I'll believe it when I see it—that is, a third broadband option to cable and DSL. It could be the fiber to the home (FFTH) via passive optical networks (PONs), as I reported several months ago. But a broadband option that may wind up winning third place is the newest wireless metropolitan-area networking (MAN or metro) technology known as broadband wireless access (BWA).
It's my guess that if the entrenched broadband carriers are going to get some big-time competition, the time is now. Not only do U.S. broadband Internet access customers pay more than most other subscribers in high-tech countries, the U.S. also seriously lags in the rollout of broadband to the population.
The cable and telecom carriers have successfully maintained their monopolies. While providing fast access to millions, they've stifled competition, kept prices high, and left a huge segment of the population in the dark without high-speed broadband access. After all, business is business. With the recent approval of the new IEEE standard for BWA, that's about to be corrected. The hope is that we're on the threshold of a broadband revolution.
When you stop to consider the entire wireless landscape, you see personal-area networks (PANs) such as Bluetooth, ZigBee, UHF ISM radios, and ultra-wideband (UWB) solving short-range needs. There are also wireless local-area networks (Wi-Fi WLANs) serving the enterprise, home, and hot-spot markets for medium-range access, as well as wide-area networks (WANs) provided by our cell-phone carriers with their worldwide connections.
But there are no wireless MANs. Numerous wired MANs exist, of course, either for cable-TV systems or fiber-optic networks aimed at metro service (such as linking LANs together and connecting LANs to WANs). Forthcoming PONs are the ultimate fast MAN. But such systems, despite their steadily declining costs, remain expensive because you have to dig holes or mount cable on poles, which takes time and money.
Wireless clearly answers these problems. But until recently, no one agreed-upon standard was out there. That's all changed with the IEEE's recently approved 802.16 WMAN standard, which is finally ready. Chip and equipment companies are racing to get in on this potentially lucrative new market.
THE STANDARD MAKES THE DAY
Fixed wireless broadband isn't really new. Jim Kraemer, Smart Modular Technologies' director of product development and engineering, says he worked with this fast wireless technology in the military during the 1980s and 1990s. The telecom industry has used micro-wave for many years in back-haul applications. And, the technology has found homes in satellite data transmissions for decades now.
During the past 10 years or so, efforts to commercialize this technology led to the multichannel-multipoint-distribution-service (MMDS) and local-multipoint-distribution-service (LMDS) microwave systems. Today, these services are at a virtual dead end. These more recent efforts failed because of three key reasons.
First, they were extremely expensive. That's usually the case with microwave equipment. However, thanks to the development of Wi-Fi WLANs, semiconductor processing is at a point where the manufacture of microwave radio chips has become commonplace. Thus, the cost of producing microwave radios has dropped significantly.
Another part of the failure had to do with the lack of a single standard. All of the systems had their own proprietary designs. If you adopted the technology, you clearly had to stick with one vendor. Interoperability between different systems wasn't even considered.
Third, reliable connections can only be established by full line-of-sight (LOS) links at the higher microwave frequencies. This means gain antennas on high towers and clear paths with no trees, buildings, or other obstructions. This was difficult to achieve in some applications. But now, those problems have mostly been solved with the 802.16 standard.
Adopted by the IEEE in 1999, 802.16 originally focused only on operation in the 11- to 66-GHz band. Realizing that a reliable broadband service could only be achieved at the lower frequencies, companies went back to the drawing boards to create options to the standard that could produce a wireless winner. This led to the 802.16a amendment, which included a new standard option for the 2- to 11-GHz range. That extension has been rolled into the full 802.16 2004 standard and is now the focus of most of the attention in BWA.
Better still has been the creation of the Worldwide Interoperability for Microwave Access (WiMAX) Forum, an organization of over 150 chip companies, equipment manufacturers, and carriers. The group was formed to promote broadband wireless and create interoperability testing and certification to ensure that all chips and equipment work together compatibly.
The 802.16 standard is very broad and complex, and it offers many options. Probably the most widely used spectrum will be the part below 11 GHz, and especially below 6 GHz. BWA can be implemented in the existing unlicensed or licensed bands. The most likely ranges are the 2.5-GHz licensed spectrum, the 3.5-GHz licensed and unlicensed bands, and the 5.1- to 5.8-GHz Unlicensed National Information Infrastructure (U-NII) bands.
Small wireless Internet service providers (WISPs), perhaps in rural areas, may opt for the unlicensed bands due to cost. But larger urban carriers that expect to offer a full range of broadband options like TV and Voice over Internet Protocol (VoIP) will no doubt shoot for the lower bands, just to help minimize interference and ensure greater reliability.
The 802.16 standard is designed for both point-to-point (PTP) and point-to-multipoint (PMP) operation. The PMP mode assumes a cellular-like architecture, with each cell site covering a radius up to five or six miles and handling up to several hundred subscribers (Fig. 1). While the technology's maximum range extends just over 30 miles, it's not likely to be used except for some PTP back-haul applications.
Both time-division duplexing (TDD) and frequency-division duplexing (FDD) are supported. The FDD method with separate transmit and receive frequencies provides full-duplex operation at the expense of spectrum space. Typically, more spectrum is needed as Tx and Rx frequencies are spaced from 50 to 100 MHz.
FDD also supports higher throughput, as transmissions can occur during reception at either end of the link. The TDD mode employs only a single channel and provides multiple selectable time slots for multiple users. Half-duplex is required, which presents somewhat lower speeds. This shouldn't deter its use, though. The TDD mode also results in a smaller and less expensive CPE unit, because it only needs one synthesizer.
Channel bandwidth can be selected in steps from a maximum of 20 MHz to 10, 5, 2.5, or 1.25 MHz. This enables the carrier to tailor the service and price to the customer. The bandwidth and type of modulation set the data rate.
With a 20-MHz channel and optimized modulation (i.e., 256 quadrature amplitude modulation, or QAM), rates from 70 to 100 Mbits/s are possible. It's not likely such maximum rates will be used except in special back-haul applications. Some equipment uses a 28-MHz bandwidth, which can be subdivided into 14-, 7-, 3.5-, and 1.75-MHz bands to support lower data rates and more subscribers.
As for modulation, the standard incorporates an adaptive modulation scheme that lets the radios automatically select the optimum modulation depending upon the link range, noise, and other conditions. For close-in operation, where the signal-to-noise ratio (SNR) comes in at the 22-dB range, 64 QAM is used to supply super-fast data rates. As the SNR lowers to about 16 dB, the system backs off to 16 QAM. For an SNR near 9 dB, quadrature phase-shift keying (QPSK) is selected. For the worst conditions near an SNR of 6 dB, plain-old binary phase-shift keying (BPSK) is used. Data rates back off automatically, but the link's overall maximum reliability is ensured.
The 802.16a extension calls for non-line-of-sight (NLOS) operation. Successful broadband consumer service requires NLOS so customers won't have to install and orient outside antennas or eliminate the truck rolls by the carrier to do this work. An NLOS system means that the antenna can be in or on a consumer's set-top box.
To make this happen, the standard incorporates orthogonal frequency-division multiplexing (OFDM), a modulation/multiplexing/access technology that spreads a signal over a wide band. Like CDMA, it's the key to minimizing the effects of multipath signals, diffraction, fading, and other phenomena associated with microwave signal propagation. The 802.16 standard specifies a 256-point fast Fourier transform (FFT) OFDM signal, while a 2048-point FFT option is available for special circumstances. A single carrier option is available for super-fast LOS PTP back-haul applications.
Using extensive error-correction methods also enhances link reliability. The standard incorporates Reed Solomon forward-error correction (FEC), convolutional encoding, and interleaving algorithms to identify and correct bit errors. An automatic repeat request (ARQ) feature fixes errors not caught by the FEC by simply retransmitting the data packets with errors.
Power control is another feature to optimize each link. The basestations send power-control codes to the customer's set-top box to set the received signal to a specific desired level. The power-control algorithms improve overall performance and minimize power consumption. Power used by each subscriber unit will be a function of the propagation conditions, such as range, intervening obstructions, and similar factors.
Subchannelization, a special feature of the system, helps equalize the upstream and downstream link budgets. Typically, the customer-premise-equipment (CPE) unit transmits at only one-fourth the power of the basestation. To maximize the range of the link, the CPE unit can be modified on-the-fly to reduce the number of OFDM channels while increasing the power in each.
To further improve link reliability, the standard incorporates both transmit and receive diversity as well as adaptive antennas. Using properly spaced multiple directional antennas minimizes the effects of multipath, reflections, and fading that are normal in NLOS conditions.
The key to providing the quality of service (QoS) needed for a high-end broadband service is an access method that avoids all collisions and contention usually associated with wireless systems. Wi-Fi 802.11 WLAN systems use carrier-sense multiple access with collision avoidance (CSMA/CA). Multiple users must compete for access to the limited number of channels.
These systems employ a listen-before-transmit scheme but must back off and wait if another station is transmitting. This produces unpredictable time gaps that are simply unacceptable for QoS. The 802.16 system uses a request/grant access method that's similar to what's used in DOCSIS cable modems. Then, the basestation control fully eliminates incoming collisions and provides consistent and predictable delays. Such an arrangement enables voice and video to be transmitted completely and reliably.
The 802.16 standard didn't skimp on security either. The standard includes the widely used Triple Digital Encryption Standard (3DES). This 168-bit key provides highly secure encryption. However, future plans call for adoption of the even more secure Advanced Encryption Standard (AES). Security is no longer a disadvantage of wireless access.
While the 802.16a version of WiMAX will no doubt be the most widely adopted BWA system, the older system will find applications, especially in back-haul operations. Several popular operating points reside in the 10- to 66-GHz band, including the 10.5-, 25-, 26-, 28-, 31-, 38-, and 39-GHz bands, the so-called millimeter-wave bands.
OFDM isn't used, and modulation options include 64 QAM, 16 QAM, and QPSK. Bandwidth options are 20, 25, and 28 MHz. With a full 28-MHz channel and best modulation, the data rate can range from 32 to 134 Mbits/s. Typical range and cell size is one to three miles, depending upon frequency and propagation conditions.
APPLICATIONS FOR BWA
The most desirable application for BWA is "last (or first) mile" consumer broadband access. It will provide a third choice for consumers who can already get cable or DSL. Competition from BWA could help reduce costs and thereby greatly expand broadband to more homes.
Given that infrastructure costs are heavily based on installing a network of basestations and not laying cable, it appears likely that a carrier can make money in this arena. This is especially true if advanced services such as VoIP telephones and video services are offered.
No large telecom carrier has yet to commit to such a system, so Verizon and SBC are stepping in to test it. More likely, though, additional carriers will look at this potentially competitive service next year. Some smaller startup WISPs are emerging to serve selected areas, too. Most likely, the first systems may show up in some rural areas and smaller cities now underserved by broadband.
The first application of these systems could involve back-haul operations—that is, providing microwave links between cell-phone basestations and the carrier and from Wi-Fi hot spots to the carrier. It can also serve as back haul for its own multipoint cells. This application helps eliminate the high cost of running T1 lines. In fact, a whole new business is emerging to supply T1-equivalent service to small businesses via wireless means. Businesses certainly welcome such services that cut costs dramatically.
TowerStream, perhaps the leader in wireless broadband access, has already proved it can make a good profit by providing T1-like service to businesses via BWA. In business for four years, TowerStream doesn't use the new WiMAX standard. Instead, it goes for proprietary broadband systems in the 5.8- and 18-GHz range.
According to Jeff Thompson, TowerStream's president and COO, the company has BWA services in New York, Boston/Providence, Chicago, and Los Angeles. It also saved the day for both the Republican and Democratic conventions by setting up critical last-minute links and correcting some other communications difficulties with available wireless services. TowerStream will most likely adopt WiMAX as it grows.
A number of microwave equipment manufacturers offer non-WiMAX-compatible equipment for fixed networks. For example, Stratex Networks provides wireless back-haul equipment for Wi-Fi hot spots.
THE MOBILITY ISSUE
BWA is essentially a fixed broadband system. Basestations and CPE are set in place. Yet creators of the 802.16 standard seem hell-bent to build in a mobility option. Currently under development is the 802.16e or Mobile WiMAX option, which will potentially feature speeds reaching 15 Mbits/s in 5-MHz channels at a range of one to three miles.
That standard aims to create a version of the WiMAX system for use with a mobile customer unit, such as a modem-equipped laptop. In this way, users can roam within the system. If the handoff problem can be licked, the goal is wireless access in moving vehicles up to 75 mph with a data rate in the 500-kbit/s range.
Another mobility effort within the IEEE standards system is 802.20, or Mobile-Fi. This standard hopes to produce a high-speed wireless system using OFDM for any mobile application. Backed by companies like Cisco and Motorola, this could be competition for the fourth-generation (4G) cell-phone system.
WHO'S DOING WHAT?
Most companies addressing the BWA opportunity say that the product isn't quite there yet. Joe English, director of marketing for Intel's broadband division, indicates that Intel has a broadband chip, called Rosedale, in the works. It targets the CPE market for WiMAX systems. Intel is currently sampling the chip to key customers and will begin general sampling later this year. A baseband chip with standard Ethernet and voice ports, Rosedale is designed to work with RF chips from other vendors.
Fujitsu Microelectronics is hammering away at a similar chip. The chip, which supports the WiMAX standards, is expected to integrate both media-access controller (MAC) and physical-layer (PHY) sections on a single system-on-a-chip IC. The company expects an early 2005 introduction.
You needn't wait for single-chip solutions that are still in the pipeline, though, to build WiMAX radios. Numerous companies make RF receiver front ends and transmit chains. Dave Robinson of Analog Devices says that his company's AD986x series of mixed-signal front ends drops right into this application. Chuck Millet, also of Analog Devices, indicates that the ADI TigerSharc and Blackfin DSP chips are readily adaptable to WiMAX. He adds that for the basestations, a programmable solution seems the best way to accommodate all of the many 802.16 options and add or delete them as needed.
As for available equipment, there's little on hand. One exception is Alvarion, which offers the BreezeMAX 802.16-compatible system (Fig. 2). While not yet WiMAX-certified, it's WiMAX-ready—the company's equipment meets 99% of the standard and can fully comply with software upgrades later. Its CPE features standard Ethernet 10/100BaseT connection via the standard RJ-45. It supports up to 512 MAC addresses and can achieve a 12.7-Mbit/s data rate over a 3.5-MHz channel.
The CPE unit has standard RJ-11 plain-old telephone-service (POTS) ports, H.323, and SIP for VoIP service. Alvarion's basestation uses network processors to implement the traffic aggregation, classification, and connection. It supports 14-, 7-, 3.5-, and 1.75-MHz channel widths. It also works with the European Telecommunications Systems Institute (ETSI) broadband standard, which is similar to the WiMAX standard.
According to Alvarion's Carlton O'Neal, the company makes a wide range of other fixed wireless systems, including some near-WiMAX radios for the 5.8-GHz band. Other equipment vendors making proprietary broadband wireless equipment and Wi-Fi systems with their eye on WiMAX include Aperto, Airspan, Ensemble, Motorola, and Proxim. Look for many announcements beginning next year.
The year of WiMAX seems to be 2005. Not only will chips be available, but many equipment vendors also will have products by year's end. The big holdup is the WiMAX testing and certification program, which has yet to commence. It's expected to be in operation by mid-2005. From then on, look for some growth as products roll out and carriers and WISPs decide their business strategies.
As with any wireless technology, it all takes time, as was demonstrated with Bluetooth, Wi-Fi, and 2.5G/3G cell phones. Market research firm iSuppli projects the BWA equipment market to soar from $80 million in 2003 to $2.3 billion in 2008 (95.3% growth). The market for semiconductor products is expected to balloon 113% over this same five-year period.
|NEED MORE INFORMATION?|
Analog Devices Inc.
Fujitsu Microelectronics Inc.
Institute of Electrical and Electronics Engineers
Stratex Networks Inc.