Delivering on the Promise:
Scenarios for Deploying Local Access
Branko J. Gerovac
David C. Carver
Center for Technology, Policy and Industrial Development
Massachusetts Institute of Technology
to appear in the book The First 100 Feet
Harvard Information Infrastructure Project Series, MIT Press
Table of Contents
If you had $1 billion, or even $100 million, what would be a productive investment that would significantly advance broadband services to homes and small businesses? Where is the real opportunity?
A decade has passed since industry first promised broadband to the home. Fiber To The Home was then a favored choice, that is, until the cost of deployment was taken into account. Since then, a few well known favored approaches have emerged.
Each time a particular approach gains favor its promise is reiterated with increased fervor as if to say "this time we're passed the hype, we're going to do it for real". Yet, each time the deployment scenario falls short. Either the technology does not scale and is overtaken by events, or a viable business plan fails to emerge; regardless, public policy issues such as universal access and privacy loom on the horizon.
Our guiding principle is to bring the kind of network connectivity enjoyed at larger institutions to homes, small businesses, schools, clinics, and other such smaller enterprises. For several years, we have looked at local access from a broad perspective that incorporates the complex interplay of technology, business, and policy.
Realize that the current Internet is fundamentally an institutional network. The Internet interconnects institutions. Many talk about empowering the individual, but everyone currently gains access to the Internet through some institution, whether it be from a university (as we do), an employer, or through an Internet service provider (ISP). It is simply not effective currently to do otherwise.
In bringing advanced services to the individual, it is important to know what the objective is and when it has been reached. Thus, we developed and evolve a set of criteria as a tool to get beyond the hype and recognize what has been achieved[1,2]. Here, we present a subset of the criteria and examine how some approaches fair against the criteria.
Certainly this kind of analysis is being done, but not publicly. Proponents of particular approaches naturally tend to highlight what they do well and to dismiss as unimportant criteria that they don't meet. This makes performing objective comparisons difficult. The promise of local access is better recognized by bringing the analysis out into the open. Establishing a comprehensive set of criteria sets the stage for deploying systems that are technologically feasible, that have viable business models, and that advance public policy and social objectives.
It is encouraging that people are beginning to recognize local access as an issue. As recently as a couple of years ago, when we raised the question, it was dismissed as being solved by ISDN. ISDN may have been an OK first step, but it does not offer the kind of bandwidth needed for many important classes of applications. A desktop connection at a large institution offers more than 10 times ISDN bandwidth. ISDN is not being deployed in large numbers. The difficulties in providing higher bandwidth services are being recognized.
Cable modems were one of the first significant steps to move potential capabilities forward. Now Continental, @Home, and others are conducting significant commercial trials and talking of approaching the magic 1 million households served.
ADSL is also moving forward. Since the beginning of 1996, ADSL has received increasing notice as an alternative to cable modems. ADSL technology is not as far along as cable modem technology, perhaps a generation or half a generation behind, but it is catching up.
No single magic technology will be the all around winner. Rather, a collage of technologies will be deployed in different settings depending on local circumstances.
Many of the local access approaches (whether ISDN, cable modem, ADSL, etc.) are really bottom up designs. They represent technology that was developed largely for other purposes. Yet, they are being touted for interactive data communications. Though it is appropriate to leverage existing infrastructure, we need to recognize the criteria these approaches do not meet. For example, one of our criteria categories covers ease-of-use and plug-and-play - existing approaches are not ready for broad consumer acceptance.
Also realize that these technologies are not an end in themselves but the beginning of a transition. For example, telcos and equipment suppliers have pointed out that ADSL can be deployed on the existing wire plant, then as concentrations of communications traffic emerge, fiber connections can be pushed out incrementally into the neighborhood, to the curb, and to the home.
However, when stepping through a series of transitions, it is important not to carry along initial annoyances in the service that later will be difficult to resolve.
We start with a broad list of criteria - the criteria are not industry specific, they are not application specific, they are not service specific. We then analyze a particular deployment scenario to see how it stacks up against the criteria and compares with other possible scenarios.
Below, we walk though a brief sample analysis using a subset of the criteria. The sample demonstrates not only how this kind of analysis can be used to compare the more talked about approaches (such as Cable Modems and ADSL), but also how it can be used to look at new opportunities (such as Microcellular Wireless) and their potential advantages and liabilities. Indeed, it would be a mistake to view this tool as simply a way of grading different approaches. Its real power is in how it can lead us to consider potential investment opportunities and risks.
Some of the criteria are traditional architectural goodness criteria. Anyone doing a good job of communications architecture and design always has these in mind. Briefly, they are:
* Cross-industry harmonization - Different industries represent different strengths and liabilities; no one industry can deliver a complete answer.
* Digital technology - Computing, communications, and media (including consumer electronics) are adopting a common set of base digital technologies potentially putting local access on the "technology curve".
* Open architecture and Interoperability - Systems should be open, to allow for competitive evolution of individual components, and interoperable, to ensure the efficient sharing of information (audio/video, voice, and data) across technologies, equipment, and services.
* Extensibility - The pace of technological change demands that forethought be given to accommodating future functionality in the initial design of a system.
* Scalability - Scalability is important to various aspects of the system: data stream transmission and reception, equipment price/performance, usage, provisioning, etc.
This is an "old mantra" that continues to be important and needs to be reinforced. Since these criteria have been around for a long time, they sometimes are taken for granted and even forgotten entirely. Sometimes proponents compromise these criteria in order to gain a false competitive advantage that ultimately hurts them and their customers.
Our criteria fit into 3 broad categories: Technology, Business Models and Markets, and Policy and Regulation. Here we outline a subset of our longer list of criteria:
* Critical Bandwidth Steps - Critical levels in network access performance at which entire new classes of applications become possible.
* Bandwidth Symmetry/Asymmetry - Can the infrastructure be designed always to accommodate symmetry regardless whether early deployment constraints warrant some degree of asymmetry - i.e., a symmetric ready system?
* Peer-to-Peer versus Master-Slave - The Internet is inherently peer-to-peer; telco and cable systems are inherently master-slave; ...
* Privacy, Security, and Network Management - Sharing an infrastructure, especially one that is open, with neighbors and local institutions exacerbates privacy, security, and network management issues.
* Mobility - More and more, people and their activities are becoming mobile - not tied to a particular office, building, or geographic location.
* Product Technology - Many component technologies (microprocessors, DRAM, mass storage, etc.) are following well defined evolutionary curves. Local access is not yet on a curve. Some approaches are a step in the right direction, others are not.
* Deployment Realities - Existing and proposed local access systems should support incremental deployment and subsequent incremental upgrade and provisioning.
* Capital Investment - Investment models represented in the different scenarios need to be reconciled with various industries' business practices.
* Market Transformation - Television, telephone, and computing as distinct services give way to a new model where the significant distinction is between content, communications, and interaction.
* Regulatory Regime -The single most important endeavor government could undertake is to set and clearly articulate national and public policy objectives (the ground rules). Without this, the heavy private investment required will remain inherently risky and ineffective.
We've looked at the criteria with respect to many potential technologies and scenarios. Here we discuss four scenarios and show how the subset criteria help you look at, compare, and contrast the four scenarios:
* CableCo / Cable Modems - Cable has been the long standing bandwidth leader. For at least 10 years, even before hybrid-fiber-coax technology (HFC), there has been talk about the coax cable plant as a vehicle for delivering high bandwidth into the home. The core technology was used in military communications more than 2 decades ago.
* Telco / ADSL - ADSL on the existing twisted-pair copper-wire telco plant, though originally oriented toward video on demand (VOD), is being positioned as Telcos' answer to the cable modem. ADSL is an important first step for the telephone companies.
* Fiber to the Home (FTTH) - FTTH has always been a future promise. FTTH is characterized as "not cost effective now, but will be in the near future." Because of transition technologies such as cable modems and ADSL, fiber is not as prominent in discussions now in the U.S. as it is, say, in Japan where they are installing new infrastructure.
* Microcellular VHF - As a wildcard scenario, what if we were to retire 20 channels of television spectrum - prime spectrum that goes through walls and buildings - and reallocate it for interactive data communications. What could we achieve? Easy mobility and access across geographic locations in the same spectrum space, etc. However, the technology to deploy microcellular is not quite here. If you're looking for a place to invest in technology, this particular form of microcellular could be high leverage.
The following tables summarize our analysis. Look at criteria across the selected scenarios to see what are and are not important differentiators.
Look in detail at the first two criteria, critical bandwidth steps and bandwidth symmetry/asymmetry. An important aspect of bandwidth that we've recognized in our work for at least 15 years is that bandwidth effectively comes in steps where each step enables a new class of applications. We refer to this as the "1,3 Rule" because the steps roughly follow an exponential progression of 1, 3, 10, 30, 100, 300,... The following chart shows the primary applications associated with each step. Of course, there is some overlap from step to step, and these are not hard and fast boundaries, but rules of thumb. The chart continues beyond 30 Mb/s to ever more involving applications.
"1, 3 Rule" of Thumb
There is an important interplay between critical bandwidth steps and the second criteria bandwidth symmetry/asymmetry. ADSL and some cable modem approaches are inherently asymmetric. Though the down stream channel is fairly high bandwidth, the upstream channel is relatively low bandwidth. Thus, inherently symmetric applications are precluded, such as high quality video conferencing that grandma would be comfortable using to view the grandkids.
The critical-bandwidth-steps concept has important implications to services that spread across a range of steps. ADSL performance is limited by crosstalk in the wire bundles and varies widely depending on distance from the central office and line conditions. Cable modem performance is limited by the degree of sharing encountered at any given instant and thus can vary widely due to congestion. This is more than a "your mileage may vary" issue. Certain applications may work for your neighbor but not for you, or you may find you can look at video clips at 2am but not at 7pm.
Indeed, sooner or later local access systems will commonly support video conferencing and the ability to originate graphically-rich multimedia content directly from home servers. We estimate this crossover point to be a ~3Mbps symmetric service. This is the point at which a user can both originate and receive at least one of any given data type that is in common use today. Above 3Mbps becomes a matter of how many simultaneous data types can be serviced.
This may seem like setting the bar at an outrageous height today, especially considering the limitation of current home equipment. Given the rapid pace of technology, it's really only a matter of when this happens. Thus, this sets the timeframe in which investments sunk into interim approaches must show their return.
This is not to say that asymmetric service cannot be initially deployed or that it would not be a substantial improvement on the kinds of service currently available. The question is when will it be technically feasible and competitively necessary to provide a symmetric service and how do you position yourself in the meantime. Even if you start with an asymmetric system, you want to look for ways to deploy a system that is architected to be symmetric ready regardless if it is initially provisioned as such.
Some criteria are not substantial differentiators between approaches. For example, compare product technologies - whereas a few years ago no practical solutions existed, now a number of technologies are becoming cost effective and viable. Though encouraging, none of these technologies seems to be a breakaway universal solution. ADSL modems could cost half as much as cable modems, but you need to deploy twice as many. Microcellular modems can be expected to follow the same cost curve as cable modems. Any of these technologies can be deployed as appropriate to the situation. The prospect that multiple solutions will likely move forward raises important interoperability and portability issues.
Consider capital investment and deployment realities across the selected scenarios, and recognize the very different business models under which these industries operate.
Cable is generally debt financed, so in order to invest in upgrades the company needs to issue debt. Bankers look to justify new debt largely on the sales and maintenance of conventional cable television services. On this basis, cable companies can usually justify building fiber out to 2000 home nodes where at 10% penetration a user effectively shares an Ethernet with 200 other potential users. The next step, 500 home nodes, can be achieved by upgrading node lasers and electronics to get 40% penetration with the same level of sharing. To improve quality of service, the level of sharing must be reduced: by reducing penetration, perhaps by charging more for the service, by further subdividing the plant, and/or by adding channels.
Telephone companies, on the other hand, are largely self-financed, putting large amounts of upgrade cost into the rate base. Since ADSL adapts to line conditions, it claims potential continuous deployment up to an estimated 60% penetration. Revenue at these higher penetration levels likely justifies building out fiber closer to the home.
We can also look for new scenarios that exhibit more of the beneficial characteristics we are looking for. In the microcellular scenario, we speculate using retired television spectrum to deploy a new data service. Upcoming digital television transmission, because it is more spectrum efficient, promises to greatly enhance the services that broadcasters can offer and at the same time release a substantial portion of spectrum for new uses.
This approach could remove the partition between mobile, portable, and stationary systems. Many of the necessary component technologies exist to build a microcellular system: wireless LANS, packet radio networks, and digital broadcast systems. Such a system could be symmetric capable and easily offer shared access to 10s of megabits per second within each cell. Microcellular product technology would follow similar cost curves to cable and ADSL modems. Cells size would be scalable to permit incremental service deployment and upgrade and to optimize frequency reuse - initially a small number of cells, and subsequent cell subdivision to meet increased demand.
The reason for pursuing television spectrum is that it has a number of good transmission qualities, e.g., it penetrates walls. Spectrum currently considered for data services is poor quality - it is non penetrating and typically requires line of site transmission, which nullifies many of the sought after benefits of wireless.
From a business perspective, this scenario requires significant upfront investments in developing the technology, in petitioning and lobbying the FCC for spectrum allocations, and in paying for auctioned spectrum licenses. Obviously, this approach is speculative, perhaps even a long shot, but the potential payoff is substantial making it a possible candidate for venture capital or the equity markets.
From a policy perspective, it is useful to draw a comparison with the origins of television. Fifty years ago the federal government allocated a large chunk of high quality RF spectrum to provide free and universal access to what was then a non-existent service: television. And, for better and worse, television has had an enormous impact on our society. It was truly a high leverage investment. Even though today cable and satellite television systems are available, terrestrial broadcast television is still considered important.
What could the government do today to achieve a future payoff of this magnitude? Many of the same arguments used 50 years ago (e.g., creating an informed citizenry) and recast today for terrestrial broadcast television can be applied equally as well to creating a wireless universal data service. This would go beyond creating an informed citizenry to creating an involved citizenry. Given the anticipated benefits, it is certainly worth allocating high quality spectrum to promote a National Information Infrastructure.
|Critical Bandwidth Steps
"1, 3 Rule"
|10 Mbs down / 10 Mbs up
30 Mbs down / 3 Mbs up
|ADSL: Up to
8.192 Mbs down /
640 Kbs up
|51 or 155 Mbs (fractional)
|Bandwidth Symmetry/ Asymmetry||Choice of symmetric & range of asymmetric||Inherently asymmetric||Symmetric||Potentially
|Privacy, Security, and
|Shared link in neighborhood
Hard to isolate catastrophic failures
|Dedicated link more private and secure
Slight regulatory advantage
|Mobility||Speculative...||Only cellular offers inherent mobility|
|Product Technology||Three chip solution
Twice the cost
|Single chip solution
Twice the modems
|$30-50 loop fiber xmtr/rcvr||Undefined technology
Expected costs comparable to cable modem
|Deployment Realities||HFC promises a single plant for TV, telephony, and data
upgrade DLCs, run new fiber where necessary
(long term: FTTH)
|Extensive initial build
|Initial cell sites
Incremental cell subdivision
|Cable Co||Telco (xDSL/FTTH)||FTTH||Microcellular|
|Capital Investment||Initial plant upgrade justified on improving CATV costs and service||85/15 rule 85% of upgrade taken out of rate base (California formula)
|Massive up front investment||Up front spectrum auction cost|
Broad shift to separating carriage and content
Internet is inherently a non-brokered service
History of content provider relationships
A national broadcast infrastructure (high bandwidth)
||Potential for true peer-to-peer system|
Transition from service based monopolies to competition
|Municipal regulation (telecomm act raises questions) Federal and State regulation||Common carrier protection on 1st Amendment issues||Investment protection?
Old regime: regulated monopoly
The full variety of solutions for providing high bandwidth local access services have yet to be adequately explored. Even the brief analysis above suggests a few interesting observations. For example, the microcellular approach stands out in that it addresses many of the criteria other approaches do not. Thus, it is a worthy candidate for investment and national attention. What other interesting approaches have yet to be considered?
Our work towards evaluating alternative approaches is based on an open set of objective criteria. (1) An open and objective set of criteria clarifies what is meant by local access, and provides a context in which to discuss the issues. (2) Analysis using the full set of criteria dispels hype, and exposes the degree to which each approach can fulfill its promises. This enables meaningful comparisons across proposed alternatives, suggests transition strategies for existing systems, and encourages the creation of what is really needed to deliver on the promise of local access to the information infrastructure.
This work is sponsored in part by the Defense Advanced Research Projects Agency and the National Science Foundation under contract NCR 9423889.
Branko J. Gerovac and David C. Carver are in the Center for Technology, Policy and Industrial Development at the Massachusetts Institute of Technology. They are codirectors of the Research Project on Local Communications Infrastructure. They investigate the interplay of technology, business, and policy, in the development of the National and Global Communications Infrastructure.