Cover Image

PAPERBACK
$118.00



View/Hide Left Panel

Page 304

38
Residential PC Access:  Issues with Bandwidth Availability

Kevin C. Kahn, Intel Corporation

Abstract

The preeminent interactive information access device in the business world today is clearly the personal computer. Via the PC, individuals access the exploding array of information sources both within their businesses and throughout the Internet. PCs are rapidly penetrating the consumer environment as well, and we strongly believe that the PC will become the center for national information infrastructure (NII) access for the residential consumer. The intelligence encoded in PC applications and their improving human interfaces will make PCs the tool of choice for ubiquitous consumer information access. However, to achieve their great potential in this role, residential consumer PCs must have access to adequate communication bandwidth. This bandwidth must be priced to be attractive to the residential consumer. The consumer must be free to access all information services using this bandwidth.

High-bandwidth access to information services must get the same level of attention with respect to public policy as providing competitive television and telephone delivery systems. This attention is needed to ensure that the bandwidth, access, and consumer choices are made available in ways that promote the growth of consumer NII use.

This paper develops these themes by examining the forces that are driving corporate PC network access to the NII and the various network topologies being discussed for the residential environment (HFC, FTTC, etc.), as well as possible future service directions. We then develop what we feel are the critical requirements for bandwidth availability for residential PCs.

Statement of Problem

The personal computer has become a ubiquitous networked communications platform within the business environment. In addition to traditional communications applications such as electronic mail, it is now becoming the base for all sorts of information access and personal communications tools. These tools enable a new level of business activity that includes everything from World Wide Web access for general information acquisition to various levels of electronic commerce and personal conferencing. Key to this development of widespread, cost-effective information and communications applications have been a number of important technologies. First among these has been the deployment of high-bandwidth network connections in both the local and wide areas, utilizing high-volume components, based upon open standards. Free and open access to this bandwidth has permitted any software or hardware developer or information service provider to easily enter these businesses. For example, there are numerous suppliers of network connection hardware and software, a growing number of competing conferencing products, and the beginning deployment of multiple on-line news services from traditional and nontraditional information publishers utilizing the World Wide Web. The resultant competition and interactions are leading to rapid development of the business use of the developing NII.

The same dynamics must be allowed to operate within the critical residential environment to support the development of individual utilization of the NII. While there has been a lot of focus on the larger infrastructure developing to support the NII, particularly in the context of the Internet, there has been less focus on the data access issues of the ''last mile" or access network to the residence. What discussion has occurred seems to revolve largely around video-on-demand entertainment services and telephony. We are concerned that while public policy may operate to guarantee competitive entertainment services and telephony over cable or telephony infrastructures, it might not guarantee the deployment of the reasonable levels of openly available, discretionary



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 304
Page 304 38 Residential PC Access:  Issues with Bandwidth Availability Kevin C. Kahn, Intel Corporation Abstract The preeminent interactive information access device in the business world today is clearly the personal computer. Via the PC, individuals access the exploding array of information sources both within their businesses and throughout the Internet. PCs are rapidly penetrating the consumer environment as well, and we strongly believe that the PC will become the center for national information infrastructure (NII) access for the residential consumer. The intelligence encoded in PC applications and their improving human interfaces will make PCs the tool of choice for ubiquitous consumer information access. However, to achieve their great potential in this role, residential consumer PCs must have access to adequate communication bandwidth. This bandwidth must be priced to be attractive to the residential consumer. The consumer must be free to access all information services using this bandwidth. High-bandwidth access to information services must get the same level of attention with respect to public policy as providing competitive television and telephone delivery systems. This attention is needed to ensure that the bandwidth, access, and consumer choices are made available in ways that promote the growth of consumer NII use. This paper develops these themes by examining the forces that are driving corporate PC network access to the NII and the various network topologies being discussed for the residential environment (HFC, FTTC, etc.), as well as possible future service directions. We then develop what we feel are the critical requirements for bandwidth availability for residential PCs. Statement of Problem The personal computer has become a ubiquitous networked communications platform within the business environment. In addition to traditional communications applications such as electronic mail, it is now becoming the base for all sorts of information access and personal communications tools. These tools enable a new level of business activity that includes everything from World Wide Web access for general information acquisition to various levels of electronic commerce and personal conferencing. Key to this development of widespread, cost-effective information and communications applications have been a number of important technologies. First among these has been the deployment of high-bandwidth network connections in both the local and wide areas, utilizing high-volume components, based upon open standards. Free and open access to this bandwidth has permitted any software or hardware developer or information service provider to easily enter these businesses. For example, there are numerous suppliers of network connection hardware and software, a growing number of competing conferencing products, and the beginning deployment of multiple on-line news services from traditional and nontraditional information publishers utilizing the World Wide Web. The resultant competition and interactions are leading to rapid development of the business use of the developing NII. The same dynamics must be allowed to operate within the critical residential environment to support the development of individual utilization of the NII. While there has been a lot of focus on the larger infrastructure developing to support the NII, particularly in the context of the Internet, there has been less focus on the data access issues of the ''last mile" or access network to the residence. What discussion has occurred seems to revolve largely around video-on-demand entertainment services and telephony. We are concerned that while public policy may operate to guarantee competitive entertainment services and telephony over cable or telephony infrastructures, it might not guarantee the deployment of the reasonable levels of openly available, discretionary

OCR for page 304
Page 305 bandwidth required to energize an affordable and comprehensive personal computer-based information infrastructure. The deployment of bandwidth that can be utilized in an open manner to access NII services from the home needs to be encouraged. For data, this access should be via general packet protocols (based on the Internet standards of TCP/IP) that permit users direct interaction with any service they desire. Specifically, data services from the home should come to be viewed like existing telephony services that allow consumers to access any services anywhere in the nation. It may be that for other multimedia services, such access would be better provided via circuit-oriented services such as switched ATM virtual circuits (although promising work is ongoing on utilizing packet protocols for this as well), but in any case the same principles of openness and full connectivity must apply. We also believe that the development of industry-driven standards or implementation agreements for attaching to and utilizing this open bandwidth will be critical to the creation of a competitive environment for consumer information applications. The government role should be to encourage rapid industry convergence on such de facto standards or implementation agreements that can later be the basis of more formal de jure ones. This paper attempts to lay out some requirements for the development of consumer PC information services broadly within the NII. Background The Business Environment as an Illustrative Example As a point of reference let us first consider the typical business access being deployed to allow PCs to utilize the developing NII. The typical networked PC in the office is attached to a 10-Mbps Ethernet or similar performing Token Ring network. (International Data Corporation estimates that in 1994, 73 percent of the business PCs in the United States were attached to local area networks.) While this bandwidth is shared with the other users of the network in most cases, switched networks are beginning to be deployed that replace the shared access with dedicated access. In addition, technologies such as 100-Mbps Ethernet are appearing that will also greatly improve the bandwidth available to the individual user. Beyond the local environment, most large corporations are deploying reasonable bandwidth into the Internet at large. This combination makes available relatively high-speed access to information and services whether local to the business desktop or remote. In addition to the bandwidth that is available, another key aspect of the business desktop is the development of standards. Industry-driven network connection standards have driven down the cost of connecting to the network from the PC. For example, Ethernet controller cards for PCs now typically sell for less than $100, operating systems increasingly come with protocol software built in, and a growing number of applications comprehend networking capabilities. Platform software standards have made it possible for creative developers to build software independent of the nature of the specific network. For example, the Winsock de facto standard for accessing network services on the Microsoft Windows Operating System is allowing application developers to focus their efforts on enhancing function in their products rather than on adapting those applications to a variety of different, incompatible, network services. Network protocol standards have allowed end-to-end services to operate over a wide variety of network implementations. In particular, general packet networks have allowed a wide variety of data services as well as new applications such as video conferencing to begin to operate without special provisions from the providers of the networks. An entrepreneurial developer need not make deals with a variety of platform and network providers to begin to deploy an application in this environment. Neither is an interested business consumer restricted from beginning to take advantage of such new services by the choices offered him by various network suppliers. Key aspects of the business environment are that it has been almost entirely industry driven and motivated by competition. De facto standards or implementation agreements have been rapidly developed and only later evolved into de jure standards. Throughout the evolution of the business and the associated standards, the intellectual property of the participants has been respected by the processes. It is interesting to note that the more ponderous de jure first approach to standards represented by the International Telecommunications Union (the official international body for defining standards within the telecommunications industry) has been considerably

OCR for page 304
Page 306 less successful in affecting the U.S. environment than have the more chaotic and rapid moving processes of private company consortia or the Internet Engineering Task Force (the de facto group that defines Internet standards). A couple of other interesting aspects of packet networks are typical in the business data environment. End systems are always available to be connected to—that is, they are always "online." It is thus reasonable to build applications that initiate connection to a user's system from a service at any time. This is different from an environment where, unless the user has initiated a connection to the network, information cannot find its way to the user's end system. Related to this is the fact that packet networks inherently support multiple simultaneous connections. That is, an end system can have an unlimited number of distinct applications running at the same time, each of which is using some of the physical bandwidth to connect to another distinct end system. There is no notion of the line being busy. Again, this flexibility opens up many possibilities for applications that operate in parallel with each other over the same physical connection. It allows service-initiated contact with an end system even while the user is doing other things over his connection. The Current Residential Environment We can compare this business situation to that seen in the residential environment, first by looking narrowly at what is really available today, and then more importantly at what is being tested and deployed for the near future. Today's electronic service to the home consists of a number of options. First among these is the existing plain old telephone service (POTS). This is universally available and provides a completely open but relatively low bandwidth access to data services for the consumer. Using 14.4-kbps and more recently 28.8-kbps modems, the consumer can connect to any information service or network. Traditional choices have been the private information services such as CompuServe, Prodigy, or America Online. However, there are a growing number of general Internet access providers available via this route as well. These provide general packet access into the Internet and thus to any information service in the developing NII. An improvement over POTS in terms of available bandwidth is an integrated services digital network (ISDN) line, which potentially provides up to 128-kbps access over a single wire. Fewer information service or Internet access providers have thus far deployed ISDN access, and the availability and practicality of utilizing this level of service varies considerably around the country. Like POTS, this is a circuit-oriented service that must be initiated by the consumer. That is, unlike the typical business connection, an information service usually cannot autonomously contact a consumer's PC to provide information or service. Also, while more flexible in some respects than POTS, ISDN is for the most part not useable by multiple separate applications simultaneously, thus further limiting the range of applications that can utilize it. ISDN can be used effectively to connect an end system to a general purpose router that provides a point of presence for the Internet or other general purpose network. Used in this manner, ISDN may provide our best short-term hope for general purpose residential access. Also, since ISDN can provide fast connections to another end system when compared with POTS, it can approximate the speed of being always connected for certain types of applications. While in principle POTS can also be used in this manner, the combination of increased bandwidth and connection speed makes ISDN much more practical in such a configuration. In contrast to these point-to-point telephony based services, the cable industry has focused on high-bandwidth broadcast services for video. The cable infrastructure is typically highly asymmetric in its bandwidth with much higher speed available into the home than out. Currently, in fact, most cable systems have no return path available. Also, cable is unlike the telephone where a consumer can call anyone or access any service on any other existing telephone system. The services provided today via the cable infrastructure are generally chosen by the cable system operator or according to the "must carry" rules. Cable systems are generally closed systems that encompass everything from the content source to the set-top box, which provides a conventional television signal to the appliance. There are beginning to be trials of data services over the cable infrastructure. However, in the spirit of the existing industry, these may tend to be for services driven and chosen by the service provider rather than the consumer. In the case of cable TV programming, the limited number of available channels means that the MSO

OCR for page 304
Page 307 must select which channels to broadcast. For data services, however, open packet access means that beyond the provision of the data bandwidth, there is no need to limit access to content in this manner. On the positive side, the cable infrastructure brings the promise of relatively near-term residential broadband service. Cable could certainly offer the advantages of high-bandwidth, multisession, always-on data delivery. In particular, by offering high-quality Internet access, cable could provide much of the flexibility we desire. It is important that PC access to broadband information services follow the telephone model rather than the broadcast TV model of access. Analysis, Directions in the Residential Environment, and Forecasts Access Network Technologies Two major groups of companies are vying to provide digital services to the residential environment: the existing local exchange carriers and the cable entertainment companies. The former are building on their expertise in operating a highly reliable telephony network by offering higher-bandwidth services through ISDN and through cable overlay systems that are capable of supplying broadband residential service. The latter are leveraging their existing investment in a physical plant oriented to deliver many analog television channels so that they can begin to offer digital entertainment channels and various levels of broadband data communications to the home. As a part of this enhancement of services they frequently wish to offer telephony services via this network as well. This competition should be a positive force for consumers in the long term, provided that it offers real competitive choices to the consumer as well as to service providers. A number of physical architectures have emerged for deployment by these companies. The most popular of these include the following: • Hybrid fiber coas (HFC): In this scheme, optical fiber is used to bring signals from a head end to neighborhood nodes. At these nodes, the signals are placed on a coax system for distribution to the residences. The cable industry target for residences served by a single coax segment is 125, although in early deployments the number is larger. Since the coax is a shared medium, the homes on a single coax segment share the available bandwidth. For most current deployments, the amount of bandwidth available from the home (either on the cable or via a separate POTS path) is much smaller than that available to the home. HFC is particularly attractive as an upgrade path from existing analog cable systems. • Fiber to the curb (FTTC): In this scheme, optical fiber is used to bring signals from a head end to a pedestal on the street that is within a relatively short distance of a collection of served homes. At the pedestal, signals are placed on coax or twisted pair lines to the home. While the optical fiber bandwidth is shared (as in HFC), the drop to the individual home is not. • Fiber to the home (FTTH): In this scheme, optical fiber is deployed all the way to the home, where it is converted as appropriate to provide digital services. At this point, this scheme is not being pursued to our knowledge in the United States, although it has been proposed at times in other national markets. • Asynchronous digital subscriber line (ADSL): For areas where the length of the lines from the last point of electronics to the served homes is bounded, ADSL provides a technique for utilizing existing telephony infrastructure to carry higher data rates. It uses more complex signaling electronics over existing copper to provide data rates capable of supplying digital video. Beyond these wired topologies, experiments are also beginning that use wireless technologies. For example, direct PC utilizing direct broadcast satellite technology provides an asymmetric bandwidth connection similar to using cable combined with POTS for a return path.

OCR for page 304
Page 308 Bandwidth and Parallelism A common theme among all of these approaches is that the allocation of bandwidth into the home is different from the allocation of bandwidth out of the home. From a purely architectural perspective this is unfortunate, since it is an asymmetry that makes an artificial distinction between providers (who need high outbound bandwidth) and consumers (who can get by without high outbound bandwidth). In the full NII where telecommuting will be common, video telephony will be widespread, and a true cottage information industry will become an important part of the national economy, this asymmetry will become a barrier. However, as a practical matter for the short-to midterm, the key will be reasonably high bandwidth into the home and moderate bandwidth out. This arrangement will cater to information retrieval applications where the bulk of the data flow is inbound with much smaller flows out to make queries. Provided that the outbound bandwidth is high enough, it can also support limited video telephony and telecommuting applications. We and others have demonstrated the practicality of such application at ISDN speeds (128 kbps) via products like ProShare™ videoconferencing and RemoteExpress™ remote LAN access. Nevertheless, higher rates would greatly improve performance as well as support more parallel activities. For the long term it is critical that residential services support a full range of simultaneous activities within the home. It should be possible for children to be involved in remote education activities while one parent is actively telecommuting, the other is accessing entertainment or shopping services, and power or security monitoring is operating in the background. Solutions that force artificial limits on conducting such parallel activities will make it impractical to depend upon network access as an integral part of the home. Furthermore, each of these individual activities may require simultaneous sessions. A child may be connected both to school and to a library system; a telecommuter may be accessing corporate databases while also checking travel arrangements with an airline. A shopper may be doing comparisons on products offered by competing providers. From a technical perspective, the solution to these multiple access issues is either the provision of multiple virtual circuits or a general packet network interface (or more likely some combination of these technologies). Openness of Access A key to allowing these applications, even in the face of the sorts of bandwidth available in the near-term deployments, is to put the use of the bandwidth completely under the control of the consumer. Consider two approaches to providing access to information services. In the first, a cable system contracts to make available one of the online service providers (say, CompuServe) using some of its digital bandwidth to the home. From the consumer's point of view a bundled package is offered that allows direct connection to the service for some fee. The service being provided by the cable operator is CompuServe and not generic online access. If consumers wish to subscribe to a different service, they cannot use the cable bandwidth to do it. Furthermore, if a new provider wishes to enter the business and have good access to customers, it may need to arrange business deals with many different network service providers. The comparable situation in the telephony industry would be that the range of people consumers could call from home would be affected by their choice of local or long distance carrier. The alternate approach is typified by the existing low-bandwidth telephony access system and the growing Internet service providers. Here, the customer gets access to the general packet switched network and uses that to provide common carriage of data to various information service providers. ISDN can be used in this configuration to provide moderate rate connection to a router from which general Internet connectivity can be provided. Note that the owner of the access network is not involved in the selection of the available services. That selection is between the consumer and the ultimate provider. Today, this access is the norm for telephony-class access. As we move toward higher-speed access networks, it is important that the model exist for broadband as well. Again the comparable situation in the telephone industry is that anyone can establish a business phone number and then advertise directly to potential consumers for business without further involvement of the network providers. Where cable companies choose to provide Internet access on the cable, this fits this model as well.

OCR for page 304
Page 309 The argument here is not that cable operators should be precluded from offering packaged services. They should certainly be allowed to offer them as well as local content, provided that they also provide consumer-controlled access to competitive offerings. Openness should exist in both directions. That is, providers should be able to attach to access networks to offer their services to consumers (similar to requirements for allowing video service providers to access "video dialtones"). At the same time, consumers should be able to attach via their access networks to providers who have not chosen to directly connect to that access network. Openness of Content and the Development of New Services We do not need to wait for the implementation of all digital networks and the general NII to see examples of why we are concerned with the issue of open bandwidth access. We can find today in analog distribution systems an interesting example of how fragile the relationship between openness of network capabilities, content provider innovation, and application creativity can be. This example appears in the use and distribution of material in the vertical blanking interval (VBI) of broadcast video signals. Note that the issues in this example are more complex than open access to bandwidth since they deal with bandwidth ownership and contracts between MSOs and content providers. However, the example does serve to illustrate how subtle issues in the nature of the distribution networks can affect the creation and deployment of new applications. Standard U.S. television signals must provide a period of time in each frame to allow the electron beam that paints the screen to move from the end of the screen back to the top in preparation for painting the next set of lines. This period is the VBI and can be seen as a black bar if the vertical hold of a television is misaligned. Since this part of the signal is not used for carrying the picture, it can be used to carry other sorts of data without disturbing the video broadcast. The most common example of this is closed captioning, which is carried in a part of the VBI. As broadcasters begin to think creatively about their content and work with application developers, other possible uses of the VBI to enhance the broadcast material can emerge. Examples include digital forms of the stock ticker seen on some news networks, digital references to sources of additional information to augment the content of a show, digital indexing information to assist in selective viewing of the material, and undoubtedly many more. However, any such program-enhancing use of the VBI must reach the end consumer to be useful. While the 1992 Communications Act required that program-related material in the VBI should be carried by local cable operators, exactly what constitutes program related is debatable. As a result, there is no guarantee that a content provider who finds a creative use of the VBI to provide, in conjunction with an application developer, an enhanced service will actually be able to deliver that service to the consumer. This issue can be viewed as an example of the difficulty of defining what openness in access means, even within the existing rather limited sorts of broadband distribution that exist today. A service may no longer be a two-party transaction between the provider and the end customer. It may now involve at least three parties and become dependent upon the cooperation of the access network provider. This substantially raises the hurdle to entering the business with a new service, particularly one that may be of interest to only a small percentage of consumers. Even though a broadcast channel is already carried by the vast majority of access network providers, it may be necessary to reengage all such network providers to even begin to offer a service. This certainly provides a barrier to innovation that need not be present. It is exactly this sort of barrier that we are concerned not be erected as digital services begin to be deployed. It must be possible for innovation to be between the provider and the consumer over an open digital highway. Conversely, it must also be possible for the network service provider to benefit from the expanded use of the network. Openness of Equipment Another key difference between the current directions in the cable access networks and the existing telephony networks is important to the rest of this discussion. This is the issue of open equipment connectivity and ownership of customer premises equipment (CPE). Today, the consumer owns the telephone, the TV, the PC,

OCR for page 304
Page 310 and the modem for the PC. However, it is generally the case that the cable access provider owns and provides the set-top box. Furthermore, the lack of standards for the physical network may make it more difficult to find an alternative supplier of CPE. This is quite different from the telephony world, where the access network provider terminates his network at a demarcation point, providing a standard form of connection to which customer-selected and customer-owned CPE can be attached. The combination of the ownership and the standards aspects of this arrangement have together spawned a vigorous industry in CPE suppliers. For telephony, the connectivity standard is the RJ-11 connector and the set of electrical and tone signaling standards supported over it. As a result, PC access to networks via modems can be achieved with the same equipment, anywhere in the country. The importance of the openness of CPE is beginning to be understood. For example, a press release from Congressmen Bliley and Markey issued on March 16, 1995, states, Restricting consumers' ability to purchase and own cable "set-top boxes"and other communicationsinterface equipment is like putting a straight-jacket on technologicaldevelopment.… There's noincentive for improvement because there's no competition.… Today'sadvanced phones only happenedbecause of a healthy, competitive retail market—and so did therevolution in computer modems and faxmachines that followed. However, there is more to energizing the broadband CPE industry than just allowing retail purchase of the set-top box. Consumers must see an advantage in purchasing the equipment, and vendors must see a large market to be served in order to invest in developing products that will attract consumers. Neither of these will happen very quickly without the development of standards for the point of attachment of the equipment. Consider how likely consumers would be to purchase feature-laden phones if they could not expect those phones to work after they moved homes. Likewise, consider how likely a company would be to develop an innovative new phone product knowing that it could be sold only to that set of consumers who were served by some specific phone companies, and that the company would also have to deal with unhappy consumers who discovered that the product ceased to work after they changed providers. Our concern is that consumers be able to buy PC equipment and applications that will effectively attach anywhere in the United States to the NII, and that they be able to freely move such equipment and applications with them. Industry-driven discussions are under way in various industry groups (e.g., the ATM Forum) toward the establishment of such de facto standards or implementation agreements. Affordability A considerable concern exists about the affordability of an open consumer broadband data service. Clearly, the bandwidth must be priced at a level that will allow reasonable access to a broad spectrum of users. True open competition should cause this to occur, as can be seen by looking at how other uses of the bandwidth in the access network might be priced. For example, consider a higher-level service that has been proposed, namely video on demand (VOD). This service promises to deliver to the consumer an online replacement for videotapes (and eventually probably much more interactive sorts of experiences). Assuming that one uses MPEG2 to compress a 2-hour movie, then this service must deliver on the order or 4 to 6 Mbps for 2 hours to the consumer at a price competitive with what she can rent the movie for today (on the order of $3.00). VOD is an individual service delivered to a single consumer at a time. Thus it is reasonable to expect that competition will drive the cost of similar downstream bandwidth of an unrestricted sort to be similar. (Actually, the resources to deliver the unrestricted bandwidth are less since no server is involved.) Clearly, this accounts only for the access network provider part of the unrestricted service, but it at least provides a starting point. It is harder to estimate what a consumer should be paying for upstream bandwidth, but a similar sort of analysis for a bundled service that makes more upstream demands (perhaps shopping or interactive gaming) should provide an estimate there as well. We are not trying to suggest any particular price structure for consumer-controlled bandwidth, nor are we suggesting price regulation. Rather, we are trying to suggest that competitive forces should cause it to be priced in a somewhat similar manner to similar levels of bandwidth utilized for services bundled by the local access network provider.

OCR for page 304
Page 311 An Architecture for a Long-Term Residential NII The actual deployment of broadband services to the residential NII will be evolutionary from today's physical networks. Economic and practical considerations preclude any revolutionary replacement of access networks with some uniform and full-service digital infrastructure. As a result of this evolutionary reality many of the requirements we have for a long-term infrastructure cannot be fully met in the short term. Indeed, some of the standards that we believe will be critical to providing the full potential of the NII are still in embryonic stages. Nevertheless, it is important to develop a vision of where we would ideally like to wind up if we are to generate public policies that encourage development to proceed toward an end-point architecture that can realize all the potentials of the NII. We believe that the architecture should be broken into three parts: core networks, access networks, and home networks (Figure 1). Core networks are those networks that provide generalized connectivity both locally and across wide geographical areas. They correspond in today's telephony infrastructure to the long-distance carriers plus that part of the local exchange carriers' infrastructure that carries traffic between the central offices that serve customers. Access networks are those networks that connect the core networks to residences. For example, the HFC networks that are being considered for deployment to connect regional hub systems to homes are access networks. Finally, home networks are those that operate within the residence. image Figure 1 Elements of an architecture for realizing the full potential of a national information infrastructure. ATM, asynchronous transfer mode; HFC, hybrid fiber coaxial; FTTC, fiber to the curb; FTTH, fiber to the home; and ADSL, asynchronous digital subscriber line. Note that any of these networks can be nonexistent in a specific deployment. In an existing cable system using HFC to deliver only broadcast-type services to dedicated set-top boxes in a home, neither a core network nor a home network may exist. Likewise, one could view the current telephony infrastructure as often not having an access or home network, since for practical purposes it generally appears that the CPE connects directly to the core network. We expect that core networks will typically be those with symmetric bandwidth. Considering today's trends, these are likely to be ATM-based networks. We also believe that it is in the interest of the carriers that provide core networks to encourage use of bandwidth by consumers. There is today competition among core network providers in the telephony long-distance business, and we would encourage such competition to continue into the broadband world. Given this competition, the easy access by the consumer to multiple core network providers, and the desire of those providers to sell their bandwidth, we believe that the requirement of sufficient, affordable, consumer-directed bandwidth will be met in these networks. Furthermore, there is already much momentum in the standards associated with communications over these networks (ranging from TCP/IP to ATM) so that effective consumer utilization of this bandwidth seems possible.

OCR for page 304
Page 312 Home networks are not yet at all well defined. However, as one sees more digital information appliances deployed within the home, the desire to interconnect them will increase, and this will give rise to various sorts of home networks. Displaying images from security cameras on a TV or PC, driving an interactive program on a TV from a PC, or operating a multiplayer game across multiple TVs or PCs in one or multiple homes are all examples of applications that will involve a home network. Home networks are likely to be fairly symmetric, if for no other reason than that there will likely not be a single special "source" to drive any asymmetry. Access networks may well remain asymmetric regarding bandwidth due to the economics of some of the popular access network topologies. For example, HFC will likely continue to supply much more downstream bandwidth than upstream bandwidth for the foreseeable future. We are most concerned about the evolution of the access networks since they are the critical bridge between the consumer and the core networks. While the consumer can choose what to deploy at home, and there is already a path to active competition in the core networks, the access networks may have much less natural competition. Once one understands these three related networks, it becomes clear that the critical issue for them is to define stable interfaces that can allow them to evolve independently while still delivering to the consumer a high level of service. The connection standards between the access networks and the core networks are already getting some amount of attention via the discussion of open video service providers. That is, this interface is the one that a service provider will be concerned about, whether the service is a local video store or a general long-distance carrier service. The connection standards between the access network and the home network have received less attention. This is the interface discussed above in the section titled "Openness of Equipment." We believe that it will be critical to elevate the discussion of this interface and build policy that actively separates the home network from the access network via appropriate standards for the reasons discussed above. In addition to the standards, this is also the interface across which we believe we must guarantee reasonable consumer-directed bandwidth. If we define effective standards at each of these points, then it should be possible for an information industry to develop that supplies consumers with equipment and applications that allow wide exploitation of the NII. In our vision, consumer hardware has a common socket for attaching to the NII. It uses common protocols for interacting with NII services. These protocols operate in the packetized Internet world and in the home to allow easy access to the educational and information resources of the Internet. The typical home has access to sufficient bandwidth to make access to these network resources a pleasant experience. The choices of what network services are available to the residential consumer are essentially unbounded. The local access network provider may choose to package and sell some services, thus making them easier to use, but there are no roadblocks to open consumer access to the NII at large. The key to all of this is the existence of the interface agreements that allow development on either side of the interface to progress independently and that do not overly constrain the sorts of implementations permissible between the interfaces. Without such industry-driven standards or implementation agreements we will be in danger of one of two extremes. On the one hand, their lack will cause the coupling of what should be distinct parts of the NII, thus slowing development of what should be independent parts. On the other hand, their overspecification will stifle creativity by admitting only a single family of products that can operate within the NII. With open bandwidth and protocols, intelligence can exist at the periphery of the network as well as inside it. An innovator with a good idea can create an information service as an end point on the network and sell to the consumer an access tool that resides at the consumer's system. In doing so there will not be any impediments due to a need to negotiate business deals with various network providers. This openness will lead to an opportunity for much greater innovation than that possible with an architecture that gives preference to the provision of service intelligence only inside the networks. Recommendations—Requirements for Widespread PC Access 1. Provide reasonable levels of bandwidth to and from the home at consumer costs. Data bandwidth into the home should probably at least mirror what is available today for the corporate user accessing the Internet. This argues for a minimum of T1 (1.5 Mbps)-level rates to each home, and more likely, for peak levels of at least Ethernet (10-Mbps) speeds. Obviously more is better, but the key issue is that the consumer must be able to

OCR for page 304
Page 313   access rich multimedia content at speeds that make the operation a pleasant one. Initially, it is acceptable that the bandwidth from the home be less than that into it. However, extremely low outbound bandwidth will limit the efficiency of any interaction. Furthermore, there are certainly interesting applications that will be enabled only by relatively high outbound bandwidth as well (for example, work at home or video security). Ultimately, the access network should not overly constrain who can be the providers of services and who can be only consumers. With the right bandwidth levels, the NII can truly energize a grassroots, distributed economy. The first step in this direction can come from the rapid, ubiquitous deployment of ISDN to the residential environment. While it does not meet the full requirements, it can quickly get us moving toward the long term and can allow interesting applications to begin to be deployed as it is embraced by existing service providers. 2. Permit consumer control over the use of this bandwidth to conduct simultaneous digital interactions with multiple services via open packet protocols. It must be possible for intelligence in the user's system to efficiently access multiple services in the NII to generate value at the end system by the combination of services. The openness of the protocols and of the connectivity they provide is absolutely key to enabling the development of new services and to competition among service providers. Consumers should be able to access any service in the NII, just as today they can access any telephone from their residential telephones. The choice of local access provider should not preclude this accessibility. 3. It must be reasonable for multiple PCs and other information appliances within a residence to be simultaneously accessing the Internet or other broadband services. The fact that a child is conducting an educational interaction with some service on a PC should not preclude a parent from conducting a financial transaction at the same time. It is desirable, in fact, that all home information appliances be viewed as equals to that there is at least logical symmetry between originating information inside or outside the home for consumption inside or outside the home. 4. Encourage sufficient standards to facilitate a commodity consumer business on at least the national level for PC connectivity. Competition drives costs down and service levels up. However, meaningful competition cannot take place unless a large marketplace becomes available to the potential suppliers. Consider the differences in corporate networking costs today versus 15 years ago. Standards such as Ethernet have energized an entire industry, with the result that the cost of connectivity for an end system has approached $100. In the consumer market segment, the standards in place for telephone connectivity have done the same thing for products that attach to an RJ-11 socket. This benefit will not accrue to NII connectivity if every regional or metropolitan market segment requires different consumer equipment for connectivity. For example, the consumer is not served well by the multiplicity of ISDN "standards" across the country. A resident of the United States should be able to move NII equipment anywhere in the country and use it effectively, just as can be done today with other appliances. It is important to note, however, that the most effective standards in the business world are those that have been initially industry-driven, de facto ones that only later were codified into a de jure form. (For example, the nearly ubiquitous Ethernet and TCP/IP came about in this manner, while the OSI protocols represented an attempt to drive standards from the de jure side first.) The role of the government should not be to impose standards but rather to create policy that facilitates the rapid development of appropriate industrial ones. 5. Permit consumer ownership of the periphery of the network in terms of what kind of equipment is connected. (Of course, this should not preclude an access network provider from leasing equipment to the consumer as an alternative.) Similar to the previous point, we can see an entire industry that has innovated based on the ability of companies to develop products to address consumer needs, without needing the permission, or worse, the active participation of intermediaries to become successful. Look at the innovation that has occurred since the Carterfone decision opened up access and ownership of end equipment for the telephone network. The evolution of capabilities offered by the personal computer is another example of market-driven innovation. Locking access networks into an environment where only provider-owned or provider-sanctioned interfaces are permitted to be attached to them recreates the old phone network and its constrained equipment competitive environment.

OCR for page 304
Page 314 Conclusions It is not clear to us at this point that the natural and policy forces at work in the emerging NII will achieve all the important results outlined above. There is, unfortunately, something of a chicken-and-egg problem concerning open residential use of the NII. The emergence of innovative NII applications utilizing residential broadband capabilities depends on the existence of those capabilities, while justification for the deployment of those capabilities requires hypothesizing the existence of those applications. As a result, we have a situation where to some extent policy choices may need to precede market development. We do not suggest a move toward a highly regulated environment, since ultimately that is completely counterproductive to the development of a new industry. However, we do suggest that policymakers need to find ways to encourage the development of an eventual architecture that supports the full potential of the NII. To some extent this notion is already present in the types of trade-offs being made to balance the cable industry's and telephony industry's developing competition in each other's businesses. We are simply encouraging policymakers to take a broader view of this set of problems that looks beyond this level of competition to include the rest of the infrastructure that the consumer will need to become a full citizen on the NII. Policymakers should see that more is involved than allowing content providers open access to consumers or considering what the competitive trade-offs should be between allowing cable systems to offer dialtone and telephone companies to offer video. They also need to look at the provision of general data services from the perspective of the consumer. We believe that more attention should be given to the provision of open, standardized, commodity-priced network access to the NII at large. We believe that only with this capability can we tap the PC's full potential to become the consumer's interactive access point to NII services and bring to the residential consumer the dynamics that have so dramatically benefited the corporate PC user.