Bridge makers try to cross over to the mainstream
In the wireless data business, as elsewhere, there are hard sells and easy sells. Wireless bridges are a comparatively easy sell
Are commercial carrier services an alternative to the enterprise ownership of wireless bridge equipment?
In the wireless data business, as elsewhere, there are hard sells and easy sells. Wireless bridges are a comparatively easy sell. By any measure of return on investment, a wireless bridge is a winner—provided, of course, that it matches the throughput requirements of the enterprise. Any equally reliable, equally fast leased-line solution for connecting geographically separated parts of the enterprise is going to cost more in the midterm than a privately purchased wireless bridge—that is, within a two-year span, worst-case. And, in some instances, a wireless bridge can pay for itself in reduced monthly fees within a single month.
"No question that wireless bridges are the easiest wireless networking devices to recommend," says Eric Erickson, vice president of marketing for Aironet (Akron, OH). "The light comes on first with wireless bridges because you only have to compare T1 prices with the purchase price of the bridge equipment."
Industrial-strength bridges support network management, longer distances, and a variety of wireless scanning devices.
So why aren't wireless bridges ubiquitous? "Not enough people know about them," explains Joel Kmetz, product manager at Solectek (San Diego), which has designed bridges for a decade. "We still have to spend a lot of our time doing missionary work."
Though many wireless experts share Kmetz's views, indications abound that a change is in the offing. Where as before enterprise bridges have mostly been the province of relatively small manufacturers, three major telecommunications manufacturers have entered the business in the last few years: ADTRAN (Huntsville, AL), Southwest Microwave (Tempe, AZ), and Adaptive Broadband (Sunnyvale, CA—formerly California Microwave). Furthermore, a rapidly growing number of Internet service providers (ISPs) such as PSINet and Time-Warner have chosen to include unlicensed wireless bridges in their networks while venture finance firms have recently invested in wireless bridge manufacturers.
Category redefining itself
The wireless bridge has traditionally been defined as a relatively long-range, point-to-point radio link, generally located out of doors. Until recently, such devices resembled indoor radios in basic construction. A bridge might use a more powerful amplifier than a network interface card (NIC) or an external PC radio modem and would probably have a directional antenna of some considerable physical size in contrast to the stubs attached to NICs. The radio-frequency (RF) component itself would tend to resemble that incorporated within the transceivers used at the desktop.
Today, bridges provide key network-management functions as well as software transport protocols not present in the wireless links used for the individual data terminal. Indeed, some products include an entire router and incorporate the kind of redundant amplification and RF sections used in cellular base stations and licensed microwave towers.
Perhaps the biggest distinction today is between bridges using an optical medium and those propagating radio waves.
Glenayre's Tsunami 10Base-T bridge provides license-free wireless interconnection over long distances without decreasing bandwidth.
Optical bridges are by far less common and seem destined to retain their minority status. At the same time, optical products have become price competitive with RF components—not the case, a few years earlier—and at least one such product has achieved a speed advantage so compelling that in certain applications it is likely to be without significant RF competition.
The optical category is itself subdivided into two groupings: diffuse infrared bridges and laser bridges.
The first type uses essentially the same type of transmission technology as an infrared remote control: short pulses of noncoherent infrared light. Diffuse bridges operate at fairly short ranges, normally a few hundred yards, but offer impressive throughputs, as high as 30 Mbits/sec in some instances. Prices are in the thousands of dollars per connection. There are relatively few manufacturers, including Lightpoint Communications (Boulder, CO) and InfraLAN Wireless (Acton, MA).
Laser bridges use laser light as their medium of transmission and exceed the diffuse type in both range and throughput. Extreme ranges for some models can extend to miles, and throughputs are commonly in the hundreds of megabits. Manufacturers include Eagle, AstroTerra (San Diego), Canon (Englewood Cliffs, NJ), and others. A new model recently announced by Lucent uses the same dense wavelength-division multiplexing (DWDM) technology seen in fiber-optic backbones whereby several carrier frequencies are transmitted and received simultaneously. This technique permits a claimed maximum data rate of 10 Gbits/sec over the air, a figure that seems plausible given the performance of existing DWDM fiber-optic linkages.
In addition to superior speed—now being challenged by RF bridges in the case of the diffuse optical bridge—the optical connections offer superior security and absolute immunity from electromagnetic interference. The bridge itself radiates no RF energy to interfere with other wireless devices, nor is its own performance affected by such energy. Because transmissions are concentrated in extremely narrow beams, interception and eavesdropping are practically impossible.
Despite their high speeds, especially in the case of the laser bridges, neither optical type has found wide acceptance. Where 30-mile links are now commonly achieved with RF bridges, little progress has been made in extending the range of their optical counterparts, either laser or diffuse. But more significantly, optical bridges are strictly line-of-sight subject to interruptions in the presence of interfering light sources as well as dense fog, heavy precipitation, dust storms, and foliage. For this reason, free-air optical solutions have found little acceptance on the carrier level, and carrier applications are increasingly driving technology development within the overall wireless bridge category.
RF, the dominant category, also forms a number of well-defined classes. The two most fundamental groupings are RF bridges using licensed spectrum and those operating within the unlicensed bands.
Licensed microwave bridging equipment has typically been expensive and aimed at the carrier community. Enterprise class products are a fairly recent phenomenon, and only a relative handful of companies is active in this market, including Racon (Seattle, WA), Southwest Microwave, Glenayre Western Multiplex (Sunnyvale, CA), TTI (Cleveland, OH), and Adaptive Broadband. Most of this equipment operates in the 23-GHz band.
Nearly a gigahertz of bandwidth is available at 23 GHz over several separated channels. Several hundred megabits of throughput are theoretically possible, but nothing on the market has achieved such speed to date, though TTI has announced a 100-Mbit/sec full-duplex product. Licensed equipment for private use is priced in the tens of thousands of dollars per link, and ranges are relatively short—normally under 10 miles. The growing interest of the carrier and ISP communities in 23-GHz equipment ensures quick evolution.
The Wi-LAN 30-Mbit/sec bridge uses a simultaneous multicarrier signaling scheme.
Unlicensed bridges, by far the dominant type today, may occupy any of several bands, including 902 to 928 MHz, 2.4 GHz, 5.1 GHz, 5.2 GHz, and 5.8 GHz. Most currently operate at 2.4 and 5.8 GHz, with the latter rapidly gaining popularity. The 5-GHz bands are much less prone to interference than the others and have generous allocations of bandwidth. But there is still some shortage of original equipment designed for these frequencies.
Unlicensed bridges have dramatically increased their throughput and distance in the period since we published our last technical update. Two years ago, only one product approached the 10-Mbit/sec Ethernet standard, and most products did not exceed 2 Mbits/sec. Now, WaveSpan and Glenayre offer 100-Mbit/sec products, with other companies to follow. Several sell systems exceeding 20 Mbits/sec.
Distances have also improved markedly over the last year. Twenty-five and 30-mile separations between bridging points are now almost routine, while 10 miles was a stretch as recently as last year. The wireless bridge is becoming a locus for distributed network intelligence and concentrator for network traffic. It is fast evolving as the key element in a metropolitan area network.
The notion of using unlicensed spectrum for providing common carrier services has generally been credited to San Jose-based Metricom, which began to build its network in the early 1990s and currently operates in three major metropolitan markets. For years, Metricom, whose particular service offering requires massive physical infrastructure, struggled to build out its network and secure right-of-way in metropolitan markets, and its slow progress discouraged others to follow its example. But in late 1998 and through this year, a multitude of ISPs, most located in tertiary markets, have initiated service, many using the high-speed, high-capacity bridging products that have only recently become available. Such companies include Global Pacific Internet throughout Orange County, CA; PSINET in over a dozen cities in the South; Nobell Communications in Austin; Last Mile throughout much of Pennsylvania; and DirectNet in South Florida. Each has succeeded in attracting thousands of subscribers and competing effectively with digital subscriber line and T1 in the small business and small office/home office markets.
According to most manufacturers, the rapid evolution of the bridge is primarily a result of one factor—the increasing interest of ISPs and competitive local-exchange carriers in unlicensed wireless solutions for last-mile connections. Representatives from BreezeCom and Aironet say more than 300 ISPs nationwide have purchased wireless bridging equipment to date. While enterprise users are not necessarily candidates for much of the new carrier class equipment, they are in a position to benefit from their advances. Just as auto racing has been said to "improve the breed" of the passenger car, competitive public network services are helping refine equipment designed to operate in the unlicensed bands.
New ways to radiate
For years, wireless bridges occupying the unlicensed space used the same general spread-spectrum technologies employed in the indoor environment: direct-sequence and frequency-hopping. Then, two years ago, RadioLAN (Sunnyvale, CA) introduced a 5-Mbit/sec plus bridge using a single frequency and transmitting within the newly available 5.8-GHz band. Since then, many have followed.
"You don't need spread spectrum in the UNI band," states Ken Ruppel, director of product management for Glenayre Western Multiplex. "That's something that was mandated for 2.4 GHz where controlling interference is such an issue. In fact, you don't want spread spectrum because it limits the throughput as the price for increasing robustness. UNI is dedicated to data, so we're free to use more spectrally efficient techniques."
Generally those techniques are simply phase or amplitude modulation, though sometimes of a surprisingly high order. The WaveSpan Stratum 100 100-Mbit/sec bridge, for instance, uses a 256 QAM (quadrature amplitude modulation) system, which means that the receiver recognizes 256 different signal levels. The new WiLAN 30-Mbit/sec bridge uses a simultaneous multicarrier scheme, while the RadioLAN equipment now uses pulse-position modulation. Another method for improving throughput is to transmit on two or three channels simultaneously. OTC is currently the only manufacturer combining three channels at once (for a total of 33-Mbit/sec throughput).
Another strategy, as yet only used by ioWave, is to employ adaptive array antennas where smart circuitry is used to shape beams according to who is using the network at any one time. Theoretically such technology could improve spectral efficiency some fiftyfold, although much more modest gains are likely initially.
Some of these techniques can be combined, so the possibility of transmitting several hundred megabits per second, a speed approaching that of LMDS networks or even Synchronous Optical Network, is real. The category topped out at 2 Mbits/sec four years ago, and now has achieved 50 times that throughput, best-case.
Perhaps the biggest changes in wireless bridges have occurred in the areas of overall network architecture and management.
"A move in the direction of multipoint bridging marks an important development in the category," notes Mark Bosse, vice president of marketing at RadioLAN, the first manufacturer to offer a product with a data throughput approaching 10 Mbits/sec. "We're doing it and so is C-SPEC and WiLAN. It's something the carriers are pushing for." And indeed many ISPs do report using bridges with sectorized antennas essentially as access points for a multitude of individual subscribers. On an enterprise level this same technology would allow the administrative headquarters for a hospital, municipality, or school district to reach many widely scattered locations with a single bridge.
A bridge used in such a manner defines the hub of a cell in much the same way that an access point does in an inbuilding local area network. But to function effectively in an outdoor environment serving multiple subscribers at widely varying distances and usage patterns and throughput requirements, the bridge must have far greater network-management and channel-reuse capabilities than an inbuilding base station.
Private bridges: a good choice
The most advanced wireless bridges and distribution points are costly. The 100-Mbit/sec bridge costs tens of thousands of dollars per link and is apt to remain expensive for some time to come. But on the low end, the price of bridges has dropped to less than $2,000 for a simple T1-speed connection, sufficiently inexpensive to pay for itself in a couple of months at current T1 pricing. And overall, the price per bit of throughput of wireless connections has been steadily declining while the distance and reliability of the equipment as well as ease of installation have seen just as steady an improvement. With the growing use of wireless bridges in public networks, we may expect further price reductions through economies of scale and increased competition. In the not-too-distant future, the private bridge may become the medium-distance data link of choice.
David Sweeney is a technical columnist for Wireless Integration, a sister publication.
This article originally appeared in the Sept-Oct 1999 issue of Wireless Integration.
Smoke signals: Solectek's new vertical for wireless connectivity
Since 1975, recognized Native American Tribal Nations have enjoyed greatly increased autonomy within their own lands. But with sovereignty have come numerous challenges, including modernizing the basic infrastructure of roads, water supplies, and communications. The last has been a particular problem because tribal lands often lie in remote and sparsely populated areas, distant from central-office facilities, and they sometimes straddle a number of local access and transport areas. In such instances the cost of conventional landline connections can be prohibitive.
Solectek, whose headquarters are located adjacent to tribal lands in San Diego County, recognized the problem and created a solution that even cash-strapped tribal governments could afford. With a line of price-competitive wireless routers and bridges designed to operate at distances of up to 25 miles, Solectek proposed to construct what were in effect cooperative local Internet service providers that would help put the Nations firmly in control of their own destinies.
To date, Solectek has put in three networks comprising wireless bridges: a 17-bridge network in Isleta Pueblo south of Albuquerque, an even larger network spanning the Chickasaw Nation in eastern Oklahoma, and a sprawling system serving the Sioux Tribe of Lower Brule in South Dakota.
All of these systems are works in progress, but each is already considerably larger than most municipal or enterprise systems. The Sioux network, for instance, extends to more than 300 users and provides security via video surveillance as well as Internet access. The Chickasaw goes even further, providing local-area-network extension for businesses, Internet access, and intranet linkage for Tribal agencies. Says Mark Abbott, manager of Smoke Signals, the Chickasaw Nation-owned computing and networking company, "We plan to increase our use of video and run video over Internet protocol [IP]. Our plans are to implement videoconferencing in the regional centers to facilitate meetings with the Nation." Abbott also indicates distance-learning and telemedicine are in the works.
The Sioux of Lower Brule have even more ambitious plans, says Steve Sievert, information systems manager for the Tribe. "I'd like to run voice and fax over IP. It would be great if we could install a PBX [private branch exchange] and replace all of the public phone lines."
The well-managed network
You can't readily extend the Ethernet model over a metropolitan area," notes Joel Kmetz, product manager at Solectek (San Diego). "You have to go for more sophisticated protocols. When you have users widely scattered around the hub, the ones closer will get much higher throughput when the ordinary Ethernet CSMA [carrier sense multiple access] protocol is employed because they'll experience fewer collisions. You need to substitute a polling protocol where you don't transmit until spectrum is available."
Polling protocols—or reservation protocols that accomplish much the same thing—have in fact been incorporated not only in the Solectek products, but in bridges from WaveSpan, Rooftop Communications (Mountain View, CA), Adaptive Broadband, WiLAN, OTC Telecom (San Jose, CA), NEC (Irving, TX), Air Data WIMAN (Naples, FL), and others. The Adaptive Broadband and NEC products are notable for using the Asynchronous Transfer Mode (ATM) protocol, which has become almost ubiquitous in equipment made for the licensed LMDS bands, but hasn't found much acceptance to date in the unlicensed sector. Uniquely, the Air Data WIMAN product uses a wireless version of frame relay, while both Aironet and OTC support Token Ring. All such systems permit, to varying degrees, prioritization of transmissions, and, in the case of frame relay and ATM, also provide for guaranteed bit rate and controlled latency for multimedia services such as videoconferencing and packet voice.
Interestingly, provisions for polling are not to be found in the IEEE 802.11 wireless LAN compatibility specification. Accordingly, achieving 802.11 compliance is not a major objective for most manufacturers. "When you're talking about ranges beyond 1,000 ft, almost everything out there is proprietary," says Eric Erickson, vice president of marketing at Aironet (Akron, OH).
"802.11 doesn't work very well in the outdoor setting," agrees Pete Bonk, general manager of WaveRider USA, which now incorporates TTI. "The algorithm is optimized for about 800 ft, and you get significant throughput degradation beyond that point."
At least two of the companies using polling strategies happen to do so within a basic physical network architecture that departs from the familiar hub and spoke model. Rooftop Communications and Nobell Communications (Austin, TX) both make products intended to operate within a mesh architecture reflecting the Internet multipoint-to-multipoint model where all individual nodes can choose among various pathways based on polling information. In this scheme each node of the network incorporates a router and the capability of transmitting to any adjacent node.
Equipment following the hub and spoke point-to-multipoint model must also incorporate at least rudimentary routing capabilities. Indeed, some of the wireless bridges on the market today are especially well-endowed with network functionality. The WaveSpan products, for instance, not only include routers with each bridge, but also distribute Web servers at each of the access points. The primary purpose of distributing intelligence throughout a bridge-connected network is to manage bandwidth effectively and make certain individual nodes are equitably served. Certain manufacturers already claim impressive progress in this area. "We can manage 256 simultaneous connections from a central point," says Steve Viegas, director of sales and marketing for OTC.
Nobell Communications' all-wireless ISP
Nobell Communications' all-wireless ISP Nobell Communications (Austin, TX) represents a phenomenon that scarcely existed a year ago when the firm turned on its first customer. The idea of using high-speed unlicensed bridges to build a public network was almost completely untried and to many, fanciful. But according to CEO Bert Johansen, the company is already well beyond the proof-of-concept stage. "We've demonstrated four nines reliability and we're looking to improve that. We've never lost a customer, and we've stolen a lot of subscribers from Roadrunner [a cable-modem service]. We plan to open markets all over the United States, and we're looking to sign up millions of customers eventually."
Johansen admits that some proprietary solutions are involved in operating his network, however. "We tried off-the-shelf equipment and ended up designing our own. We use a commercially available radio, but we include our own router at every node." Each such router incorporates extensive network-management capabilities and can transmit to any other node in the area depending upon network traffic patterns.
"What we're doing is different from what other wireless ISPs [Internet service providers] have attempted," Johansen continues, "or different from what DSL [digital-subscriber-line] providers have done, for that matter. Most access providers are cherry-picking customers. We're the only ones out there really serving both a residential and business market, and doing so reliably. We've got day traders using our network, and they cannot afford to have any downtime whatsoever. Our wireless solution works for them. Unlicensed networks can be very reliable, and we've proved that. Build it correctly and you don't need service-level agreements or guarantees. People will simply know it's reliable."