The broadband wireless market is experiencing significant growth and could be worth an estimated $20 billion world-wide over the next few years. Vincent Tobkin, director and co-leader of Bain's telecoms and technology practice and David Sanderson, partner and member of this practice, comment on developments.

Now that AT&T and Microsoft have staked their claims to CATV networks in the United States and portions of Europe and many telcos are gearing up for DSL deployment, a big question is: what major networks will be built and subsequently acquired? Or is the network building business at an end-point (at least until the planned broadband satellite networks, such as Teledesic and SkyBridge, are successfully deployed)? We foresee another major network opportunity on the horizon - terrestrial wireless networks designed primarily for broadband services.

The first major initiatives in this area were launched in the United States by Winstar, Teligent and Nextlink (and others), all offering wireless broadband services at different wireless frequencies. Similar spectrum licences are now being awarded elsewhere in the world. Wireless broadband IP services are now being offered and should experience substantial growth. Wireless technology offers the potential for true bandwidth on demand, lower initial deployment cost, provisioning flexibility and the ability to serve customers where CATV, DSL, or fibre services are technically infeasible, legally precluded, prohibitively expensive or simply not yet available. Current estimates for the wireless broadband market suggest that this market could reach $20 billion world-wide over the next few years.

This new technology opportunity is being pursued primarily by entrepreneurial teams with funding from venture capital, equipment vendors and other capital market sources. This is further highlighted by the fact that most new spectrum awards are going to new companies, rather than existing network operators. Ultimately, as in the CATV and cellular industry segments, we foresee eventual industry consolidation with substantial profits accruing to the network pioneers - another potential gold rush.

Background

In the US, broadband wireless has been authorized within a frequency spectrum of 2.4, 5.8, 18, 24, 28 and 38 GHz. At the time of the first grants of 38 GHz spectrum in the US in the 1980s, these grants were free or very low cost - as the commercial promise of these previously military frequencies was unclear. Now market values are known. While the low end of these frequencies (2.4 and 5.8 GHz) in the US are available for free, unlicensed shared use, the recent LMDS spectrum auction (28GHz) generated close to $600 million of fees for the licensing of the spectrum. While much of the spectrum in the US has been allocated, internationally these same frequencies, in many regions, are becoming available for pioneer licences.

In some ways it is like the early days of cellular, with regulators not fully informed, entrepreneurs obtaining licences cheaply, equipment just starting down the volume scale curves, and early customers enjoying uncrowded spectrum. Few of today's participants understand the full potential of this opportunity.

Enablers

This opportunity is enabled by favourable developments in wireless components, radio technology and relevant manufacturing technologies. Moreover, the potential for terrestrial broadband wireless has been enhanced by abundant spectrum licensing, dramatic increases in broadband demand and problems with the deployment of other broadband technologies. As a result, there is mounting capital market interest in terrestrial broadband wireless. The beneficial impact of all these developments is neither clear-cut, nor yet widely understood. But their combined impact should enable a substantial number of additional business successes over the next three-five years.

These developments begin with advances in the semi-conductor technologies used at these frequencies. Gallium Arsenide (GaAs) semi-conductors are used at all frequencies assigned to broadband wireless communications. GaAs manufacturing is now shifting from 4 to 6 inch wafers, as volumes increase for cellular and optical semi-conductor components. This increase in the wafer size used in manufacturing can produce as much as a 56% reduction in component costs.

In addition Silicon Germanium (SiGe) semi-conductors are now being developed for wireless communications and are being introduced commercially by several companies, including IBM. SiGe components are usable at frequencies under 5-6 Ghz and have the potential to lower costs 60% below those of GaAs. Finally, work by TRW and others in Indium Phosphide (InP) semi-conductors holds promise for commercial devices for broadband frequencies well above those now being licensed.

Several radio technologies originally developed for military applications are being applied to broadband wireless. The first, of course, is CDMA technology already in general use for cellular. Next is the use of phased array antennas to provide automatic and dynamic aiming of antennas, without mechanically moving parts. Another significant antenna technology is sectoring. This process involves the use of several antennas on the same tower (with pie-slice-shaped, rather than circular, radiation patterns), to facilitate simultaneous re-use of the same spectrum in multiple quadrants around the antenna, yielding a 2-16x increase in the capacity of each radio tower. Wide bandwidth amplifiers and variable bandwidth modulators permit rapid changes in the capacity/wireless link, providing bandwidth on-demand and dynamic adjusting for weather conditions. CDMA also provides a high level of security for high bandwidth transmissions, which can be further enhanced with software-based encryption.

Manufacturing technologies are being developed to lower the cost of commercial components used for these systems. First is the development of monolithic microwave integrated circuits (MMICs) to standardize the previously highly customized process of making high frequency microwave transmitters and receivers. This technology is initially being deployed in 1.9 GHz cellular phones and leading edge optical transceivers. Also significant is the developing use of plastics and other-low-cost assemblies for the construction of high-precision antenna surfaces and outdoor housing.

This scientific progress is being matched by regulatory support for terrestrial broadband wireless. Specific assignments vary by country, but the overall trend is very favourable. Specific examples include the following spectrum assignments. Spectrum has been assigned for both Local Multipoint Distribution Service (LMDS) and Multi-channel Multipoint Distribution Service (MMDS) in the US over the past 10 years at frequencies from 2 to 38 Ghz. Similar spectrum assignments are now under way in Spain and France. Interest in these spectrum assignments is widespread in other European countries (using MVDS), in Japan, South America and Canada. There is also active consideration of re-tasking wireless CATV spectrum at 2-4 GHz for use by Internet protocol (IP) broadband services. Recent acquisition activity by both Sprint and MCI has highlighted the potential for this wireless CATV spectrum. In the interest of accelerating service deployment, many of these spectrum assignments have been issued for free, or at prices well below their likely future value.

The authorization of unlicensed broadband spectrum at 2.4 and 5.8 Ghz will enable the deployment of campus broadband wireless networks. Entrepreneurs are also planning to build community or nation-wide networks using this unlicensed spectrum, in much the way Metricom has used unlicensed spectrum to build its Ricochet business in the US.

On the demand side, major demand increases for high bandwidth Internet access, growth in IP telephony and demand for high-speed connections by corporate networks for work-at-home applications are driving the need for pervasive broadband network access. The drivers of these trends are well understood. However, there are aspects of this demand, particularly with those users best able to pay, that are particularly well suited to broadband wireless access. The most obvious is the use of wireless data connections at work locations and emergency sites, where other wired access means are unavailable or are no longer working reliably.

Wireless broadband is most attractive where there is no available fibre, DSL cannot be provisioned, and cable plant does not exist to provide broadband access. It is also attractive for fibre operators that need access to a few, low-density valuable users, without having to work through central office facilities or over-build their own network to those few locations. This is particularly needed in countries where the PTT copper plant is not usable by third parties for upgrading to DSL access service.

The bandwidth-on-demand features of wireless broadband are expected to create demand among premium users. These features permit the purchase of a fixed bandwidth connection on a typical flat-rate plan, both with and without quality of service guarantees. However, should a user temporarily need higher speed guaranteed access, say to transmit large files or for video conferencing, unused bandwidth can be made quickly available for a premium price, up to the full bandwidth of the transmitter. If sufficient unused bandwidth is unavailable, other users not paying for premium access can be switched to lower bandwidths temporarily. Similar functionality can be implemented in wireline broadband networks, such as ION by Sprint, but doing so has been complicated and more expensive than wireless bandwidth allocation.

Lastly terrestrial wireless broadband can serve as the access technology of last resort by operators of fibre, DSL or CATV access networks to reach customers, who are not yet served by these other access mechanisms and who will probably not be served in this way for the foreseeable future. Estimates of wireline connections unsuitable for DSL services range as high as 60% in many US localities.

These situations generally occur where Pairgain-like line sharing has been used, where the network is inferior in terms of design or maintenance, or when distance from the central office is beyond the range served by current DSL technology. These conditions are also particularly prevalent in developing countries.

Several hurdles remain. These include a specific line-of-sight requirement and a 3-5 kilometre limitation in transmission range (depending upon the frequency used). Many users in office buildings and apartments will also need to negotiate roof rights, much as they have done for satellite TV services. Terrestrial broadband wireless will require many more antenna locations than are now used for cellular coverage within the same region. However, the required antennas are rather compact compared to cellular antennas and the use of flat-panel phased array antenna makes them far less noticeable. The development of antenna real estate companies, formed by the acquisition of existing antenna sites, should also facilitate wireless broadband radios sharing cellular sites.

New services

Wireless broadband services are being designed to offer users many benefits. The initial uses of wireless broadband spectrum have involved the offering of traditional T1/E1, T3/E3, or partial T1/E1 connections to central offices or other carriers' points of presence (POPs). Some corporate customers also use this service for private network connections between facilities. These connections are unaffected by wireline network failure and offer complementary back-up capacity for mission-critical systems. They also provide high-speed links to buildings on the periphery of urban areas that are not yet connected to fibre.

The next service to be offered will be 1-45 Mbt/s IP connections for Internet access and voice-over-IP back to Internet Service Provider (ISP) points of presence. For many office buildings and homes, this wireless link will be the quickest to install and also the most cost effective, particularly in less densely-populated suburban areas. This service will be competitive in those markets that are still monopolized by PTTs. In more open markets, it will face DSL competition, where the copper plant is usable without expensive build-out.

We expect terrestrial broadband wireless services to start offering levels of service based upon guaranteed bandwidth, with the option of increasing bandwidth upon demand for a premium fee. Operators may also start to prioritize data within an IP data stream, utilizing emerging Internet standards. One such combined data service would support IP-based servers/key systems of the type being offered by Vertical Networks.

We also envision wireless operators offering a version of bandwidth-on-demand as a back-up for DSL and CATV IP services, for broadband applications and networks requiring fail-safe, fault-tolerant operations. Depending upon the cost trends for customer premise equipment video servers and IP streaming technologies, wireless broadband services could also be used to provide local stations to customers using satellites for their primary TV service.

Competition

Existing wireline operators, both telco and CATV, have not vigorously pursued broadband wireless licences or services using unlicensed bands. A few telcos have looked to wireless broadband for use in providing telephone services, where wireline facilities are either delayed or not economic. US WEST has displayed the greatest interest in and leadership among US local exchange carriers (as it has also done in DSL services and technology).

But most existing carriers have not yet aggressively pursued wireless broadband opportunities. Moreover, they are generally disadvantaged in pursuing new licences, given some of the regulatory and auction guidelines. Also in many cases, disputes within carriers' businesses from existing broadband businesses (@Home-like business in CATV companies or DSL business units in telcos) have prevented entry into broadband wireless businesses.

Existing cellular wireless operators have not done much better. They are occupied by voice-line competition and find both the data nature of these new businesses and their non-mobile attributes too challenging. Most cellular operators are busy trying to initiate narrow-bandwidth IP services or are waiting for UMTS /3G cellular systems, to the exclusion of terrestrial broadband wireless.

Wireless broadband pioneers have generally been entrepreneurs with a regulatory, technical or business vision who rely upon their other existing businesses or venture capitalists for start-up financing. A few recent start-ups have also relied heavily on vendor financing in excess of their equipment costs.

Conclusion

Terrestrial broadband wireless services have a valuable role in complementing broadband services provided directly by fibre, DSL or CATV. Their unique service features, start-up economics, and cost/technology drivers make them a credible competitor for broadband services. Their dual role in providing primary or back-up telecoms access and unique broadband IP services afford them both near-term revenue and long-term promise.

Recognition of the business opportunity that these services represent has so far been limited to entrepreneurs, some of whom have been handsomely rewarded. Despite the stock market success of the earliest of these services, regulators in many countries are still issuing pioneer licences at significant discounts.

The challenge for existing carriers that may eventually want to participate in this business is substantial. They need to decide when to enter and how to nurture this line of business, even though it competes with their other lines of business. We believe that the teams now starting new terrestrial broadband wireless businesses will achieve stock market success and benefit from many liquidity options.