|Fixed wireless broadband is at the stage that microprocessor led PCs were 20 years ago. The dynamics of FWBB is that a myriad of technical innovation is taking place that is rapidly pushing through was has been perceived to be the barriers to mass-market acceptance an profitability. A critical factor in this new revolution is similarly through the use large-scale integrated circuits. The technical innovations when combined with the “Moore’s Law” dynamics of increased functionality with reduced cost is fast approaching a business case that is far superior to wired, fiber or any other physical plant termination to the home and small business customers.|
First let’s look at the perceived barriers to FWBB, then the types of new technology that will be used to get around them and finally at some companies who are now or soon will be ready to sell commercially viable systems.
The first perceived barrier has been the cost of deploying FBBW MMDS to the home and small business customer. I say ‘has been the cost’ because this is really no longer the case. The first large scale U.S. deployments were done by Sprint in 15 cities. Sprint chose cities were it was relatively easy to deploy mega cell architecture MMDS systems from Hybrid Networks. Nonetheless, Sprint says that sales exceeded expectations, capturing up to 25% of broadband subscribers in a relatively short period of time. This is about 3 times higher penetration rate than expected. The cost per customer premises equipment is about $600. Hybrid Networks system is LOS and requires at least one truck roll for professional antennae and cable installation and PC connection. The average cost of a truck roll is about $200, bringing the total installed cost per CPE to around $800. Allocation of head-end equipment and service costs adds another $50-$200 depending on saturation levels.
The second major obstacle to ramping up of FBBW as a competitive alternative to DSL and cable is the lack of a current mass market and low volumes of production. This leads to a current dilemma, or “chicken and the egg” situation: Costs for FWBB are still relatively high when all costs are added togetherto be a mass-market consumer product, (although it may be argued that they are already competitive with the high costs of real world deployment experience of DSL and cable). The underlying component technologies used in FWBB systems are highly scalable using ICs and automated manufacturing methods. However, the constituent technologies are changing rapidly such that tolling up for each product change at low volumes of production proves inefficient and highly expensive.
This chicken and egg scenario has been particularly true for the new Orthogonal Frequency Domain (OFDM) based systems that are just entering the MMDS market. One of the primary competitive arguments against this new technology is that it requires a powerful DSP to perform Fast Fourier Transform calculations and signal discrimination processing. I guess you could have said the same things about the PC revolution 20 years ago – you just can’t pack enough power into a device cheaply enough to make it functional or commercially attractive.
Another barrier to FBBW is the need to have line of sight for mega-cell DSSS type 1st generation systems. The advantages of DSSS systems are large coverage areas from a single tower location – up to 35 miles and therefore lower initial head-end costs in favorable geographic regions. The major disadvantage is that antennas must be mounted on the outside of the building and pointed directly in line of sight to the tower or through the use of complex bridging of reflected signals. This is similar to DSS system installation but often more difficult given the relatively shallow angle of the signal. The “reach” of DSSS systems is touted as high as 70% but practically is much closer to 50% coverage.
A final perceived limitation for FWBB is the limitation of available spectrum. This is commonly thought to limit the practical throughput because spectrum must be shared within a cell or segment of a cell. For instance, if 6 Mbs is available it is thought that it must be split up among the hundreds or thousands of users to unacceptable throughput levels.
FWBB is on the threshold of a revolution that culminates from the intersection of complimentary technological developments combined with powerful low cost DSP and communications cores for VLSI chip solutions. The business case has been successfully proven with more expensive and application constrained DSSS based systems while OFDM systems have recently exceeded expectation in field trials and early deployments. The new systems remove all of the barriers to mass-market success:
OFDM is a non line of sight technology. Systems have been recently introduced by Breezecom and are entering trials at several other companies that include a user installable CPE unit. Some of these systems have antenna that are installed inside the building, on a shelf or table next to the PC. Little or no aiming of the antenna is required to achieve 1-3 Mbs connections. This eliminates truck rolls and thus reduces the cost of installation dramatically compared with cable or DSL systems. Even the new user installable DSL and cable systems require line conditioning and head-end provisioning. Breezecom, NextNet, IoSpan, Wi-Lan, Cisco, Malibu Network, Raze Technologies, VYYO have OFDM systems entering the market now or over the next 18 months.
The cost issue is being addressed through the use of recently available DSP cores that are both inexpensive and cheap. Along with custom signal processing circuits an firmware, ICs can now be fashioned that reduce the cost of implementing OFDM technology. NextNet, IoSpan, Raze have developed their own custom VLSI DSP chip set solutions. Third party OFDM chip sets are also being developed with first working silicon already reported. Another contributor to reduced costs is plate type antenna arrays. OFDM systems will use these low cost antenna arrays to provide multi-path transceivers that can be built integral with the OFDM modem. This industry appears to be able to develop along the lines of “Moore’s Law”; a halving of cost and a doubling of capability every 18 months. NextNet proposes that the cost of the CPE is currently about $500 and will sale down to as low as $100 in five years. Other companies claim a price trend of about $250 per CPE within five years. In any case, the proposed pricing seems highly competitive.
The field of OFDM competitors will soon mushroom as privately held start-ups enter the market over the next few months.
Sprint and Worldcom have confirmed the reach of the new NLOS systems in field trials as being as high as 90% currently. Both companies have delayed further deployments of FBBW so that they can evaluate the offerings from several of the companies entering the field. Early statements from these companies indicate that they will to forward with deployments using OFDM technology although it is not at all clear which company or companies they will go with. Complicating the decisions are several enhancements to the basic OFDM technology that promises to increase capacity of the systems dramatically or offer greater effective throughput and management of IP broadband traffic.
The final argument against FWBB is that “there is only so much bandwidth available and it isn‘t hardly enough to carry all the traffic the Internet needs to grow in the future”. What could be said about 1st generation FWBB systems cannot be said about enhanced 2nd generation OFDM technologies or even enhanced first generation systems. The basic concept behind various technologies that increase capacity of a given spectrum in a given geography is “spectrum or frequency re-use”. Unlike linear fixed wire systems, the airwaves are relatively boundless in terms of spatial differentiation. For instance, IoSpan is developing a version of OFDM that uses multi-path transmitter towers and CPE receivers. Put simply, the CPEs are able to gather and discriminate signals using the same spectrum but originating from various directions. Because the primary data signal is comprised of discrete data packets, these can be split up and routed to antenna that are spaced one wavelength or more apart and transmitted simultaneously over the same frequency. These signals can even be split up centrally and sent to transmission towers located in different parts of the cell. The CPE receives these signals with the multi-antenna plate array and the DSP extracts the data and recombines the component data packets. This provides a linear multiplication of apparent bandwidth. Other methods, such as SpeedCom’s “packet-hop” and Wi-lan’s Vine technologies can be used to multiply the effective bandwidth of both 1st generation and OFDM systems. These similar methods use each CPE in a way that they become and transmitter and IP router for each other CPE in the local network. Usually two or more user locations tie into the Internet through connection to the transmission tower or through a fiber optic or T1-T3 fixed connection. In the later case there is no need for a head-end wireless system.
I believe that FWBB is finally at the stage of development to explode onto the scene during the next two years. The climate for capital expenditure will probably continue to look bleak for at least six months. Meanwhile, major carriers are diligently evaluating the new systems and readying their plans for deployments expected to kick off by the 1st quarter of next year.
Here are some links to companies, articles, etc.: