Reshaping fiber`s "last mile"
A new technology that combines sdh-based optical transmission with electrical distribution could lead to a variety of opportunities for fiber in the local loop.
Mitchell Cotter New World Paradigm Ltd.
One of the lecture sessions at the recent Distributech conference (January 13, 1998) was "The Promise of Fiber: Real or Imagined?" A new, hybrid, "last-mile" cable that carries both optical and broadband electric signals to end-users provides an affirmative answer to the question. The cable delivers a substantial amount of the bandwidth for end-users, which has for so long been only a potential of fiber-optic technology. This cable and the system in which it operates provide each user with two separate, continuously active paths (a send and a receive), each of which carries a 622-Mbit/sec channel that operates much like Synchronous Digital Hierarchy (sdh) sts-12.
While current protocols operate very much via a connection-based concept, this new system does not have "dial- tone" type operation. A user "sends" whenever his terminal gear places the information`s address bytes into the sdh data frame along with the information bytes to be sent. In a like manner, whenever the user`s terminal receives a packet set of data addressed to the user it decodes and directs it to the appropriate end device attached to the user`s system.
Many such individual simultaneous "messages" are possible within this structure--and this data flow may be active without any human interaction, 24 hours a day. The number of such activities is so large that communications could become a very different process from what is commonplace today. Each user could easily have half a dozen voice channels ("telephone" lines) and barely dent the system`s capacity. Even several simultaneous multi-megahert¥wideband Internet sessions would still only lightly tax the system`s capacity. When the architectures of information transmission systems are re-examined from this perspective, many advances are possible. Availability of such bandwidth on an economical basis moves communications into new territory.
The "last-mile" cable project began over five years ago with an analysis of the basic system limitations in last-mile communications. The new approach focuses on solving the last-mile problem for electrical signals because we believe that cost-effective all-optical systems for the last link to the user will remain elusive for many years. Some new electrical signal propagation techniques were discovered that make wideband performance possible. These highly proprietary developments also involve new materials and processes. The cable carries these electrical wideband signals no more than 2 km from a local node--i.gif., the "last mile." These nodes are linked by fiber-optic paths to their "hub." In preparation for the future use of optical transmission to the user, the new "last-mile" cable includes four singlemode fibers. Fiber is relatively low in cost, and the opportunity to provide for the future is easily accomplished--which is especially important because installing the cable is the most costly part of such an effort.
By providing electrical wideband paths in an appropriate network architecture, wide-bandwidth information capabilities become immediately available to the largest number of users. The figure shows a typical group of local nodes distributed about a service area, which may be some portion of the neighborhood serviced by a switching and transfer point (stp). The stp is the hub of such a local distribution system. Here, the information packets for each of the node`s user sending frames (sts-12) are directed outward onto the appropriate paths of the global network to their intended addresses. This same stp site reads incoming packets addressed to users and directs them to their destinations. This is a somewhat higher level of routing function with a few new twists. If no messages are going out or coming into the stp, then the entire local feeder network would be running "empty." However, the level of activity will in all likelihood be vastly greater than that at which present networks operate. To this point, the many new kinds of unattended services, wideband video, and other possible functions have not yet entered the communications realm.
The architecture may employ multiple wavelength-division multiplexing (wdm) fiber-optic links between the local nodes. Each node will feed several wideband last-mile drops to users. Present designs serve 64 to 1000 users per fiber for each direction in the local node loops. Such links may be rings or multiple tessellating networks of fibers that connect the nodes to the stps. It is not yet clear what combination of wdm and multiple fibers will be deployed because it appears to depend on the density and number of users in an stp group.
Whatever number per fiber is employed, the system provides economical performance at all these levels. The costs are adequately recovered and the system operates profitably at only very moderate basic charges per user--no more than $100 per month for connection, wideband Internet, basic video, and "phone" service. For all the reasons discussed, this new architecture invites new ideas in pricing for the users and services connecting to the network. In any case, there will be much more fiber required as the new system comes into full operation on a global scale.
Possible future scenarios
With everybody having this amount of bandwidth and since the paths are continuously active, what happens to the volume of data flowing in the network beyond the stps--the existing long-lines carriers and "interoffice" paths? Very little in the beginning, if data traffic remains the same. However, the amount of data being sent and received will likely soar at a rate never before experienced, even with the recent Internet expansions. But current concepts of what communications is all about are still dominated by the architectural legacy of connection-based, human-to-human functions. This is no longer an adequate description of what is happening. Facsimile and computer data transfers already account for the majority of data flow. In the future, this shift will certainly increase substantially as voice becomes less and less of the total traffic.
However, what this development enables are new kinds of communications. Vast numbers of new services and functions for both individuals and businesses are enabled by the permanent, continuously active "information highways" created with this last-mile approach. Today the world`s total data-transmission rate is 1013 to 1014 bits/sec. Estimates of the impact of the new last-mile network technology on that transmission rate suggest a rapid ascent within the next decade to more than 1019. Such data flows ultimately require great expansion of long-line fiber-optic capacities. Extensive expansion of data packet-switching capacities throughout the world must follow. The new 128-bit address structure for the new Internet (v6 proposed) is an example of the kinds of changes already afoot that make these opportunities very close at hand. Along with this is the rapidly evolving multi-casting network architecture and protocols that will make all manner of broadcast possible.
This will call into action all of the high-density wdm technologies that can be applied and further push the efficacy of the optical modulation techniques used by optical carriers. Optical signal modulation techniques are only now beginning to develop the sophistication equivalent to that which the pressure of signal density produced years ago in RF electromagnetic propagation. Few optical carriers today even reach the sts-192 level of 10-GH¥double sideband modulation of the optical carrier. Fiber itself is relatively cheap and it is yet to be seen just what combinations of multiple fibers and modulation strategies will provide the best economics and signal performance. The optimum will certainly vary with the particular service and length of the paths.
Important in the overall architecture of this new realm of communications is the need for increased system reliability. This need is not limited to the requirement for path redundancy, as with ring and multiple tessellation topologies. Individual network elements must rise to new levels of reliability if the network envisioned is to succeed in providing all of the new services that should operate without the close attention of the people using them. Communications is reaching so deeply into the fabric of the activities and critical needs of the world that it demands significantly more reliability than has ever before seemed necessary.
The need for reliability and the permeation of communications into every part of the economy and society will continue to reverse the trend that had seen optical-fiber manufacturers forced to close plants or reduce production. The development of this new system architecture and the rapid deployment of the hybrid cable for the short lengths of last-mile drops require millions upon millions of feet of fiber. To this must be added the node-feeding redundant paths (also many millions of feet) and the considerable fiber expansion of the global network capacities. In this latter vein, new organizations are already entering the field of providing long-line fiber capacity to all comers.
There exist still other performance opportunities in the choice of fiber materials that can provide other characteristics for the differing service distances and operating bandwidths that the new architecture implies. The lessening of optical loss, so often pursued as a prime objective, may not be nearly as relevant for many such needs, opening possibilities for new materials and methods. Possibly more important may be an emphasis on bandwidth, other spectral transmission windows, dispersion, durability, and environmental performance.
At the user`s end of each last-mile connection is the interface system coupling the 622-Mbit/sec paths to the various systems and devices of the user. The quantities of such "boxes" or computer-like subsystems that potentially will be necessary are staggering. Every user would require such a function distributor. The very size of these applications generates new economies of scale that attract even the semiconductor industry, which is used to large-scale investment for volume production. Indeed, integrated circuits and other semiconductors would be required in greater magnitudes.
The amount of fiber interface equipment in the many millions of nodes to service the user drops also reaches new proportions. This appears to be just what was required to reach cost economies of tens to hundreds of times in these functions. One need only consider the use of lasers in CD players to see the rapidity and economy achieved with substantial volume. The volume required to fulfill communications needs dwarfs that of laser use in CDs.
Summary
These trends suggest that all communications needs can be best met by a single connection that brings together all needs: voice, video, video-on-demand, Internet, intranet, local-area and wide area-networks, enterprise connections, and the host of many new services that can flow back and forth to users without human attention. As significant as the explosive expansion of the Internet has been, the development of new "passive" services will astonish us.
The sponsor of this New World Paradigm development is khamsin Technologies, San Diego, CA, which will establish a demonstration and test facility of the architecture described in this article and the new last-mile cable in a small-scale network system so that licensed vendors will be able to use the facility to develop and test services or other products to ensure compatibility with the last-mile technology. Continuously connected, symmetrical, wideband, two-way, sdh-based packet Internet protocol-like channels for everyone play a crucial role in defining the potential of future fiber-optic communications. u
Mitchell Cotter is chief scientific officer at New World Paradigm Ltd., Raleigh, NC.