November 28, 2004
WiMAX: Broadband Wireless Access And Beyond
Proponents want WiMax to be as successful in the metro as Wi-Fi has been on the premises. Is that possible?
Metro-area microwave has been around for decades, supporting everything from point-to-point private links to video-on-demand. Most recently, non-line-of-sight (NLOS) technology and quality of service (QOS) enhancements have made microwave viable for last-mile access as well. With new standards in place-the IEEE 802.16 family in the US and the very similar ETSI HIPERMAN in Europe-the WiMax Forum is working on the same kind of interoperability guarantees, commodity silicon and industry-wide marketing for metro-area microwave that have made local-area wireless, i.e., Wi-Pi/802.11. such a success. (WiMax is short for Worldwide Interoperability for Microwave Access.)
Wi-Fi was portable from the start, so it didn't compete directly with wired Ethernet. In contrast. WiMax competes in fixed broadband wireless access (BWA) applications with proprietary microwave and optical products, as well as cable and DSL (see "Free-space Optics In The Metro"). Customers with functioning solutions in place don't feel pressured to replace them.
The other primary application anticipated for WiMax-portable/ mobile broadband wireless-is shaping up to be a hotly contested niche, with new cellular, Wi-Fi and optical technologies emerging to challenge WiMax. In addition, WiMax faces spectrum availability uncertainties, with the FCC undecided about opening up more bands and perhaps reallocating some underutilised bands.
Nonetheless, WiMax start-ups are coming out of the woodwork. The WiMax Forum's website lists more than 130 members, including more than 30 equipment vendors, plus scores of component and test equipment providers. Cisco joined the Forum in September, so at least one of the startups could hope to be acquired, but most of the rest probably won't be around in five years.
FIGURE 1 For Line-Of-Sight Transmission, The Entire Fresnel Zone Must Be Unobstructed
Meanwhile, both Intel and Fujitsu have announced WiMax chips, legitimizing the technology and fueling the hope that rapid, inexpensive, interoperable WiMax products will stimulate experiments and acceptance, a la Wi-Fi. Let's look first at the technical improvements that give WiMax so much potential, then check the views of a few early adopters, proponents and analysts.
Kicking The Fresnel Football
Microwave, which includes electromagnetic frequencies from about 1 GHz to about 300 GHz, is the basis of radar, microwave ovens and many other technologies. (In World War II, sailors warmed themselves by sitting in front of radar antennas. They didn't know they weren't just warming up-they were being cooked.)
In the networking realm, frequencies above 11 GHz are used for high-capacity, point-to-point transport, but these applications require an unobstructed line-of-site (LOS) between antennas. An entire football-shaped "Fresnel zone" must be clear of buildings, trees and other interference (Figure 1). Over longer distances, even the curvature of the earth can cause problems. LOS constraints have limited microwave's utility in many urban and rural settings.
New microwave systems, including 802.16/HIPERMAN, that operate below 11 GHz are more powerful and don't require a clear LOS. Their signals pass through windows and walls, bounce off buildings and trees, and still arrive intact. Like Wi-Fi, these systems employ Orthogonal Frequency Division Multiplexing (OFDM), a modulation technique that resists interference and distortion.
Together, NLOS transmission and OFDM make the new microwave systems practical as BWA options for rural and suburban locations that are too far from a central serving office for DSL or cable access, and for inner city high-rise offices that lack fiber. "Fiber may reach lots of [office] buildings, but it often doesn't reach tenants high up in office towers," said Keith Doucet, vice president of marketing and planning for Redline Communications, a maker of broadband wireless equipment. "With microwave, you can put an antenna in the office window for non-line-of-sight broadband service."
Besides an increased geographic reach, the WiMAX technologies in particular have some specific QOS capabilities that previous BWA, Wi- Fi and other wireless technologies lack. These capabilities are expected to become more important in the future, as customers combine more of their voice and data applications and traffic (see "Serious QOS Capabilities"). In the meantime, pre-WiMax broadband wireless implementations are mostly being used for Internet access, municipal networks and private data applications.
Will o' The WISP
Wireless ISPs (WISPs) like TowerStream are eager users of new BWA technology. TowerStream sells wireless Internet access based on pre- WiMAX equipment from Aperto and other suppliers; its service is available in New York City, Chicago, and in the greater Boston and Rhode Island metre areas. Prices start at $500 per month for T1- equivalent service; 100-Mbps service lists at $5,000 per month.
FIGURE 2 TowerStream's Boston-area Network Includes Six Base Stations
TowerStream claims more than 700 enterprise customers. To reach them, it surrounds each metro area with microwave base stations. The Boston area, for example, has six (Figure 2). Each base station extends service over a 13-mile radius. Overlapping coverage provides redundancy in case of base station failure, and high-capacity wireless links between the base stations form a self-healing, SONET- like wireless ring. Several of the base stations also connect to Tier 1 ISPs for Internet access and to LHCs for access to the PSTN. Base station redundancy and the self-healing ring give TowerStream the confidence to guarantee 99.99 percent uptime.
To connect an enterprise, TowerStream installs a small transceiver on the customer's roof, in a window or on the side of the building, and points it at the nearest base station. Careful tuning and NLOS capability minimize interference from intervening obstacles and other microwave traffic. Customers can use BGP to include the TowerStream link as part of their multihomed Internet access, and it's not unusual for them to start with wireless as backup for wireline.
Typically, however, customers drop their wirelinc connection when they see how reliable wireless is, according to Philip Urso, TowerStream's founder and CEO. "Over 95 percent of our customers use wireless as their primary connection," he said.
Urso suggested that wireless Internet access appeals to enterprises for three reasons: First, $500 per month is a good price and, second, customers like the fact that everything is separate from the usual landlines, including the links between base stations: It's hard to take out a rooftop antenna with a backhoe.
Finally, microwave goes in fast compared to landlines. Installation normally occurs within two weeks, although TowerStream will guarantee 48 hours. Urso likes to brag about the time they brought a hotel-based conference online between 9 a.m. and noon.
Allegany County Goes Wireless...Twice
Municipal governments and school systems are also interested in the new microwave systems, primarily to keep their operating expenses as low as possible. "In school systems, it's easy to get onetime funding, but not ongoing funding," explained Jeff Blank, supervisor of networking for Allegany County public schools in western Maryland, in the foothills of the Appalachian mountains.
In 1994, the school system partnered with local government to install early LOS microwave products from Alvarion, then known as BreezeCom. The network, called AHCoNct 1, brought 45Mbps data connections among the sites and Internet access to about 90 buildings, including the Board of Education, the City of Cumberland, public libraries and the county government.
"The ongoing cost was next to nothing," said Blank. "Compared to Tl lines, our cost-avoidance was around $70K per month."
AllCoNet I was so successful that the county is now building AllCoNet 2, based in part on Alvarion's latest NLOS product line. AllCoNet 2 comprises 16 base stations, each supporting three point- to-multipoint frequency bands: 900 MHz for computers in police cars and other mobile applications, 2.4 GHz for small businesses, and 5.8 GH/, for larger enterprises. Licensed frequencies are used on the backbone to avoid interference, and the OC-12 (622-Mbps) microwave links between the base stations form a redundant SONET-like loop. Officially, each base station serves a 5- to 8-mile radius, but unofficially, links up to 11 miles are possible.
The primary motivation for AllCoNet 2 is economic development: By creating a broadband infrastructure, the county hopes to attract new businesses and encourage ISPs to bring affordable broadband Internet access to county residents. The major telecom provider's rates were not competitive, users are too spread out to attract DSL or cable providers, and the county could not afford $189 million to lay its own fiber.
With an installation cost of just $4.7 million and minimal operating costs-"Lunch money," said Blank-an updated microwave solution made the most sense. Al\lCoNet 2 will provide Internet access to government and other non-profit users, and will charge low fees to any commercial ISPs who want to offer services to business and residential customers.
A Few Speed Bumps, Then A Virtuous Cycle
There are several WISPs, like TowerStream, and many successful municipal and private microwave deployments, but continued market growth will depend on attracting new types of customers: more business and residential BWA users, and even portable/mobile users. To penetrate those market segments, WiMax will have to be both cheap and functional.
One industry observer said the long-term ideal is 50/50/50 service: 50-Mbps wireless links that reach distances up to 50 kilometers for a CPE price of $50. That ideal is years away, with current services more like 4 Mbps over 4 kilometers using CPB that costs about $400.
FIGURE 3 Worldwide Point-to-Multipoint Sub-GHz BWA Equipment Sales
Recent developments, however, are moving things in the right direction. First of all. the IEEE has continued to grow the 802.16 family of standards. Originally, for example, the Wireless MAN Air Interface specification included only LOS transmission in the 10- to 66-GH/. range. In early 2003, amendment 802.16a added NLOS transmission in the 2- to 11-GHz range and a speed/distance maximum tradeoff of either 70 Mbps or 31 miles.
Standardizing what most equipment vendors already offer in proprietary form. 802.16 also includes QOS features and 3DES encryption. A consolidated version of the standard, including all the amendments, was approved in June of this year as IHEE 802.16- 2004 (see "Sorting Out the Standards.").
With 802.16 and its ETSI counterpart, HIPERMAN, as the foundation, the WiMax Forum, founded in 2001. is working toward international multivendor compatibility by creating interoperability profiles and appointing test labs. Approved labs should begin testing vendor equipment in late Q1 or early Q2 2005. "WiMax certification will be similar to Wi-Fi and CableLabs," said Mohammad Shakouri. vice president of the WiMax Forum.
The job of the WiMax Fomm is complicated by the diversity of spectrum allocation rules around the world. Each country or region makes its own rules concerning which parts of the microwave spectrum may be used, which are licensed, and which are unlicensed (also known as license-exempt). WiMax interoperability efforts will begin in the 2.5 GHz, 3.5 GHz and 5.8 GHz bands because those serve the most countries.
WiMax certification is intended to promote customer confidence, multivcndor interoperability and inexpensive components. "WiMax and 802.16 will let chip vendors commoditize the physical layer so that equipment vendors can focus on higher layers," said Alan Menezes, vice president of marketing at Aperto Networks, another maker of broadband wireless equipment.
In fact. Intel has already sent samples of its WiMax system-on-a- chip to key customers. Codenamed "Rosedale," the chip supports the IEEE 802.16-2004 MAC and OFDM PHY layers, a 10/100 Ethernet MAC, security processing and a TDM controller interface. Fujitsu has also promised a single-chip WiMax product for early 2005. The availability of industry-standard chips, plus Intel's marketing strength, could fuel a virtuous cycle of increasing volume and decreasing prices, just as it did for Wi-Hi.
WIMax And WFI? WI-Not?
WiMax and Wi-Pi are frequently mentioned in the same breath, and there are some similarities beyond their nicknames-both are wireless (duh!) and both use OPDM-but actual relationships are still developing.
One frequently mentioned possibility is the use of WiMax to backhaul ISP traffic from Wi-Pi hotspots. It's a fine idea, but there aren't any obvious technical synergies. WiMax, or for that matter any BWA technology, simply offers an attractive alternative to leased lines for point-to-multipoint services, backhaul included. BWA's advantage is that the ISP doesn't have to wait for or pay a monthly rate for a LEC access line. Of course, these benefits would apply not just to Wi-Fi hotspots, but to any premises business and any LAN technology, including fiber or Cat 5 copper.
In fact, BWA could prove a bypass boon to all competitive carriers, not just hotspot providers. With the changes in UNR-P regulations, alternative carriers are more eager than ever for ways to reach customers without the cost and rigmarole of colocation in the LRCs' central offices. As WISPs like TowerStream demonstrate, BWA is a viable alternative. One scenario described, and no doubt dreamt of, by equipment providers is the use of BWA by ILRCs to deliver voice and data services in each other's territory.
The other frequently mentioned match-up of Wi-Pi and WiMax-for mobile or portable broadband wireless access-is more exotic and farther in the future. Today's WiMax and 802.16 specs cover only fixed wireless links.
Late next year, the IRHR 802.16 working group will issue a new amendment (802.16e) that describes a mobile or portable version of BWA. The idea is that you'll be able to open a VPN, for example, connecting to your Manhattan office while strolling through Central Park (if a mugger doesn't grab your PC). In the ultimate Wi-Fi/ WiMax marketing vision, laptop computers, PDAs and other portable gear will ship with both Wi-Fi and mobile WiMax chips built in. The two wireless services will somehow be integrated, as vendors are trying to integrate Wi-Fi with cellular data access today, and you'll be able to go virtually anywhere, indoors or out, and maintain your wireless Internet connection. Someday.
Will WiMax Rule The Broadband Wireless Market?
Even without extraordinary applications, worldwide broadband wireless equipment sales are projected to grow to over $2.5 billion per year by 2008, according to Emmy Johnson, principal analyst at Sky Light Research (Figure 3) Much of this growth will occur overseas and in rural parts of the U.S., she said, where microwave can be used to reach users sooner and more cost-effectively than DSL, fiber or cable.
Johnson doesn't expect WiMax to dominate metro wireless in the next five years to the extent that Wi-Fi now dominates wireless LANs. She predicts that more than half of all broadband wireless equipment sales in 2008 will be for portable/mobile applications, most of which will not be WiMax. Overall, she said, only about a third of the equipment sold in 2008 will be WiMax. She attributes the slow growth to the many choices customers and carriers will have for portable/mobile applications and the wide choice of solutions they already have for BWA applications. And, as Aperto's Menezes pointed out, "Non-standard equipment will work side-by-side with standard WiMax gear," so there's little motivation to upgrade the existing BWA gear.
The wild card in the game is spectrum allocation. Right now, most of the 2.5-GHz band in the U.S. is licensed to Sprint and Ncxtel for Multipoint Metropolitan Distribution System (MMDS) service. Originally conceived as an alternative to cable TV, then re- purposed for Internet access, MMDS technology never caught on. Sprint and Nextel could use their MMDS spectrum for BWA, or in a more extreme scenario, the FCC could take back idle spectrum in the 2.5-GHz band and auction it off all over again.
Meanwhile, the WISPs and other BWA providers share the unlicensed 3.5-GHz and 5.8-GHz bands. These are subject to more interference than the licensed bands, although Redline's Doucet maintains this is not much of a problem. "In unlicensed bands, service providers cooperate to divide up the region and the spectrum," he said. Moreover, the FCC has been gradually opening more spectrum in the unlicensed bands, and interested parties are lobbying the agency to reassign frequencies currently reserved for other purposes.
Resolving the spectrum issues would remove some uncertainty from the BWA market, but reducing the price of equipment also would help. According to Emmy Johnson, BWA CPE prices, which averaged about $425 in 2003, should drop to around $120 by 2008. On the other hand, she expects the average base station price to increase from about $25,000 today to about $40,000 in 2008 as the base stations add radios and coverage to support more users.
BWA in general, if not WiMax in particular, is bound to succeed in some fashion. Its short installation intervals and low-cost operations are a winning combination. Add sufficient spectrum availability and excellent QOS, and fixed wireless bypass applications might become quite popular-even without interoperability or standards.
Whether WiMax becomes as successful as WiFi is a separate question. The ultimate size of the WiMax market will depend to a large extent on the popularity of mobile data applications, as the real volume can't come until WiMax chips go into PCs. In the mobile/ portable arena, WiMax will have many competitors. Even if 802.16 shuts out 802.20 and other data-oriented mobile technologies, there are still battles to be won with Verizon's Evolution-Data Optimized (EV-DO) and other 2.5 and 3G cell phone technologies for Internet access. On the other hand, if voice eventually becomes just another data type, WiMax's superior QOS may give it a better chance of becoming the Wi-Fi of metro-area wireless
Companies Mentioned In This Article
MRV Communications (www.mrv.com)
Skylight Research (www.skylightresearch.com)
Wi-Fi Alliance (www.wi-fi.org)
WiMAX Forum (www.wimaxforum.org)
Portable/mobile applications may have a tougher time than fixed broadband access applications
Free Space Optics In the Metro
Microwave isn't the only metro-area wireless technology available. Fre\e space optics (FSO) systems use near-infrared laser light to carry point-to-point digital signals. FSO transceivers create physical-layer links that operate much like optical fiber, but send the light through the air instead of a glass strand.
Like microwave, FSO can be deployed quickly. According to Isaac Kim, director of optical transport for MRV Communications, FSO systems can be aligned in less than half an hour. "The biggest delay in FSO deployment is the sales cycle," said Kim. MRV has been installing FSO solutions since the early 1990s, Kim continued, "but the market still isn't comfortable with the technology."
On the plus side, FSO is subject to fewer government regulations than microwave, since it doesn't clash with radio frequencies. That makes FSO useful for cross-border links and other installations that would otherwise get tangled in red tape. And FSO has a much narrower beam than microwave, making it difficult to intercept or jam.
On the minus side, FSO doesn't reach as far as microwave, and its reach is affected by fog, snow and other atmospheric conditions. A typical 155-Mbps link, for example, can reach seven or eight kilometers in very clear air, but only about five kilometers under normal conditions. And equipment for a Fast Ethernetspeed FSO link from MRV costs about $5,000, which seems expensive compared to WiMax's $400 CPE and $25,000 base station (no base station is needed with FSO)
Serious QOS Capabilities
Per-flow quality of service (QOS) is probably WiMax's strongest differentiator from other wireless access technologies. Wi-Fi has IEEE 802.11e, which supports limited prioritization on a single connection between the endpoint and the Wi-Fi access point. WiMax, on the other hand, allows multiple connections between a subscriber station and a base station, and each connection can have its own QOS attributes.
IEEE 802.16 defines four types of QOS:
* Unsolicited Grant Service (UGS), designed to support constant- bit-rate applications, such as T1 emulation and voice over IP (VOIP) without silence suppression.
* Real-Time Polling Service (rtPS), for applications that generate periodic variable-size packets, like MPEG and VOIP with silence suppression.
* Non-Real-Time Polling Service (nrtPS), which supports applications like FTP that generate variable-size packets on a regular basis.
* Best Effort (BE) Service, for low-priority applications like Web surfing.
To implement the QOS levels, the base station polls the subscriber stations for bandwidth requests and schedules the requests it receives. The frequency and regularity of the polling depends on the QOS type of each subscriber station's connections. For example, rtPS connections receive periodic unicast polls, while BE connections are never polled individually, but must respond to multicast "contention request opportunities" or piggyback their requests on data traffic.
Once a subscriber station is allotted airtime, the usual queuing techniques decide which packets get to use it. The result is efficient bandwidth utilization and ATM-like support for a wide variety of applications
The WIMax Forum hopes to emulate the Wi-Fi Alliance, but WiMax has some different hurdles to clear
WIMax offers QOS capabilities that the other wireless technologies don't
Sorting Out The Standards
WiMax is to 802.16 as Wi-Fi is to 802.11. That is, the IEEE defines the standards, while the WiMax Forum and the Wi-Fi Alliance promote real-world interoperability. The IEEE 802.16 Working Group has been hard at work on "the Wireless MAN Air Interface for wireless metropolitan area networks" since July 1999. Here's a summary of the group's output:
-802.16, approved December 2001, specifies the PHY and MAC (including QOS) layers for line-of-sight wireless MANs in the 10- to 66-GHz range.
- 802.16a, approved January 2003, amends 802.16 to include wireless MANs in the 2- to 11 -GHz range for non-line-of-sight connections up to 31 miles at speeds up to 70 Mbps.
- 802.16e, approved December 2002, adds 10- to 66-GHz system profiles to aid interoperability.
- 802.16REVd, issued only in draft form and sometimes referred to as 802.16d, revises and replaces 802.16, 802.16a and 802.16c.
- 802.16-2004, approved June 2004, is the officially released version of 802.16REVd.
- 802.16e, targeted for late 2005, will add mobility features to 802.16-2004.
ETSI HIPERMAN, the European
Telecommunication Standards Institute's standard for broadband wireless MANs in the 2- to U-GHz range, shares MAC and PHY specifications with IEEE 802.16. The main difference is that HIPERMAN covers only the 2- to 11 -GHz range, while 802.16 goes all the way up to 66 GHz. The WiMAX Forum is working with both IEEE and ETSI on testing to ensure worldwide compatibility
Wireless Internet access costs less and goes in faster than wireline alternatives
WIMax could offer a bypass boon to competitive local service providers
Bob Bellman, president of Brook Trail Research, specializes in new technology introduction and education. He can be reached at [email protected]
Copyright Business Communications Review Nov 2004