Telecom companies, startups, and others have been promising high-speed broadband Internet connections to our PCs for several years, yet only a fraction of homes and small businesses actually have that luxury today. But progress is under way. Local carriers and competitive start-ups have begun to roll out digital subscriber line (DSL) service, many local cable-TV companies offer high-speed access, and a handful of wireless companies now provide various wireless access schemes.
A tangled mix of issues caused the slowdown in rolling out widespread broadband. Clearly, the main reason is economic rather than technological. While many businesses have suffered failures and major setbacks, especially in DSL and wireless, the whole broadband field is moving ahead nicely. A major need still exists for broadband access, and demand continues to outstrip supply. Required is the right economic model that lets suppliers meet their profit hurdles while simultaneously satisfying the user with high speed at a justifiable monthly fee.
Although we think of broadband access primarily to the Internet, a fast connection makes other applications possible. It provides the potential for services like video on demand (VOD) and streaming audio. Digital or IP voice telephony is expected to eventually replace our old familiar analog telephone service. Gaming and videoconferencing are other potential services not yet offered. Also, once a home has a broadband connection, a home gateway can be added to supply service to multiple PCs as well as to other multimedia devices.
With approximately 6.45 million cable-modem subscribers as of mid-2001, cable access still dominates. DSL comes in second place with about 2.91 million subscribers in the U.S. Various fixed and mobile wireless services have several hundred thousand subscribers. It's expected that by the end of 2001, the number of cable subscribers will rise to about 8.4 million, and DSL will reach around 4.1 million. This is only about 8% of U.S. households.
Most agree that it will take a critical mass of at least 20 million broadband connections to make some of the proposed services profitable. For now, the industry is concentrating on fast Internet access, but VOD and VoIP are clearly in the future.
As most major cities have major networks in place, cable companies provide an ideal medium for high-speed data. Each 6-MHz channel can download data at up to 30 Mbits/s using 64QAM modulation, and at 27 Mbits/s using forward error correction (FEC). Such rates aren't achieved at the subscriber's cable modem, but up to 2- or 3-Mbit/s rates are possible.
The success of cable for fast Internet access is primarily due to the favorable characteristics of cable systems and a well-organized cable industry. Most cable systems are a hybrid fiber coax (HFC) arrangement with a main fiber trunk cable distributing the content to many coax cable nodes. These feed the drops to each home. Cable bandwidth is organized in adjacent 6-MHz channels designed to carry TV signals. Coax bandwidth is as much as 860 MHz, so over 120 channels are available. Only a few are allocated to data applications.
Another positive factor is a set of standards developed primarily for carrying data over cable. Developed by CableLabs with industry cooperation, the Data Over Cable Service Interface Specification (DOCSIS) defines the physical layer (PHY) and the media access (MAC) layers necessary to carry IP data. This standard gives semiconductor and equipment manufacturers guidelines by which to make products that meet the standard and interoperate among each other.
Aside from slowdowns that users experience when many subscribers simultaneously access the bus-like network, another downside to cable access is a severely limited upstream bandwidth. Cable systems are very asymmetrical. They can download signals at up to 27 Mbits/s. While the maximum theoretical upload speed is 10 Mbits/s, it's rarely achieved because most cable companies limit uplink speed to between 256 and 384 kbits/s.
Although the upstream bandwidth is limited to the 5- to 40-MHz segment of the cable's spectrum, the most useful range is 20 to 42 MHz. Moreover, upstream channels are only 2 MHz wide where QPSK modulation is used. Because the upstream spectrum is noisier, QPSK provides a far more robust transmission method than the 64QAM used in the downstream mode. Most Internet access is heavy on the downloads and light on the upstream, so this limitation isn't significant. But with more users on the system and many uploading large files, the upstream capabilities are severely overtaxed.
Cable companies are handling this with equipment upgrades. As users crowd the lines, some companies split nodes to reduce the total number of users per cable. Some systems also are upgrading from 64QAM to 256QAM, which increases the data rate in a 6-MHz channel to a raw 43 Mbits/s, or 39 Mbits/s with FEC. To deal with the upstream problem, cable companies are going to 16QAM instead of QPSK.
Further cable improvements will take place when equipment implementing the new CableLabs standard DOCSIS 1.1 will become available in 2002. This updated standard will provide features that permit operators to offer different levels of speed. In other words, operators could provide guaranteed bandwidth and quality of service based on price. The new standard will also allow for packet fragmentation and better security. It will implement VoIP, giving cable operators the option to offer telephone service on their systems. Many cable operators believe that by bundling voice, video, and data services, consumers will simplify their bill paying and save money in the long term.
In the coming years, cable will no doubt continue to hold its number one position. But many predict that DSL installations will increase at a faster rate than new cable subscribers so that eventually it may pass cable as the number one broadband supplier. But cable-TV providers aren't sitting by idly either.
For example, Broadcom, the leader in cable-TV and modem ICs, recently announced several chips that will take cable to a whole new level, says John Gleiter, director of cable modems at Broadcom. The BCM3345 is a single-chip cable modem with a tuner, modem, and MIPS controller that meets full-U.S. and European DOCSIS versions 1.0 and 1.1. It features Broadcom's Propane packet acceleration software, which greatly increases the number of upstream users and upload speeds. This chip also includes a 10/100 Ethernet PHY/MAC with MII interface and a USB 1.1 interface.
Broadcom's new BCM3360 is a single-chip broadband gateway with a cable modem, a firewall router, and a home networking system. It contains an integrated Network Address Translation firewall, IP address Dynamic Host Configuration Protocol, and server and router software.
Also, the BCM3360 has an integrated Home PNA 2.0 (Home Phoneline Networking Alliance) interface, enabling the existing telephone lines to act as a 32-Mbit/s multimedia network connection throughout the home. It concurrently delivers voice, video, and data. Additionally, it supports wireless networking interfaces for the IEEE-802.11b and Bluetooth protocols.
For the cable's head end, Broadcom has matching chips. The BCM3138 is a universal advanced TDMA dual burst receiver, while the BCM3034 is a universal advanced downstream modulator. Both are designed for Cable Modem Termination System equipment at the head end. The BCM3138 receiver works with the new faster upstream modem chips described above to boost upstream speeds using TDMA and 64QAM techniques.
An example of one part keeping cable competitive is Conexant's CN9414 single-chip cable modem. Nine manufacturers have now received CableLabs DOCSIS certification using this chip. It features Direct IF sampling at 44 and 36 MHz, an integrated ADC and DAC, FEC, an embedded ARM9 microcontroller, an integrated 10/100 Ethernet MAC, and a 12-Mbit/s USB transceiver. The chip is fully programmable so that cable operators can provide field upgrades to the customer.
Recently, STMicroelectronics and Microtune teamed up to provide a DOCSIS 1.1 cable modem reference design. Based on STMicroelectronics' STV0396 single-chip cable modem and Microtune's MicroTuner, it's optimized for high performance at a very competitive price.
Under a recently announced partnership between Motorola and Broadband Innovations (BI), Motorola will use BI's new 3210A CATV upconverter module in its digital CATV head ends and HFC hubs (Fig. 1). The part accepts IF inputs and converts them to any one of 137 channels in the 54- to 876-MHz range. According to BI's vice president of marketing and sales, Larry Stark, this patented upconverter is half the size, uses 66% less power, and has lower phase noise than all prior upconverter designs.
Texas Instruments (TI) recently announced two ICs designed to boost the upstream capacity of cable systems. The chips offer a TDMA solution with enhanced modulation to boost upload capacity by 50%. Both chips incorporate 64QAM modulation, Reed-Solomon FEC, and a 5.12-Mbaud rate.
The TNETC4522 dual-channel burst receiver is designed for the head end. It incorporates TI's proprietary increased capacity ingress cancellation technology. This DSP-based IC cancels ingress and burst noise, a common scourge on upstream channels.
The TNETC4042 is the integrated MAC/PHY chip for the customer's cable modem. Together, the two chips solve cable's toughest problem of limited upstream bandwidth, allowing multiple-system operators to offer fully symmetrical upstream and downstream services. The chips are fully DOCSIS 1.0 and 1.1 compatible.
Most cable systems are for individual consumers, but that will soon change. Consider Advent Networks new patent-pending Ultraband system. This cable system provides a dedicated 40-Mbit/s symmetrical connection for subscribers on present HFC cable systems. Current contention-based, shared-access DOCSIS-compatible connections aren't suitable for businesses as speeds vary considerably depending on the traffic.
The Ultraband system doesn't use DOCSIS specifications. But it can operate alongside of a DOCSIS system. Ultraband uses the cable spectrum differently along with a different modulation scheme to achieve its ultra-fast operation. As Advent's CEO Geoff Tudor says, "Ultraband is an alternative to fiber to the home and it certainly makes cable a viable broadband alternative for businesses."
No other broadband technology has received as much press coverage and hype as DSL, and more companies are making chips and equipment for DSL than for any other broadband technology. Yet while DSL is finally available in many areas, it lags cable in deployment, mainly due to economic and technological reasons. The recent economic downturn has forced many of the competitive local exchange carriers (CLECs), or digital CLECs (DLECs), out of business, or at best into a cutback and regroup mode. Some major equipment manufacturers (Nortel, 3Com, and others) have dropped out of DSL altogether.
Nevertheless, DSL is a strong competitor in the last-mile broadband war games. Development continues to make this technology even more useful and robust. It's a most logical choice for everyone's broadband connection because of the ubiquitous twisted-pair local loop. Even though standards were developed and many new products were created, the process of transmitting high-speed digital data over the local loop has proven far more difficult than expected. A complex mix of problems has slowed DSL deployment, frustrated the carriers, and infuriated customers. Key problems include:
The reach problem: Local-loop connections to telephones vary from a few thousand feet to well over 20,000 feet. Standard ADSL can generally operate at up to 15,000 feet. At distances beyond 18,000 feet, ADSL is virtually useless. This problem is further aggravated by the presence of loading coils and bridge taps, which severely limit transmission distances. This problem has eliminated a huge number of potential subscribers.
Noise and crosstalk: In addition to the attenuation and distortion caused by variable line length, noise and crosstalk seriously limit a DSL connection. Because most twisted pairs are bundled into large cables, they experience crosstalk from other lines. The longer the line, the higher the noise level and crosstalk.
Interoperability or lack thereof: The existing ITU standards for DSL haven't guaranteed interoperability among equipment of different manufacturers. One manufacturer's CO DSL access multiplexer (DSLAM) won't always work with the DSL modems from others.
The difficulty of provisioning: Most carriers offering DSL have found it far more difficult to deploy and provide DSL than they expected. It takes time to check out a potential subscriber's line, install the equipment and software, and make it all work together. In most cases, a truck roll is necessary to get the consumer up and running. This adds both time and cost to the rollout of DSL and severely affects profitability.
The multitenant problem: Some of the most frustrated potential DSL users live in apartments or have offices in multitenant buildings. Only now is the problem of providing DSL connections to multiple tenants being solved.
Access to lines by CLECs: The key to making the CLEC business work is a provision in the Communications Act of 1996. It permits CLECs to access the consumer's local-loop lines at the facilities of the incumbent local-exchange carriers (ILECs), which are mostly the regional Bell operating companies (RBOCs). So while the law mandates access, actually acquiring it and space in the local COs for equipment hasn't been easy. Squabbles over space, equipment positioning, and particularly the location of the splitters required by the ILECs to separate the DSL equipment and voice lines are common.
The reach issue, with its attendant problems of noise and crosstalk, is being solved in a variety of ways. First, the ILECs are discovering new and better methods to test the lines and judge their applicability to DSL. Second, a number of companies now offer special line-testing hardware and software.
The most widespread provider of DSL in the U.S., SBC Communications, is solving the line-length problem by setting up broadband gateways in local neighborhoods to shorten the CO's reach. SBC runs fiber-optic cable from the CO to the broadband gateway where DSL customers are connected over much shorter twisted-pair lines. This opens the market to many thousands more homes than previously possible over standard local-loop connections.
One of the best solutions to the reach problem has been the development of a new variation of DSL called symmetrical high-speed digital subscriber line (SHDSL). This newly approved ITU standard offers symmetrical-speed DSL service over existing single twisted-pair lines. By using trellis-coded pulse-amplitude modulation (TC-PAM), it grants a distance advantage of 25% to 35% over standard discrete multitone (DMT) DSL. This new technology offers a fully reach-adaptive speed range of 192 kbits/s to 2.3 Mbits/s. SHDSL is expected to be especially attractive to businesses because of its superior compatibility with T1/E1, ATM, and other legacy technologies.
Semiconductor companies are ad-dressing the inoperability problem as well. In cooperation with the DSL Forum and the InterOperability Lab testing facilities at the University of New Hampshire, slowly but surely, DSL equipment manufacturers are discovering and solving the problems that prevent interoperability.
As for provisioning, DSL installation becomes faster and easier with equipment improvements and resolution of interoperability issues. This reduces the number of help-desk calls and truck rolls to get a subscriber up and running. Carriers are beginning to provide self-install kits with more-detailed instructions that help subscribers install the modem and software and get it working with little or no assistance.
New provisioning software has also been developed by BroadJump. SBC and others are using this autoprovisioning software to speed up and simplify the provisioning of DSL lines.
One of the biggest players in DSL is Analog Devices (ADI). It continues to roll out new DSL chips, like the Anaconda xDSL chip set. ADI's AD6497 is a 16-port CO chip with an integrated line driver and receiver that reduces the need for external components by 35% and lowers power dissipation to between 1 and 1.5 W/port. Smaller size and lower power are key features for future DSLAMs as the number of DSL subscribers grow and CO space and power become scarce.
On the customer-premise equipment (CPE) side, ADI also launched the Eagle two-chip set, one of the smallest and lowest-power sets available (Fig. 2). One chip integrates industry-standard bus interfaces with an ADSL DMT data pump. The second IC is a mixed-signal device that fully integrates the analog front-end and line driver and digital receive filters. The chip set is available with any choice of three standard interfaces: ATM UTOPIA, PCI, or USB.
New ADI ADSL line drivers also are available to provide a variety of power-supply voltage options, reduce power consumption, and improve ADSL's reach. On the CPE side there's the AD8018 5-V device, which eliminates the need for a separate dc-dc converter, and the 8019 ±12-V device. On the CO side are the AD8016ARP and AD8016ARE drivers. A new receiver, the AD8022, is available.
An interesting approach to next-generation ADSL hardware and software is on its way from Bandspeed Inc. of Austin, Texas, with R&D in Melbourne, Australia. Bob Burke, Bandspeed's marketing director, says the company's forthcoming ADSL chip set and reference design for CPE modems improves overall performance by increasing receiver sensitivity, improving noise and crosstalk immunity, increasing the channel's tolerance to impairment, and improving synchronization reliability and spectral compatibility. This technology should significantly extend the ADSL service boundary for carriers, immediately expanding revenue opportunity without additional infrastructure investment.
SHDSL is one of the hottest areas of DSL, with Infineon Technologies and Mindspeed working on SHDSL chips. Infineon's Socrates PEB 22622bSHDSL transceiver implements TC-PAM line coding that makes SHDSL so attractive. It also includes a 2B1Q line-coding mode, giving the chip compatibility with older HDSL equipment.
Mindspeed, Conexant's networking division, recently introduced its CX28975 ZipWirePlus transceiver. It too offers both TC-PAM and 2B1Q modes. The CX28985 is an octal SHDSL transceiver with built-in line driver and frequency synthesizer for CO applications. Interestingly, this chip's embedded controller runs Mindspeed's LoopWizard sequenced time-domain-reflectometry software. This combination lets the carrier determine line length while detecting and locating opens, shorts, bridged taps, and loading coils that would degrade or even prohibit service deployment. Mindspeed and Infineon recently demonstrated the full interoperability of their SHDSL chips.
Virata, another SHDSL player, recently announced its Aluminum series. The Aluminum 200 DSL processor is a complete transceiver with modulation, demodulation, and filtering. The Aluminum 200 AFE contains line drivers and receivers with an ADC and DAC and related filtering. This chip set also supports the HDSL2 standard and 2B1Q operational modes.
STMicroelectronics just announced the CopperWing two-chip set, aimed at CO product manufacturers. The STLC60844 is an analog front end that integrates eight analog transceivers, while the STLC60845 is a DMT transceiver incorporating eight data pumps with two ARM7 processors. On the CPE side, STMicroelectronics also introduced the Unicorn chip set. The ST70137 is a controllerless ADSL DMT transceiver with both PCI and USB interfaces.
Another strong DSL chip maker, TI, recently introduced the AC5 octal DSL infrastructure chip set, the AU5 CPE chip set, and the AR5 ADSL chip set for routers, IADs, and residential gateways.
Very high-speed DSL (VDSL) technology has also been receiving lots of attention, such as Telco Systems' EdgeLink system. It consists of the VTU10 Termination Unit, which provides a symmetrical 10 Mbits/s on a standard twisted-pair line. When used with the EdgeLink V24S VDSL Switch, a complete VDSL system can be implemented for 24 users. Telco Systems is addressing the markets of multitenant and multidwelling units (MTU/MDU) in apartments, hotels, college dorms, and small office buildings.
Also a VDSL vendor, VDSL Systems produces the iVALO modem card, router, and DSLAM. The system can be configured for 13/13-Mbit/s symmetrical or 26/3-Mbit/s asymmetrical configurations of up to 2.5 miles on a single pair. Using two twisted pairs permits setting up a 26/26-Mbit/s symmetrical system over 2.5 miles.
Finally, Teradyne recently announced its Celerity DSL test system (Fig. 3). Designed for use by ILECs and CLECs to test every potential DSL line, it can greatly speed up DSL deployment by accurately and quickly determining which lines can and can't be used.
Broadband won't be dominated by any one technology. During the next few years, the broadband technologies will sort themselves out in the marketplace. It won't be easy for the industry, but in the long run, the consumer will benefit.
|Useful Broadband Web Sites|
|Companies That Contributed To This Article|
Analog Devices Inc.
(781) 461-3333 or
(902) 468-6397 or
Infineon Technologies Corp.
(949) 483-4600 or
(781) 551-0300 or