The next wave: Wireless/wireline hybrids?
Once disparate broadband access strategies are beginning to coalesce around a tightly integrated, hybrid wireline/wireless platform that could pose significant new challenges to the dominant wireline players in cable and telephony.
The idea of mixing fixed wireless and wireline components into a single network to accomplish wide and rapid penetration of the local marketplace started to take off this past spring when NextLink Communications Inc., one of the concerns cellular pioneer Craig McCaw is backing, suddenly bought out the LMDS (local multipoint distribution service) holdings of WNP Communications, the leading bidder in last year's auctions with 40 licenses in major markets nationwide. "That was a shot that was heard throughout the industry," says Tom McCabe, vice president of SpectraPoint Wireless LLC, the new firm created in the buyout of Bosch Telecom's LMDS unit by Motorola Corp. and Cisco Systems Inc. Now, with MCI WorldCom and Sprint acquiring MMDS (multichannel multipoint distribution system) spectrum in markets representing about 26 million households each, and AT&T quietly moving forward on various other wireless fronts, the message to the vendor community has been amplified many times over: hybrid wireless/wireline is the wave of the future.
"It's not just a matter of the radio link on the air interface providing bandwidth on demand," says Margaret Marino, vice president for network development at NextLink. "It's a question of, 'how can I make sure that I can bill for that service and provide for the OSS (operations support system) over that interface the same way I do over my wireline connections'."
NextLink is now the biggest player in LMDS, which operates at the spectrum tiers of 28 GHz and 31 GHz. Along with acquiring A block (1.15 GHz) licenses to 39 markets representing a population of 98 million people from WNP, NextLink has taken full control of 13 additional A and 29 B block (150 MHz) LMDS licenses by buying out Nextel Communications Inc.'s 50 percent interest in the two companies' joint wireless broadband venture, NextBand.
"The ability to combine fiber and wireless to reach small- and medium-sized businesses gives us an enormous advantage over companies that are either all-wireline or all-wireless," asserts NextLink spokesman Todd Wolfenbarger. In the wireline mode, he notes, the company can cost-justify extending fiber to buildings only about a quarter of a mile from the fiber rings it is installing throughout the metro regions of the country, which would mean it would have to lease T-1 facilities from telcos for deeper reach into these markets if it didn't have access via wireless broadband.
"Wireless gives us buildings two-and-a- half miles away from our rings and allows us to own the facilities," Wolfenbarger says. Eventually, he adds, the company will be able to extend the reach deeper, possibly moving fiber to the first points of wireless connection as the revenue streams build, and shifting the wireless transmitters further out to encompass ever more territory.
Wireless-wireline integration is also key to the MMDS strategies of Sprint and MCI, though their agendas differ from each other and from NextLink's.
"We intend to accomplish our local service goals using a patchwork of overlapping networks, including DSL (digital subscriber line), fiber, wireless, cable or other means," says Robert Finch, vice president of strategic development at MCI WorldCom. His company's move into MMDS assumes the platform will become an integral part of this local mesh, he says.
In MCI's case, use of the MMDS spectrum being acquired with the buyouts of Prime One and CAI Wireless and its affiliates will be focused on meeting the surging demand for high-speed data services in the small business and institutional markets, Finch says. "We also want to serve residential customers," he adds, but makes it clear the service will be designed to accommodate serious users with business-oriented applications.
"We hope to provide multimedia and video services," Finch says, but he indicates this is not a top priority.
Sprint, too, wants to be able to mix the use of DSL with MMDS to deliver ION (Integrated On-Demand Net-work) services in the local market, says Kevin Brauer, president for national integrated services at Sprint. In fact, he adds, the company wants to be able to integrate with its mobile PCS platform as well. "In bridging services like unified messaging, you lose the distinction between wireline and wireless," Brauer says.
But, in contrast to MCI, Sprint, with its acquisition of People's Choice TV and American Telecasting, will also design services, including TV entertainment, to accommodate the residential market, Brauer says.
America Online is making clear it believes the ability of carriers to integrate across multiple access platforms is vital to its plans to deliver broadband services ubiquitously across the country. "As we look at broadband, we're looking at a tapestry of network platforms," says George Vradenburg, senior vice president at AOL. Even if AOL wins the battle for open access over cable, it will still need the other platforms, he adds.
Many pieces vital to such visions are coming together, starting with advances on the wireless side that provide for the provisioning flexibility and quality-of-service options that are common to wireline platforms. Moreover, some entities are already focusing on the integration issue with new products that employ IP (Internet Protocol) and other formats over ATM (asynchronous transfer mode) as the primary means of tying transmission elements together at the local network edge points.
For example, Motorola Corp. and Cisco Systems Inc. have agreed to pool resources in an effort to create an IP-based, industry-wide framework for integration of the wireline and wireless domains. The two companies have pledged to spend $1 billion on the process, which they say could take as long as four years, though they're already beginning to deliver initial product lines built to suggested specifications of the framework.
"The reason we're moving with Motorola on SpectraPoint is we want to get the best of breed RF (radio frequency), fixed wireless broadband integrated into that architecture," says Skip Stritter, director of market development for broadband wireless at Cisco.
SpectraPoint has a strong headstart in the direction of facilitating multi-platform integration via IP and ATM, having been working with Cisco for most of the past year to integrate router switches and other IP components into the LMDS access system. "A lot of thought has gone into the integration of our system with Cisco's products," says SpectraPoint's McCabe.
The big performance gains in LMDS and other spectrum applications of point-to-multipoint wireless broadband start with greater flexibility in the use of bandwidth than was possible heretofore. For example, McCabe says, SpectraPoint will introduce product next year that supports dynamic changes in modulation from the more robust, low bits-per-Hertz levels to the noise-sensitive, high-capacity levels as weather conditions change or customers shift the tradeoffs they want to make between bandwidth efficiency and quality of service.
"The radios we're shipping with our 2200 system (slated for commercial rollout in the fourth quarter) support QPSK (quadrature phase shift key) and 8, 16 and 32 QAM (quadrature amplitude modulation)," McCabe says. "By mid-2000, we'll have software that supports dynamic shifts from one modulation level to the next." One of the innovations vendors are applying to LMDS and other fixed wireless products is the ability to transport everything in ATM while using time division multiple access multiplexing technology to dynamically alter the amount of bandwidth devoted to any one user's needs, McCabe says. This way, all the users on a single RF LMDS channel, now at up to 45 megabits per second and moving to 155 Mbps next year on SpectraPoint's system, can pay for services on an as-needed basis, allowing service providers to more efficiently allocate bandwidth that's not in use from one moment to the next.
Another glimpse of what is to come can be seen in the plans of Ensemble Communications Inc., a San Diego-based startup that demonstrated the first iteration of its "adaptive IP" platform at Supercomm in June. As described by Ensemble marketing Vice President Carlton O'Neal, the system uses a number of innovations to greatly simplify the end-user connection process while allowing service providers to deliver a given channel of frequency on a shared-use basis to multiple buildings.
"With current generation systems, you have to engineer every link to the customer individually and allocate bandwidth on a fixed basis to every link," O'Neal says. "With our system, customers will be able to take our CPE (customer premises equipment), hook it up and turn it on, just like people do today with cellular and PCS systems."
Another leading LMDS supplier, Newbridge Networks Inc., is promoting tight integration of multiple delivery platforms via TDMA/ATM interface cards, which employ compact design capabilities obtained through the firm's recent agreement to acquire Stanford Telecommuni-cations Inc. The Newbridge MainStreet Xpress ATM switch is designed to aggregate LMDS, MMDS, DSL, and eventually, cable, each with its own interface card, at a local service node which can also serve as the wireless base station, says Bernard Herscovich, vice president for broadband wireless at Newbridge.
"Our vision is to support a multiple-service network," Herscovich says. "Basically, we've made the ATM switch a base station by designing an interface card that directly integrates ATM functionality into the RF signal, which turned out to be a very difficult thing to do."
At the same time, the company is developing a voice-over-DSL interface in conjunction with the DSL card that slides into the same ATM switching module that is used for the wireless components, thereby providing for the complete integration of voice and data services over various wireline and wireless platforms at the base station/switch, Hersovich says. This new voice capability for DSL will use the GR-303 switch interface that allows the call provisioning and feature functionalities of central office switches to be supplied to the packet voice streams in the DSL connection, which is the approach to voice delivery being taken by most players in the DSL arena.
Where MMDS is concerned, operators are just now moving to the digital TV delivery capabilities that will allow them to go from about 30 channels of analog service to 100 or more digital channels. And they are also just introducing two-way, high-speed data capabilities, widescale deployment of which depends on what FCC officials at press time say is the imminent issuance of orders in the reconsideration process attending last year's grant of two-way authority to the MMDS sector. At the cutting-edge on both the video and data fronts is one of Sprint's targeted acquisitions, PCTV, which now has 160 channels of video along with a 60 megabit-per-second downstream/10 Mbps upstream data service underway in Phoenix, according to PCTV CEO Matthew Oristano. The two-way data capacity "can be repeated ad infinitum with various forms of spectrum multiplexing," he says.
Expanding that data carrying capacity is a top priority for both MCI and Sprint, starting with use of multiple transmitters and sectorization of those transmitters, in contrast to today's deployments where one transmitter, as in Phoenix, serves an entire metropolitan area. Sectorization means that each area served by a specific transmitter is divided into segments served by separate beams. In some vendor iterations, all of the sector beams operate over the same 200 MHz of frequency through use of some means like reverse polarization to prevent interference at the sector edges, while, in other design approaches, adjacent beams use different frequencies in an alternating pattern that allows only 100 MHz to be used over any one beam. This latter approach, used, for example, by Spike Technologies, permits a much higher degree of sectorization, where, in Spike's case, a dozen sectors divide the territory into small slices of users contending for the 100 MHz of available bandwidth within each slice.
Ironically, MMDS, with a signal reach of 35 miles, could readily serve a metropolitan region with a minimum number of transmitters, if it had the amount of bandwidth made available at the much higher frequencies set aside for LMDS, where licensees must cellularize to accommodate a signal reach of only three miles or so. The spectrum allocation differences result from the fact that there are many competing applications in the low-frequency tiers where MMDS resides, whereas it was not until fairly recently that it became technically feasible to use the tiers occupied by LMDS for point-to-multipoint applications.
There's considerable confusion about how these and other wireless tiers can be scaled over multiple cells and what the effective comparative bandwidth-to-user ratios are from one tier to the other. Given the great disparities among them, which also include the 24 GHz category where about 400 MHz is available; the 38 GHz tier, where several 100 MHz blocks might be amassed in any given marketplace; and the much more bandwidth-restricted categories in cellular at 800 MHz and PCS at 1.9 GHz; the players are always looking for ways to make their chosen tier look good against the others, which has produced some interesting claims.
For example, some experts are claiming the effective available bandwidth using MMDS technology at 200 MHz per territory is on a par with the bandwidth efficiency of LMDS at 1.15 GHz per license. But such claims often demonstrate a lack of understanding of all the factors that must be considered in weighing relative efficiencies among various wireless broadband spectrum tiers, where the propagation characteristics deteriorate as the frequency level goes up, notes Bob Foster, chairman and founder of Wavtrace Inc., a supplier of wireless broadband access systems at the millimeter wave frequencies at and above 28 GHz.
"What you have to realize is that, when you go to multiple cells, the limiting factor in terms of frequency reuse becomes the signal-to-interference ratio, and that's directly related to the modulation level you use, independent of what frequencies you're operating at," Foster says.
In other words, the sensitivity to interference at 64 QAM (quadrature amplitude modulation) is the same whether the propagation frequency is 2.5 GHz or 28 GHz. This means that, with its much greater signal reach, cell-to-cell interference for MMDS comes into play at far greater spacing of cells than is the case for LMDS, so that any effort to chop a territory into a lot of MMDS cells in order to reuse the limited frequency to the point that it begins to even out the bandwidth advantage of LMDS is going to be hampered by the S/I problem. "With MMDS, you're not going to be able to get away with 64 QAM if you're counting on a reuse strategy to compensate for the limited bandwidth," Foster says. Thus, with the loss of the higher bit rate per Hertz of 64 versus 16 QAM, for example, which translates to about five bits per frequency cycle versus two, the 200 MHz of frequency loses a lot of its aggregate throughput in exchange for a move to multiple cells.
Foster says the best way to look at the two types of platforms is to view LMDS as ideally suited for high-capacity service requirements in urban areas, while MMDS is better for lower capacity requirements in less dense residential areas. "The bottom line is, LMDS has the bandwidth advantage, but MMDS has the propagation advantage, and it's best to design systems accordingly," he says.
With only 200 MHz to work with, Sprint CTO Marty Kaplan says cellularization is a likely step in his company's efforts to maximize bandwidth. "I doubt that we'll put just one stick up in an area," Kaplan says, adding that the company is also looking at use of "intelligent waveguide" technology as a means of expanding bandwidth efficiency.
MCI is thinking along the same lines, Finch says."We will break the serving areas up into smaller cell sizes," he says, noting that this would add substantial costs to the approximately $1 billion MCI is spending to acquire the wireless cable properties.
Eventually, both companies intend to use the MMDS spectrum for delivering multiple line packet voice services, though it remains to be seen how soon such services will be possible. Brauer says Sprint will focus on delivering high-speed data next year, postponing to some undetermined time in the future the rollout of voice over MMDS.
While Oristano says that PCTV has already tested use of an IP voice phone over the firm's two-way data channels and found the quality to be on par with circuit switched service, Finch cautions that it will take operators a long time to achieve the back office and operational integration essential to supporting voice services.
In fact, Finch says, the technical challenges associated with use of wireless broadband have made the topic "one of the most difficult questions for our senior management to answer." But, he adds, with demand for high-speed data connections surging and the advantages MCI sees in owning broadband facilities rather than leasing telco lines, as it must do with use of DSL technology, the company concluded it had little choice but to make a move into fixed wireless services once it became clear that two-way services were technically feasible.
Now MCI, which, like Sprint, will have coverage reaching some 26 million households and businesses through the announced acquisitions, is considering further moves in this direction, Finch says. "My guess is that you haven't heard the end of it yet," he says.