Peter Kaiser, PhD
Santec Photonics Laboratories
433 Hackensack Avenue, Hackensack, NJ 07601
Ph: 201 488 5505; Fx: 201 488 7702
The large-volume production of low-cost, highly functional, integrated and standardized optical components is a key prerequisite for the more widespread introduction of optoelectronic and multi-wavelength technologies in new network segments. This note describes the demand for optical components from Ultra-Long-Haul to Metro, Cable and Access Networks, to LANs and Dark-Fiber Networks, together with requirements and opportunities for application-specific optoelectronic components in those networks. The article is a summary of an invited paper presented at the IEEE LEOS Topical Meetings in Copper Mountain, Colorado, July 30 to August 2, 2001.
Due to exponentially growing Internet and data traffic which may comprise as much as 99% of the total traffic in the next 3 to 5 years, information networks are evolving from SONET/SDH-based ring networks to more cost-effective, packet-based mesh networks with ITU-wavelengths-compatible DWDM interfaces of the IP (Internet Protocol) routers. To further reduce the cost of bandwidth in the core networks, the trend is to eliminate opto-electronic regenerators and SONET multiplexing equipment as much as possible and to deploy large-channel-count DWDM systems together with wideband optical amplifiers extending up to several 1000 kms in Ultra-Long-Haul (ULH) systems. Access to the 2.5 and 10 Gb/s channel rates (and 40 Gb/s in the near future) is achieved with optical add-drop multiplexers (OADMs) and currently typically opaque optical crossconnects (OXCs) with electronic switching matrices. Both the low speed of electronic switching matrices and the high cost of the transponders required for such switches is gradually leading to their replacement with transparent OXCs with all-optical switching matrices.
The deployment of ULH DWDM transmission systems at 10 Gb/s and particularly at 40 Gb/s signal rates requires the use of low-nonlinearity and low-dispersion optical fibers with dynamic chromatic and polarization-mode dispersion compensation, and wideband EDFAs and Raman amplifiers with dynamic gain equalization. A reduction of the effects of fiber non-linearities can be achieved through new fiber designs (e.g., large-effective core area fibers) and through a reduction of the amplifier pump power via distributed Raman amplification and through Forward Error Correction coding.
Metro and Access networks are considered the next major growth area for optical networking technologies. The drivers for these services are fast-growing e-commerce and e-business, and increasingly Web-based entertainment, multimedia, general and customer-specific information and educational services in addition to conventional voice and data services. In addition, Napster-like peer-to-peer communication and information exchange will rapidly increase the user demand for larger-bandwidth transmission with an associated increase of traffic in the Access, Metro and core networks. Metro networks with transparent OXCs and multi-service platforms will be able to route different protocol LAN/WAN, Enterprise, Storage Area as well as multimedia and digital video traffic together with conventional SONET signals.
Much is being said about the bandwidth glut in the long-haul network where both incumbent and new network operators and service providers have installed the latest fiber and DWDM systems with up to 10 Gb/s SONET/SDH rates on each wavelength, and resulting capacities in the 100s of Gb/s range to multi-Tb/s on a single fiber. However, there is a severe bottleneck in accessing this immense transport capacity and potentially very low cost bandwidth by end-users since currently only about 12% of the 68 million Internet users in the United States have high speed Internet access. This access is currently provided by highly Asymmetric Digital Subscriber Lines (ADSL) and Cable Modems that are practically limited to not more than 500 kb/s speed per user in the downstream, and only about 50 kb/s in the upstream direction. The increasing use of ADSL and Cable Modems, however, is stimulating the demand for symmetric higher-speed and broadband access for large-bandwidth-consuming video, interactive multimedia and file-transfer services. This should lead to the deployment of fiber closer to the customer through the use of FTTC and FTTH by telecom operators, and fiber-fed smaller Service Areas by Cable Network operators involving only 75 to 300 customers (compared to 500 to 2000 users per service area today). As a result, the number of Cable users served via so-called Fiber mini-Nodes (FmN; Ref: AT&T Broadband) approaches the 16 to 32 customers served in typical APON-based FTTC systems by Telecom operators. This may lead to a convergence of the optoelectronic component technologies for both applications such as common-design ONUs (Optical Network Units) - with a resulting increase of the manufacturing volume and an accompanying reduction in the manufacturing costs. It should be noted that the FmN concept includes three or more wavelength WDM transmission to the mini nodes, and also allows for FTTH connections upon customer demand.
The use of WDM in the Access network can provide services from different service providers to a customer on the same fiber, or it can be used to route selected services to different customers in a flexible way. Because of its cost-effectiveness (e.g., requiring no thermo-electric cooling of the lasers), low-cost coarse WDM (CWDM) technology with 20nm channel spacing may advantageously be deployed in the Access network already in the near future, while DWDM channels within some of the CWDM bands could be provided for additional services to selected customers later on.
Broadband access can also be provided with low-cost Ethernet LANs up to 1 Gb/s now, and 10 Gb/s (10GigE) in the near future. In fact, due to its low cost, optical Ethernet may become the preferred transmission protocol not only for high-speed LANs/WANs but in the subscriber loop as well, once improved network reliability is being achieved through Resilient Packet Transport (RPT, see IEEE802.17). The increasing deployment of dark-fiber networks where both the fiber and terminal equipment is owned and operated (under contract) by user communities rather than Telcos - for example with Ethernet protocol - is already resulting in bandwidth offered at a fraction of the price of conventional high-speed services today (e.g., $1000 per month for 100Mb/s Fast Ethernet connections, or $10 per Mb/s - Ref.: Cogent Communications). This trend is expected to accelerate with the commercial introduction of 10GigE in 2002, whereby 10GigE transceivers are predicted to be available in the $1000 range in the near future and around $500 within the next three years (at approx. 1/7th of the cost of SONET transceivers; Ref: R. Haitz, Agilent).
While the component functionality in Metro networks is similar to that in long-haul networks, the key for success of optical networking technologies in the Metro and also the Access networks is a high level of integration of low-cost, robust, highly reliable and standardized optical components that are designed for mass production. Furthermore, as the number of DWDM channels increases into the several hundred channel range, it is necessary to design tunable devices such as lasers and filters in order to minimize the number of manufacturing codes, allow cost-effective sparing for maintenance purposes, and also permit increased optical networking functionality.
Specifically, optical component costs may be reduced through the following improvements of the component design, manufacturing or processing steps, or other means:
Great emphasis is recently being placed on automated manufacturing as the component volume increases from several 10,000 to 100,000 units or network elements in the long-haul network to potentially several hundred million devices in the Access Networks and LANs - where the sensitivity to costs is paramount. As the number of components also increases due to large-channel-count DWDM systems (such as 80 or 160 wavelengths per fiber, and beyond), component designs will have to be chosen that allow easy scalability while maintaining small size overall. Good examples are Arrayed-Waveguide-Grating (AWG)-based components fabricated with photo-lithographic techniques, and VCSEL devices which permit wafer-scale processing and testing. Potentially very-low-cost components with different functionalities are also being designed with MEMS (Micro-Electro-Mechanical Systems) technology, and emphasis is being placed on using MEMS for 2D and 3D optical switches, variable attenuators and filters, as well as long-wavelength VCSELs for low-cost fixed-frequency and tunable lasers (together with MEMS technology).
While wideband Erbium-doped fiber amplifiers and Raman amplifiers continue to be improved for ever more demanding applications in long-haul and ULH networks, compact and low-cost EDFAs with more limited performance parameters and semiconductor optical amplifiers (SOAs) are being developed for use in Metro, Cable and Access network applications.
Another important aspect for the more widespread acceptance of advanced optical networking technologies by Network Operators and Service Providers is the need for low-cost, software-driven OAM&P (Operation, Administration, Maintenance and Provisioning) of new optical network elements in the Metro and Access network environment and which frequently may have to be smoothly evolvable from, and interface with legacy network technologies and networks.
In summary, low-cost optoelectronic components comprise the enabling technologies for the ever more widespread application of optical and WDM technologies in next-generation optical networks, with particular emphasis on Metro and Access networks. This is expected to exhaust the bandwidth glut in the core network and bring affordable high-speed and broadband services to end-users directly - and thus propel the optical networking and information technology industry forward on its continuing path to success.
return to contents