Mar 1, 2001 12:00 PM,

Outta Sight

Jeremy Kintzel

Various paths lead vendors toward true non-line-of-sight technology for fixed-wireless access to data.

A Piece of the Pie

First, there may be some confusion about the threshold that constitutes NLOS. Proulx defines NLOS as the ability to reach 90% to 95% of users within cell range, with transmission rates of 2Mb/s and above. Other providers such as Cisco discuss coverage percentages as viability thresholds. Troy Trenchard, Cisco Systems director of marketing for the Wireless Access Business Unit, argues that early adopters are finding they need to hit 60% to sustain profitability.

Most everyone working on NLOS is focused on enabling signal reflection or multipath. The concept, at least, is simple: The way to get a signal around obstructions is to bounce it around them, and engineer a CPE that can create one accurate, complete signal from all of the echoes.

At 2.5, 2.7(GHz), some of the energy comes through the walls, but a lot of it comes through the windows, said Ralph Muse, NextNet president & CEO. It comes from various directions, and it bounces around. Frequently, what we find is (the CPE) is not actually pointing toward the cell site. We're pointing 180 degrees away from it.

To maximize coverage, carriers are moving toward microcells.

You're not going to have the 30- to 35-mile (radius) cells you have today, said Reza Ahy, Aperto Networks president & CEO. You're going to have cells capable of going up to 10 to 12 miles, but more likely to be in the 4- to 5-mile range.

Focus on microcells allows carriers to begin at low capacity and grow with their customer bases, reducing the massive front-end outlays of capital that have doomed a raft of technology start-ups in 2000. There's a stepping function, said Cristian Parrino, Spike Broadband vice president of marketing, which is what we like to refer to as a stairway to heaven. The system can provide a capital efficiency far superior to wireline alternatives.

Within these microcells, most carriers will depend on sectorization to increase the capacity of limited spectrum. The coverage field of any given cell is divided horizontally, much like a pie, with the same frequencies used concurrently within each slice and by different cells.

At 3.5GHz, which is one of the fundamental microwave bands we deploy in, we can generate up to 24 sectors in any given cell, said to Tony Masters, Spike CTO.

The system depends on the concurrent use of frequencies between sectors and cells, or spectrum reuse. Basically, Proulx said, when carriers depends on one big, giant supercell, once they use up all of those channels in that specific slice of spectrum that they own, they're out of capacity. There's no scaleability.

In this environment, vendors and carriers must be vigilant against interference, in fact, not merely from outside sources but from their own signals. For Iospan, the answer is multiple-input, multiple-output (MIMO) technology, which uses multiple antennas at both the base station and CPE. MIMO allows Iospan's system to detect or recreate the single, discrete signal necessary for reliable transmission in a multipath environment. The system's key is in the development of new algorithms, which allow very aggressive interference canceling, according to Proulx. There has to be some mechanism for the cells and CPEs to control interference, and reliably track and relay each user's transmission. Maximum capacity requires a 1:1 frequency reuse, according to Muse, which means being able to use every frequency on every cell site. This allows a carrier, working with well-placed microcells, to serve a large metropolitan area with only four to six channels.

The orthogonal-frequency-division-multiplexing (OFDM) protocol is an important element in the development of fixed wireless. Masters believes the suitability of OFDM to multipath transmission, as well as its incredible capacity, will lead more carriers to embrace it. The OFDM signal carries information across multiple tones, which helps lower the error rate. Ideally, each signal will be 80% to 90% clean, which increases the CPE's ability to untangle and reassemble one clean signal. Masters also believes OFDM will be an indispensable component in mobile broadband wireless, although mobile applications for these technologies will require some FCC rule changes.

Spatial multiplexing and symmetric transmission are key features that equipment providers, including Iospan, are counting on to improve system productivity. Iospan's architecture allows the system to transmit two streams of data at the same time, over the same piece of spectrum, according to Proulx. The two different streams are then received and separated, effectively doubling spectral capacity. Symmetric transmission will allow carriers to establish downstream and upstream capacity. Carriers may establish, for example, gold, silver and bronze service levels, priced accordingly dependent on capacity and support.

Simple customer-installed CPEs also are a valued goal. In many ways, the model is the development of the pet satellite dish. Customer self-installs increase roll-out speed and keep carrier costs down by reducing or eliminating truck rolls.

I got DSL (working) on the fifth truck roll, NextNet's Muse said. I don't think there's a lot of money being made there.

NextNet is working toward an indoor, affordable CPE, which doesn't even have to be mounted. It just sits in the office with the user a self-contained box that encapsulates modem, antennas, processing.

It's like Prego, Muse said. It's in there.

Many equipment vendors are working to retail a sub-$500 CPE, although Cisco's Trenchard believes there always will be a place for the high-end, high-capacity entrant.

Finally, for fixed-wireless broadband to reach its full potential, regulation and spectrum scarcity will require the systems to be engineered to other frequencies. Although the MMDS range is attractive to early adopters, equipment and carriers eventually will need cells and CPEs compatible with the 3.5GHz space to move into Europe. As well, licensees of narrower spectrum ranges, such as 700MHz, may want to use them to offer fixed-wireless service, said Bernard Aboussouan, BeamReach marketing vice president. LMDS, despite its challenges, also has its users. These ranges (generally 24GHz and above) offer particular technical challenges, but may be attractive to smaller players and larger players looking to expand into areas where MMDS is occupied.

Sprint has most of the West Coast, but they have almost none (MMDS spectrum) along the East Coast, Ahy said.

The Standards Story

Eventually, some standardization of frequencies and architectures will be necessary in order for the technology to become widespread. Industry organizations such as the DSL Consortium and IEEE are working to sort out the chaos of technologies and strategies. The consortium is an organization of vendor companies that have an open forum for agreeing to technical specifications in the 2GHz to 4GHz range, or basically the MMDS and 3.5GHz European frequency. The intent is to come to a resolution next year, perhaps by the third quarter, according to Lamar Bishop of the Gartner Group, which is facilitating the DSL Consortium.

A large consideration is currently being given to an interface that has MIMO features, said Bishop.

The chosen standard, according to Bishop, will have to be functional in a multipath environment, although he acknowledges that the DSL Consortium will not have the final say.

None of these bodies (DSL Consortium, BWIF, OFDM Forum) can produce a standard, he said. The standard has to come from an organization like IEEE. Those other organizations can work toward a specification that they feel should be standardized. Then, they can submit it to the IEEE.

With several forums working on competing specifications, some hope their efforts will come together sooner rather than later.

We'd like to see a convergence of industry forums, said Cisco's Trenchard. There have been talks.

Similarly, within the OFDM Forum, competing technologies are working their way toward a standard. WiLAN, which co-founded the forum, has proposed wideband OFDM, which uses a wider frequency band than standard OFDM systems, an innovation the organization believes will result in higher speeds and less interference. The Forum's Web site states it will encourage the broad acceptance of a single compatible global OFDM standard on a worldwide basis, but that members will not be required to support the OFDM standards to the exclusion of other standards. Wi-LAN would seem to have a vested interest in the outcome, but the meetings have attracted the interest and participation of dozens of other firms.

Not everyone is thrilled about these processes.

The standards process is a compromise by a bunch of businesses that try to figure out how to write a standard that would be the least problem for all of them, said NextNet's Muse. It's automatically going to be a sub-optimal solution.

Muse believes that standards should be codified only after market forces choose a few winners. Aboussouan likewise doesn't want to see a standard based on a lowest common denominator, but he doesn't believe that's happening, especially because some discussions have turned toward more of an OFDM-based standard. He argues that spectral efficiency should be a priority, to ensure coverage for the greatest number of subscribers within any given market.

Whatever the final specification, it likely will complement, rather than eliminate the presence of wireline broadband in the marketplace. For Cisco, the strategy is to develop fixed wireless, while continuing to develop other options. Trenchard argues that there will be cases where wireline is still the most affordable choice for customers. Some users may pay more for higher bandwidth and quality-of-service guarantees. Some residential customers are happy with best-effort Internet connectivity and multi-Mb transmission. A viable business plan can come from predicting for this mixed use. For most customers and carriers, invisibility is the goal they don't care how or why their broadband works, Trenchard said.

The services matter; the technologies don't, he said.

Kintzel (kintzel@sunflower.com) is a freelance writer based in Lawrence, KS.