Infrastructure for DSL Solutions: agcs.com
============
Infrastructure for DSL Solutions
by
Mark Emery: AG Communication Systems
One of the most talked-about areas in the telecommunications industry today is digital subscriber line (DSL) technology. DSL may be offered as asymmetrical DSL (ADSL), high-bit-rate DSL (HDSL), symmetric DSL (SDSL) or very high speed DSL (VDSL). This group of technologies is frequently referred to as "xDSL." With the focal point ADSL service, a single line can move information downstream at extremely high speeds and upstream at a slightly slower rate. It can also carry both voice and data signals in both directions simultaneously. Many interesting parts of the xDSL marketplace to date have focused on speed. However, much as a common golfers' phrase is "drive for show and putt for dough," the xDSL marketplace's analogous phrase could be "speed for show and services for dough." While speed may be the factor that draws customers initially, there must be exceptional service in order to keep them. This paper will address both aspects of DSL service as well as architectural and infrastructure concerns.
Market Factors
While DSL is certainly an exciting technology, it is competing with several other new entrants in the field and is by no means guaranteed success. There are many factors to consider when trying to choose a technology and, therefore, when trying to forecast the market growth of any product. One of these is cost. DSL has anticipated end-user service charges of $30 - $100 per month, depending upon speed, services and subscriber session count. It is forecast that for residential service in North America, fractional penetration will require $65 - $100 per month for basic service. Reaching single digit penetration will require $45 - $65 per month for basic service. Reaching double digits will require $30 - $45 per month for basic service. The residential market is certainly not the only goal for DSL providers, though. There is also a drive to reach the business market, which will be looking for things to replace existing switched 56 services, integrated services digital networks (ISDN) and T1. The business market requires high value-based service pricing where DSL will attempt to deliver services previously unavailable and/or undercut existing service pricing.
Figure 1 shows a projected penetration chart from primary market research conducted by AG Communication Systems in early 1997. Residential penetration begins to rise at about $65 per month and curves steeply upward when prices drop to about $40 per month. This fits in with some of the paradigms that have been seen in the past with the digital broadcast satellite (DBS) and cable industries. It should be noted that cable now passes more than 95% of homes, some 60% of which subscribe to the offered service.
DSL has the right offer mix, and providers are discovering what their pricing plans should be. Still, all of this must fit the economic models of the characteristic business or household in the United States.
Figure 1: Price versus Residential Penetration for Basic DSL Service
DSL Service Parameters
Other considerations, along with those basic economic parameters, are the types of services to offer. Right now, there is an enormous emphasis on data. The explosion in T1 services and the demand for trunking between today's existing switches and service access points are closely related to the delivery of data to businesses. They are also related to delivery of data to residential customers, educational institutions and so on by hauling that data to various access points. In addition to data services, there is a growing demand for video services. Those services cover low-speed and medium-speed video conferencing, NVOD and VOD.
Multiple-service sessions over a single DSL line are also very important. The reason can be seen by looking at today's educational institutions: children are being taught to use computers in kindergarten. When they get to grade school, they are writing reports by pulling data off the Internet. This paradigm will continue to expand. In fact, within a typical household, there might be two or three "consumers," one of whom is interested in talking and getting data for work applications and one of whom is, at the same time, interested in doing something in an educational institution or for an educational reason. The third might be interested in some sort of entertainment. All of these services use a variety of speeds. Hence, DSL services must offer and support this variety as well as allowing multiple simultaneous sessions over a single line. Other services that may prove desirable include advertising and security monitoring -- partially for economic reasons and partially for a richer set of vertical applications.
All this aside, first and foremost, xDSL is about delivering some form of value-based content -- either symmetrically or asymmetrically. However, providing a delivery service is not enough. Provisioning and continued service are also essential. The DSL provider must offer straightforward installation, provisioning and maintenance. It must be possible to deliver service through different physical plants such as the residential environment based on a digital loop carrier or the residential environment based on a central office and to provide service turn-up and prove-in. As these services roll out, there must be robust systems in place to handle billing, ongoing operations, service change and customer support. Essentially, DSL providers must be able to supply the same things that a customer is used to finding with the larger, more established service carriers and supply them at least as well.
It is possible to provide whole effective lines with solutions encompassing not only the technology but also the wrapper that goes around it to deliver service. Together, this is what will sell end customers. In fact, most people do not have strong feelings about what technology they use as long as it works well and they can get effective service, intelligible billing and high perceived value. A whole effective line, which includes a pair of modems, multiplexing, shelving, broadband IFC and support, may have a cost as high as two to three times that of a modem line. This makes the services provided particularly important: the value of those services in the mind of the consumer must outweigh their cost to the provider.
Infrastructure and Architecture
There are, then, many parts to DSL service. Some support service delivery, and some support control, administration and special services. First, to support service delivery, there is DSL equipment, which includes modems and multiplexors. Then, there must be integrated interfaces to data providers, which could be using T-1, T-3, OC-3 or other speeds, and multiple protocols (IP, ATM, Frame Relay, etc.). There is also a broad range of data infrastructure: edge switches, routers, service access multiplexors and so on. Finally, the service delivery support infrastructure must have integrated interfaces for other services like voice and video which will only grow in demand. Video telephony, video conferencing and composite streaming video are definitely applications for a DSL environment. So are things like TR303 or TR08 interfaces that allow direct tie-ins to switching fabric to draw voice traffic from the DSL link.
Managing and controlling all of this requires element and network management. This could extend to local or wide area networks and will include traffic engineering, network engineering, diagnostics and so on. The control and support infrastructure must allow for operator control and management of these functions and should also provide network engineering and diagnostics capabilities. Other important areas include service subscription management, billing systems and special services such as advertising and security.
An architecture that supports all of these functions is shown in Figure 2. This total DSL architecture has parts that provide billing, service control, the potential for the end user to control service and management capabilities that allow device monitoring and control. The architecture has an asynchronous transfer mode (ATM) backbone network that contains a switching fabric and access points. Finally, there are multiple DSL technologies, IP-based and ATM-based, that provide the copper loop solution comprising customer premises modem and modem technology for the central office and digital loop carrier.
Figure 2: GateWay Total DSL Architecture
Broadband Infrastructure Issues
One problem that must be dealt with by DSL providers is exemplified by the recent experiences of Internet service providers (ISPs). They have had a great deal of difficulty delivering analog speed 14.4 kbps and 28.8 kbps data services. Now, suddenly, there will be a growing pool of users trying to use data services at multimegabyte or at least hundreds of kilobytes per second. This is some 50 to 100 times faster than what has been common until now. What this means to providers is that everything in the network must scale -- end-user platforms, modems, copper speeds, multiplexors, switching, routing and data servers at the providers. The bottom line is TOTAL throughput must be taken into account -- connection levels and their management must all tie together.
In terms of switching, it is not only necessary to scale the backbone but also to create effective controls. Latency and synchronization must also be considered. The more constant bit rate services like telephony and video that are put through a DSL network, the more issues there are regarding end-to-end communications. For two-way services, audio/video synchronization has been a consistent problem and is one that must be solved. Another issue involves routing. If there are many routers, someone has to administer them. Administering any kind of a network is a significant hidden cost, perhaps even as much as 80% of total, post-setup cost over time. After routing, addressing and conversion must be considered for Internet protocol (IP), ATM, frame relay (FR) and mixed media protocols like MPEG and TDM.
Spectral compatibility may be one of the most important issues facing DSL. DSL devices deliver on the order of 10 to 100 times more power into the loop than existing network-delivered services. When multiple DSL lines are put into a 24-pair bundle, there is crosstalk between the DSL lines. The more power that is put in (which is directly related to loop length and speed), the more chance there is of crosstalk. When these lines are put next to an ISDN service or a switched 56-kbps service, which is running at one or two orders of magnitude less power in the loop, it is very easy to have service crosstalk. One of the parts of infrastructure that has to be addressed very closely in terms of rollout, then, is the amount of power and spectral frequency mapping.
Finally, from a management perspective, there are several things to be considered: network management, element management, preventive and diagnostic maintenance, administration and provisioning and billing. The issues here center on new management protocols (SNMP, CMIP, TMN), legacy management systems which are mostly proprietary, access mix by both users and administrators and the level of element- and network-level management.
Not too surprisingly, because providers are seeking to deliver a mix of services, they are interested in a mix of technologies. Two of those are Internet protocol (IP) and ATM. The IP infrastructure is fairly ubiquitous and does its job extremely well. ATM, which is far less prevalent, was designed from the bottom up with a mix of services in mind. There are advantages to each technology.
IP is inexpensive and by far the most prevalent. 10/100baseT is an industry standard, and Ethernet is basically self-learning. IP is highly effective for data traffic as shown by the LANs that are proven and working today. IP PC software and operating system drivers already interface to IP-based LANs. Finally, Ethernet and IP offer inexpensive LAN PC cards.
On the other hand, ATM makes the mixing of services like video, telephony and data far easier than it is with IP. Quality of service or QoS constructs are built into the protocol. ATM also has traffic speeds that conform to standard telephony transport rates like T3 and OC-3. ATM technology is real. Streaming video transport and compression, an array of LAN services and backbone switching are proven applications. Finally, most new software and drivers are being engineered to work with ATM.
There are applications and customers that support each protocol choice as well as some that have a need for both technologies. For these reasons, DSL must have an architecture and an infrastructure flexible enough to handle them.
Knowing that IP is probably here to stay but that many customers are choosing to start with or switch to ATM, it is important to know where ATM conversion happens. This is an issue that many people are trying to address. As shown in Figure 3, ATM conversion can involve many different options. There can be ATM all the way to the premises, which is subsequently converted to IP, voice, video, etc. Or ATM conversion can take place at the central office, and IP or CBR can be used to go out to the premises. Another alternative is to use ATM or Frame Relay on the back end of the broadband switch. All these options have implications relative to how a network should be structured and the services it will support.
Figure 3: ATM Conversion Points in a DSL Network
Applications
Before setting out to create and market a DSL system, it is important to consider how users interface with devices. Obviously, computers with browsers (standard versions of Netscape Navigator or Microsoft Internet Explorer, etc.) are basically ubiquitous. One of the primary uses that people have for these platforms is tying into data networks. For these applications, the concept of service subscription and connection management is very important. There must be access to cyber services where the user may have encrypted financial or electronic commerce. Other services may include video, v-mail, e-mail, a list of ISPs or services to select and so on. All of this should be at the fingertips of the end-user through a browser-type interface. The goal is to create a flexible, easy, point-and-click interface that is analogous to "dialing the telephone." There should also be a way to defray the network cost using advertising and messaging. This leads us to the conclusion that users should, upon attaching to the service provider's application interface, be presented with standard browser screens like that shown in Figure 4 that allow them to select a function from a suite of offered and tailored services.
Figure 4: Service Connection Management
If, for example, the user chose "SERVICE," the corresponding button would be clicked, and the screen would change to one like that in Figure 5.
Figure 5: End User Service Account Browsing
This screen would offer different services, like account monitoring, service addition, service changes or customer service. If, for instance, a user had an application that needed a different speed, "Change a Service" could be selected and then "Speeds." The next screen might look like Figure 6 where price and speed are directly selectable.
Figure 6: End User Speed Selection
Here, users can select their own maximum loop speeds. Note that throughout this process, no operators have been involved. This not only gives the user a feeling of control and convenience, but also saves the provider some of the cost involved in maintaining a customer service center. The billing record is still created and the transaction kept, but cost has been removed. This is the sort of capability that should be built into advanced services like DSL networks.
Other Requirements
Of course, there is more to a successful service than architecture and applications. Installation and customer support are two of the biggest concerns. Installation alone has three separate aspects that must be taken into account: telephony, data and premises. In addition, the provider must prepare for several different regulatory issues such as ILEC and CLEC ownership and distribution of revenues. The telephony aspects of installation include the following: central office and remote switching equipment; -48Vdc power, grounding, copper cabling, testing and service-mix interference; digital loop carrier equipment; environmental packaging; construction; and AC power issues.
When considering the data aspects, the installers must have Ethernet and ATM knowledge as well as basic LAN, router and switching knowledge. They have to understand path or tunnel provisioning versus open campus LANs. In many of the current trials, two people are being sent out to do an installation at a site. This is something that must change as the cost of having two people and a truck on an installation for two or three hours is $100 to $200. Overall, the process must be simpler, e.g., a plug-and-play kind of environment that allows people to do it themselves and still have the equipment work.
Finally, premises aspects of installation must be considered. Premises wiring is extremely important. One of the main ADSL concepts is to provide separate voice and data on a single pair if desired. If, on the other hand, there is no desire to offer them both, then it must be possible to have them on adjacent wires. Any wiring changes must also fit in with existing equipment. All of this requires a certain amount of PC and telephony knowledge on the part of the installer. One last issue connected to premises concerns is that of professionalism. The person who is installing the system will be one of the first contacts that the customer has to the supplier. That first contact must be a positive one.
Many customers are also concerned about security. Instances of hackers, unscrupulous past or present employees and others who may be out to steal from or otherwise harm a company are all too common. Unfortunately, this sort of issue is real in today's environment. When there are high speeds, large amounts of data and people browsing wherever and whenever they can, the user and providers must feel like access is discrete and under control. This means that the supplier must find a way to solve the matrix of dynamic user connections, packet or cell routing and path or tunnel security from both ends -- all without putting a huge infrastructure cost or complexity in place. There are several ways to address these issues, like directed IP protocols, tunneling and virtual circuits. Using any of them, however, requires thought before the equipment is put in, not afterwards.
As every successful company knows, service considerations are not over when the service is sold and installation is complete. It is essential to have good support systems in place that will fulfill the needs and expectations of the customer. People pay for the service itself, the delivery of the service and its quality and reliability. These things have little or nothing to do with the actual technology used in equipment. They have everything to do with the people providing the service, the timeliness of the service and the level of control the end users feel that they have. It is essential for the supplier to provide accuracy and timeliness of service delivery and billing. No customers want the aggravation of incorrect or confusing billing that may continue to affect them for months or years. When these kinds of services are put under the end user's control and done dynamically, they must be easy to use and accurately tracked and billed.
The area of customer support also includes less technical areas like first point of service assignment and contact. There should always be a pleasant, helpful voice at the other end of the line to give customers the assurance that they need. Service technicians, whether in the central office or in the field, need to remember that they are representing the providers. Also, post-sale service support is certainly as important as pre-sale assurance. If there is a problem with the equipment, there must be a friendly representative who can be accessed to fix the problem as soon as possible or send someone out, if need be, and fix the problem on premises. Finally, users must feel that they are controlling their own service and service options.
Conclusions
There is no doubt that digital subscriber line services can be an extremely useful and well-received product suite. There are, however, some major infrastructure challenges associated with DSL. First, it must provide cost-effective, reliable, easy-to-use service. Second, the service must be scalable in terms of its architecture and management. It has to integrate with today's telecommunications networks and data networks, providing an end-to-end solution that includes network and element management as well as service subscription management. Finally, DSL infrastructure must meet the needs and expectations of the end user. Beyond basic pricing, this involves many requirements in the areas of pre- and post-sale abilities and service. These capabilites need to provide methods for users to control their own service, which means selecting their providers as well as changing their service parameters. If all of these areas are accounted for, all parties involved, from supplier to end user, should be winners in the DSL environment of tomorrow.
============= |
