence some future manufacture of such magnetic hard drives in the United States. At this time, beyond the strong possibilities in HAMR and some continuing use of optical disks such as DVDs, the path for other optical data storage approaches is less clear, although there is continuing development of holographic storage for commercial archiving, with possible capacities in the 500 Gb to 1 Tb range in a disk of a size comparable to that of current DVDs.113 Other approaches remain in research at the present. ECONOMIC IMPACT Communications networks have taken on a role well beyond people-to-people voice communications; they provide the information and trade routes of the new global digital economy. They will also likely provide the sensor integration and distributed control for the power grid, transportation, and freight networks and smart enterprises and cities in the future. Optical interconnects are very likely to have a key role in future information processing technology and systems, allowing performance to continue to scale. Leadership in communications networks and interconnects, including the underpinning optical networks, both in the R&D and in commercialization, deployment, and effective use, will be absolutely essential to maintaining and enhancing economic growth and improving the quality of people’s lives in the 21st century. These optical technologies will be increasingly indispensable for the technology of the information age. This critical importance of communications networks to the future has driven action in countries around the world. Several countries, including Australia for example, have mandated broadband fiber to the home. Europe continues to pro- 112  Esener S.C., M.H. Kryder, W.D. Doyle, M. Keshner, M. Mansuripur, and D.A. Thompson. 2012. International Technology Research Institute, World Technology (WTEC) Division. 1999. WTEC Panel Report on the Future of Data Storage Technologies. Available at http://www.wtec.org/pdf/hdmem. pdf. Accessed August 1, 2012. 113  More information is available at General Electric’s Global Research website at http:// ge.geglobalresearch.com/blog/breakthrough-in-micro-holographic-data-storage/. Accessed June 26, 2012.

OCR for page 64
C o m m u n i c at i o n s , I n f o r m at i o n P ro c e s s i n g , and D ata S to r ag e 95 vide substantial research funding, as do Korea and Japan. However, the single most important threat to U.S. leadership, especially with respect to network equip- ment development and manufacture, is likely to come from China, where the g ­ overnment-sponsored Huawei ranks as number 1 or number 2 in global market share,114 followed closely by ZTE, another systems vendor from China. Already extremely innovative in its products, China is now focused on research and inno- vation. Its paper submissions to international optics journals and to the premier global optical communications conferences have increased substantially in the last 5 years.115 There is particularly strong research in Europe and recently also in Japan in the emerging silicon photonics platforms for networks and interconnects. Given that labor-intensive industries will continue to migrate their manufac- turing to low-labor-cost regions, it is imperative that the United States stay at the leading edge of optical technology at the component, platform, and system levels. This approach appears to have served the U.S. electronics industry well. To keep a substantial portion of the value chain in the United States, it will be important for U.S. enterprises to have the critically enabling intellectual property that pro- vides a barrier of entry without financial compensation. To have that intellectual property, it is essential to be at the leading edge of enabling the fundamental and applied research. Suggested areas of focus on the component side include very high speed electronics and optical components, including modulators and detectors at operating rates of 400 Gb/s and 1 Tb/s; advanced signal processing to overcome transmission impairments for coherent systems; and on-chip integration that pro- vides increased functionality while reducing size, power consumption, and cost. Such integrated chips also do not require the substantial manual assembly that is now being performed in low-labor-cost regions. In the systems and networks area, finding a new approach to cost-effectively achieve several-orders-of-magnitude long-haul and metropolitan distance transmission capacity and to rapidly get that technology to market will be critical in order for the United States to maintain a strong global leadership position. Such technology evolution will also underpin the increasing move to optics inside information processing that will be essential for the continued scaling of an information-driven economy. The successful develop- ment of an integrated platform technology, such as some version of silicon photon- ics that can service a broad range of applications and integration with electronics, could be a major enabler for U.S. economic impact. 114  Huawei. 2010. Milestones. Available at http://www.huawei.com/en/about-huawei/corporate-info/ milestone/index.htm. Accessed July 26, 2012. 115  Cao, J. 2012. A new journal in optics and photonics—Light: Science and Applications. Editorial. Light: Science and Applications 1:Online. Available at http://65.199.186.23/lsa/journal/v1/n3/full/ lsa20123a.html. Accessed July 26, 2012.

OCR for page 64
96 Optics and Photonics: Essential Technologies for O u r N at i o n Comparison Between the United States and the Rest of the World For decades the United States has been the envy of the entire world in terms of R&D in high-tech fields, including optics and photonics for communications. For decades the most prestigious international journals, conferences, and professional societies have been based in the United States. Many of the most impactful R&D advances came from the U.S. industrial laboratories (e.g., AT&T Bell Laboratories, Corning, RCA Sarnoff Labs, IBM), with several Nobel Prizes as well. The U.S. uni- versity system was unrivaled, with student researchers flocking to the United States from countries around the world. The combined R&D in optics from corporate laboratories and federally funded university labs set the bar high in terms of quality and quantity. Critical advances were made in the United States, including low-loss fibers, semiconductor lasers, optical amplifiers, and information theory. The scenario has changed somewhat in the past 20 years. In general, corporate laboratories no longer enjoy steady, long-term funding, and the main sources of university research funding have not grown at a pace consistent with that in the rest of the world. The main sources of university funding in communications and information processing, such as the NSF and the Department of Defense (DOD), tend to fund a relatively small percentage of new proposals, particularly with a 10-year focus. In the area of optical research for information processing, U.S. efforts are gen- erally comparable in size to efforts in Europe and Japan in the emerging technolo- gies for optics in information processing, but the United States has no overall lead against these research competitors. The United States lacks the larger European framework projects that help tie together a broad range of players from academia to industry. Both Europe and Japan are making substantial investments in silicon photonics technology research now. Trends in terms of the United States in relation to the rest of the world include the following: · Research funding. Nations all over the world view the United States as a benchmark in terms of research funding. In order for countries to com- pete, many nations focus on specific, strategic areas in which to invest in long-term photonics research. The United States tends not to make such strategic, long-term “bets.” The effect is that in any given area, the United States will share leadership with, and perhaps be surpassed by, research in specific countries. High-speed communications in Germany, integrated photonics in Japan, and access technologies in China are all examples of thrusts centered in various countries. · Professors. Several nations have invested in hiring distinguished researchers to prominent, well-paid professorial positions. Oftentimes, these research-

OCR for page 64
C o m m u n i c at i o n s , I n f o r m at i o n P ro c e s s i n g , and D ata S to r ag e 97 ers have made their reputations in the United States, only to be drawn away in their prime. Countries including Australia, Canada, China, and Germany have hired excellent researchers in communications with generous research funding. This trend, if it accelerates, could have a profound impact. · China. When observing statistics of prestigious optical communications journals, it is undeniable that the quantity of research from China is grow- ing rapidly. Looking at “optical communication” in the Scopus database, there are 15,003 journal publications from 1969 to 2012 from 21 common optical communications journals;116 in total, the United States represents 3,909 of those, and China, 1,303. However, since 2000 the U.S. scientists have published 42 percent of their work, or 1,642 publications, whereas Chinese scientists have published 85 percent of their work, or 1,108 publi- cations. Most of the Chinese publications have appeared since 2000. More- over, the quality gap, according to rejection ratio statistics, is narrowing. Using the number of citations that a paper receives as a metric of quality also points to China’s gaining on the United States. A comparison between the two countries with respect to publications related to the three keywords (Optics, Photonics, and Communication) in the Scopus database, using the highest-cited papers, gives the following results: currently there are 2.3 citations of a U.S. paper for each citation that a Chinese paper receives; however, 5 years ago this number was 2.8, or 16 percent higher.117 Further- more, China is aggressively funding research in optical communications and has a foothold as home to some of the world’s largest communications companies. China is poised to make great strides in the coming decade. FINDINGS AND CONCLUSIONS In spite of great progress in optical communications over the last three decades, the optics and photonics community faces a great challenge if optical communica- tions networks are to continue to satisfy the insatiable global demand for informa- 116  The selected journals were these: Optics Communications, IEEE Photonics Technology Letters, Op- tics Express, Journal of Lightwave Technology, Optical Engineering, Microwave and Optical Technology Letters, Journal of Optical Communications, Physical Review A Atomic Molecular and Optical Physics, Optics Letters, Photonic Network Communications, Applied Optics, Applied Physics Letters, Journal of the Optical Society of America B Optical Physics, Fiber and Integrated Optics, Optical and Quantum Electronics, Proceedings of SPIE the International Society for Optical Engineering, IEEE Transactions on Communications, IEEE Transactions on Microwave Theory and Techniques, IEEE Proceedings Optoelec- tronics, Journal of Optical Networking, and Photonics Spectra. 117  Data from SciVerse Scopus, www.scopus.com. Data derived using the top 25 papers and averag- ing over 5-year periods. Accessed March 21, 2012.

OCR for page 64
98 Optics and Photonics: Essential Technologies for O u r N at i o n tion processing, data storage, bandwidth, and ubiquitous connectivity that, in turn, drive global economic growth. Key Finding:  The growth in bandwidth demand over the next 10 years is expected to be at least another 100-fold, possibly much more. It is important to note that the previous 100-fold gain in capacity that came very naturally with wavelength division multiplexing has been used up; growth by means of higher bit rates per wavelength comes more slowly; hence without a new breakthrough, increases in data transmission capacity will stall. Key Finding:  Cloud services not only drive capacity demand, but also make critical the role of large data centers. This will usher in a new and important era for short- distance optical links in massive numbers to provide cost- and energy-efficient high-density interconnects inside data centers and in information processing sys- tems generally. Key Finding:  Silicon-based photonic integration technologies offer great potential for short-distance applications and could have great payoff in terms of enabling continued growth in the function and capacity of silicon chips if optics for inter- connection could be seamlessly included in the silicon CMOS platform. It is also highly likely that integrated optoelectronics is a critical development area with significant growth potential for continuing the advance of defense systems. Finding:  Nanophotonic technologies promise very compact and high-performance optics and optoelectronics that could allow such platforms to continue to scale to higher density and performance. Information processing in general and data centers will also require massive and exponentially growing data storage. Finding:  Magnetic storage will continue as the primary data storage technology, and optics-assisted magnetic techniques may play an important future role. Finding:  Communications networks and information processing consume a relatively small fraction of the global energy budget (approximately 2 percent). However, with the growing network demands, this percentage will likely grow sig- nificantly unless successful action is taken to reduce systems’ energy requirements. Finding:  Leveraging networks with applications like telepresence to reduce energy- hungry activity such as travel would reduce total energy consumption. Finding:  Optics has the potential to increase the energy efficiency in networks by, for example, replacing high-energy-loss electrical conductors in data centers and wireless backhaul or, perhaps, even on silicon chips.

OCR for page 64
C o m m u n i c at i o n s , I n f o r m at i o n P ro c e s s i n g , and D ata S to r ag e 99 Conclusion:  The committee believes that a strong partnership among industry, universities, and government agencies will be crucial to overcoming technical challenges and to ensuring that the United States leverages that knowledge to gain market leadership. Finding:  Many of the optical communications successes over the last 10 years are built on earlier research that came from research laboratories of vertically integrated companies, as well as strongly supporting government agencies. The industry is no longer integrated, but instead is segmented by material fabrication, components, modules, systems, network providers, and content and service pro- viders. In this fragmented environment there has been a reduction in industrial research laboratories, because the reduced scale makes it difficult for companies to capture the value that is prudently needed to justify investing in research. Finding:  Today’s broadband access by individual users in the United States is nei- ther high-capacity nor available at reasonable cost to a large fraction of the popu- lation. Bandwidths of 1 Gb/s represent the current state of the art in broadband access to the home in leading installations today. RECOMMENDATIONS AND GRAND CHALLENGE QUESTIONS Key Recommendation:  The U.S. government and private industry, in combina­tion with academia, need to invent technologies for the next factor-of-100 cost-effective capacity increase in long-haul, metropolitan, and local-area optical networks. The optics and photonics community needs to inform funding agencies, and information and entertainment providers, about the looming roadblock that will interfere with meeting the growing needs for network capacity and flexibility. There is a need to champion collaborative efforts, including consortia of companies, to find new technology—transmission, amplification, and switching—to carry and route at least another factor-of-100 capacity in information over the next 10 years. This key recommendation leads directly to the first grand challenge question: 1.  ow can the U.S. optics and photonics community invent technologies for H the next factor-of-100 cost-effective capacity increases in optical networks? The first recommendation of the chapter states a goal for increased capacity; the next recommendation offers a path to help achieve that goal, especially with re- spect to very short range communication, such as that required inside a data center. Key Recommendation:  The U.S. government, and specifically the Department of Defense, should strive toward harmonizing optics with silicon-based electronics

OCR for page 64
100 Optics and Photonics: Essential Technologies for O u r N at i o n to provide a new, readily accessible and usable, integrated electronics and optics platform. They should also support and sustain U.S. technology transition toward low- cost, high-volume circuits and systems that utilize the best of optics and electronics in order to enable integrated systems to seamlessly provide solutions in commu- nications, information processing, biomedical, sensing, defense, and security ap- plications. Government funding agencies, the DOD, and possibly a consortium of companies requiring these technologies should work together to implement this recommendation. This technology is one approach to assist in accomplishing the first key recommendation of this chapter concerning the factor-of-100 increase in Internet capability. The second key recommendation in this chapter leads to the second grand challenge question: 2.  ow can the U.S. optics and photonics community develop a seamless inte- H gration of photonics and electronics components as a mainstream platform for low-cost fabrication and packaging of systems on a chip for communica- tions, sensing, medical, energy, and defense applications? In concert with meeting the first grand challenge, achieving the second grand challenge would make it possible to stay on a Moore’s law-like path of exponen- tial performance growth. The seamless integration of optics and photonics at the chip level has the potential to significantly increase speed and capacity for many applications that currently use only electronics, or that integrate electronics and photonics at a larger component level. Chip-level integration will reduce weight and increase speed while reducing cost, thus opening up a large set of future pos- sibilities as devices become further miniaturized. The size and number of data centers in the United States and globally is ex- pected to grow dramatically over the next decade to address the needs of a global digital society, especially if cloud services become more pervasive. It is clear that these data centers will be the focal point for the development and deployment of new optics and photonics communications technologies, and as such will be very important for the economy. Key Recommendation:  The U.S. government and private industry should position the United States as a leader in the optical technology for the global data center business.

OCR for page 64
C o m m u n i c at i o n s , I n f o r m at i o n P ro c e s s i n g , and D ata S to r ag e 101 Optical connections within and between data centers will be increasingly important in allowing data centers to scale in capacity. The committee believes that strong partnering between users, content providers, and network providers, as well as between businesses, government, and university researchers, is needed for ensuring that the necessary optical technology is generated, which will support continued U.S. leadership in the data center business. Recommendation:  The U.S. government and private industry, in conjunction with academia, should strive to develop technology to have optics take over the role of communicating and interconnecting information, not just at long distances, but at shorter distances as well, such as inside information processing systems, even to the silicon chip itself, thereby allowing substantial reductions in energy consumption in information processing and allowing the performance of information processing machines and systems to continue to scale to keep up with the exploding growth of the use of information in society. Recommendation:  The U.S. government and private industry, in conjunction with academia, need to encourage the exploitation of emerging nanotechnology for the next generation of optics and optoelectronics for the dramatic enhancement of performance (size, energy consumption, speed, integration with electronics) in information communications, storage, and processing. Recommendation:  The optics and photonics community needs to position the United States in broadband to the home and business space. The U.S. government should pursue policies that will enable at least gigabits per second broadband access to the substantial majority of society at a reasonable cost by 2020. Recommendation:  A multi-agency and cross-discipline effort is recommended to identify the opportunities and optical technologies to significantly increase the en- ergy efficiency in communications networks, information processing, and storage. In addition, new ideas for the use of energy-efficient optical approaches to displace current energy-hungry practices—for example, travel—should be identified and supported. Greater focus and support in this area, especially at the fundamental level where companies are less likely to invest and where payoff could be huge, will be important.