Broadband wireless provides one of many physical-layer options for the operator of a public service network. Furthermore, the different wireless networking technologies themselves exhibit widely varying capabilities for fulfilling the needs and expectations of various customers and enterprises. More than previous wireless standards, 802.16 addresses a multitude of needs for the users of broadband access services, but it is not invariably the best solution for delivering broadband services in every market.

Broadband Fixed Wireless: The Competitive Context

This section strives to answer the question, when is 802.16-based equipment appropriate? It is the first and most crucial question network operators have to ask themselves when considering the broadband wireless option.

At the risk of stating the obvious, I will enumerate the rival competitive access technologies for broadband before discussing their competitive positioning vis-à-vis wireless. In the metropolitan space, wireless broadband competes with the following:

• T1 and E1 data services over legacy copper where aggregations of ordinary twisted pairs form the physical medium of propagation.
• Data services based on Synchronous Optical Network (SONET) and Synchronous Digital Hierarchy (SDH), running over fiber linkages.
• Frame relay services running over fiber or T1/E1.
• Ethernet data services running over active fiber-optic linkages.
• Ethernet data services running over passive optical networks (PONs).
• IP data services over active fiber.
• Asynchronous Transfer Mode (ATM) services over active fiber.
• ATM over passive fiber.
• Wavelength services over active fiber.
• Ethernet services over hybrid fiber coax.
• Digital subscriber line (DSL); most existing DSL networks contain components based on ATM and Internet Protocol (IP) as well as Ethernet.
• Powerline carriers where AC transmission lines carry packet data.
• Broadband satellite.
• Free-air or free-space optics where laser beams transmit information over an airlink, dispensing with fiber.
• 2.5-generation (2.5G) and 3G mobile data services, including HSDPA.
• Integrated Services Digital Network (ISDN), a nearly obsolete type of medium-speed data service utilizing double pairs of ordinary copper phone lines to transmit data.
• Storage-area networks represent a special case; most use specialized data protocols of which Fibre Channel, which runs over optical fiber, is the most popular.

Of these rivals, several are currently entirely inconsequential. Broadband as opposed to medium-speed satellite services scarcely exists as yet, and powerline carrier services and PONs are scarce as well, though both appear to be gathering impetus. Pure IP and Ethernet metro services over fiber are growing in acceptance, but they are not well established, and ISDN has almost disappeared in the United States, though it lingers abroad. Finally, free-space optics have achieved very little market penetration and do not appear to be poised for rapid growth. Other services mentioned previously-such as wavelength, 3G mobile, direct ATM services over active fiber, and metro Ethernet over active fiber-have some presence in the market but are spottily available and limited in their penetration thus far.

In this context, broadband wireless does not look nearly as bad as detractors would have it. If you consider the whole array of competing access technologies, broadband wireless has achieved more success than most. Still, it faces formidable competitors among the more established technologies, and these are T1/E1 (including fractional and multiple T1/E1), frame relay, DSL, and cable data.

Among the incumbent technologies, cable data and DSL are the leading technologies for residential services, and business-class DSL, T1/E1, and frame relay are the dominant service offerings for small- and medium-sized businesses. The largest enterprises that require large data transfers tend to prefer higher-speed optical services using both packet and circuit protocols.

Circuit-Based Access Technologies

Within the enterprise data service market, T1, fractional T1 (E1 elsewhere in the world), and business-class DSL are the most utilized service offerings, along with frame relay, which is chiefly used to link remote offices and occupies a special niche.

T1 is usually delivered over copper pairs and is characterized by high reliability and availability, reasonable throughputs, 1.5 megabits per second (Mbps), and inherent quality of service. Its limitations are equally significant. T1s cannot burst to higher speeds to meet momentary needs for higher throughputs, and they are difficult to aggregate if the user wants consistently higher throughput speed. T1s are also difficult and expensive to provision, and provisioning times are commonly measured in weeks. Finally, T1 speeds are a poor match for 10 base T Ethernet, and attempts to extend an enterprise Ethernet over a T1 link will noticeably degrade network performance.

Because it is circuit based and reserves bandwidth for each session, T1 offers extremely consistent performance regardless of network loading. Maximum throughput speeds are maintained at all times, and latency, jitter, and error rates are well controlled. Were the bandwidth greater, T1s would be ideal for high-fidelity multimedia, but, as is, 1.5Mbps is marginal in that regard.

T1/E1 is legacy access technology. The basic standards were developed in the 1960s, and the SONET and SDH optical equipment supporting massive T1/E1 deployments dates back 20 years. In terms of performance level, T1/E1 is essentially fixed, a fact that will put it at an increasing disadvantage to newer technologies, including broadband wireless. Also, the infrastructure for these circuit-based access networks is expensive to build, but, since most of it has already been constructed, it is by now fully amortized.

I do not expect a lot of new copper to be built except in developing countries, and so the last-mile access for T1/E1 must be considered a fixed asset at this time. But, somewhat surprisingly, the sales of SONET and SDH equipment for the metro core have been increasing rapidly through the late 1990s and the opening years of this century, and they are not expected to peak until 2007. Therefore, SONET and the T1 service offerings it supports will be around for a long time.

Prices in the past for T1s were more than $1,000 per month, but they have dropped somewhat, and they are now about $300 to $400 in the United States, though prices vary by region and by individual metropolitan market. Compared to newer access technologies, T1 does not appear to represent a bargain, but it is all that is available in many locales. Moreover, the incumbent carriers that provision most T1 connections are in no hurry to see it supplanted because it has become an extremely lucrative cash cow.

Because of the apparently disadvantageous pricing, T1 services may appear to be vulnerable to competition, but thus far they have held their own in the marketplace. Ethernet and IP services, whether wireless or wireline, will probably supplant circuit-based T1 in time, but as long as the incumbent telcos enjoy a near monopoly in the local marketplace and are prepared to ward off competition by extremely aggressive pricing and denial of central office facilities to competitors, the T1 business will survive. I suspect that T1 connections will still account for a considerable percentage of all business data links at the end of this decade.