You're likely to hear the term broadband just about anywhere. From satellite broadband to fixed wireless broadband to high-definition television (HDTV), broadband often means different things to different people.

While broadband can have different meanings, the bottom line is that broadband basically refers to a lot of bandwidth, regardless of application or delivery system. There is no defined minimum or maximum bandwidth and there is no specific frequency at which broadband exists exclusively.

Much of what is referred to in the consumer industry as broadband is simply good guerrilla marketing — using the term to define a product.

Because broadband refers to a wide bandwidth, it in turn implies the ability to run bandwidth-hogging applications such as real-time multimedia and high frame-rate video.

While theoretically this is true, broad bandwidth is subjective. To some, broadband can be a few hundred kilohertz, while others consider broadband gigahertz. There is a lot more involved in sending data over a fairly wide bandwidth than the marketers want the consumer to know.

Issues such as compression, two-way transmission capability and power, just to name a few, are all relative factors in the quality and loading of broadband systems. And, distribution networks have to be worked out, as well as interference and rights-of-way.

However, broadband is pressing ahead, even if the pace is a bit slower than expected. And the different broadband opportunities (satellite, TV, last mile, etc.) are beginning to emerge, each with its own particular flavor.

Broadband wireless can be looked at as having three basic levels — macro, micro and pico. Each has its own particular strengths and weaknesses, based on these criteria: application, bandwidth and coverage area. At the pico level are Bluetooth and wireless local area networks (WLANs). These technologies operate from 1 to 54 Mb/s and they are content-insensitive. They can transmit data, voice and video, and can perform file sharing and remote access within a small (typically 100 to 300 meters) zone. They are unlicensed and generally dynamic and ad-hoc focused.

At the micro level, the typical application is wireless Internet with fixed and mobile Tx/Rx points, as well as a larger coverage area.

At the macro level is cellular data — code-division multiple access (CDMA), typically.

Each has its own particular bandwidth requirements.

One promising broadband technology showing noteworthy progress at the micro level (at least in planning) is fixed wireless. What makes this so attractive is that a 6 MHz wireless TV channel, using 64-quadrature amplitude modulation (QAM), can support 27 Mb/s of downstream data. This has service providers salivating at the potential to deliver data-intensive multimedia and bandwidth-hogging Internet junk content.

This also has providers looking hard at one particular segment, fixed wireless broadband, as a wireline and cable replacement for the last mile, or wireless local loop — the last leg of the cable route from the distribution box to the home.

The last mile has been a target of opportunity for some time now but has been logistically difficult. To provide similar bandwidth by traditional means requires overhauling the wired infrastructure by upgrading hardware and stringing thousands of additional miles of fiber, coax or twisted-pair wiring — an ugly scenario.

Fixed wireless presents an attractive solution because it has the ability to rapidly introduce high-speed data access, Internet and high-quality digital data, video and voice services throughout a metropolitan area without the time, cost or delay of wired plant upgrades. And, it can do so for more than just the last mile. It is attractive for wireless L/W/M-ANs (M is metropolitan), Internet and fixed and mobile multimedia delivery. Basically, all a wireless operator has to do to be open for business is pick a frequency, and install a head-end and transmission tower.

This particular flavor of wireless broadband is about as perfect a solution as can be envisioned because it has the potential to deliver the much-heralded consumer applications being touted as the driving force behind the next technological revolution, and without any of the wired infrastructure issues.

Presently, the frequency of interest for wireless cable, Internet and WLANs is the 2.1 to 2.7 GHz band (see Table 1). There is also some activity in the 5 GHz and super-high frequency (SHF) range of 27.5 to 29.5 GHz.

The 2 GHz band is the most active due to cost and, to a lesser degree, technology (technology, of course, is becoming less of an issue at today's pace of development). The equipment for the 5 GHz, and especially the SHF band, is a relative issue — the higher the frequency, the less of, and more expensive the equipment becomes.

And, propagation issues increase with frequency. Short-term progress is more likely to occur at the lower frequency where technology is more mature and less expensive (and more robust). Eventually, however, like all technologies, the price/performance/equipment curves of the higher frequencies will flatten.

Delivery systems evolving for the broadband arena include multipoint distribution services, (MDS), multichannel multipoint distribution service (MMDS) and instructional television fixed service (ITFS). There are others in the wings, such as the ultra-high frequency (UHF) TV bands, but politics and current user allocations are slowing progress there. Eventually, however, UHF will give up some of its bandwidth, so look for this to pop dow the road.

Presently, potential fixed wireless providers have aggregated available MDS, MMDS and ITFS spectrum, in a given market, providing as much as 200 MHz of bandwidth. This is the equivalent of 33 analog 6 MHz TV channels. And, from a digital perspective, this is equivalent to 200 Mb/s of raw digital bandwidth.

Whether it is WLANs, the last mile or wireless Internet, fixed wireless is based on line-of-site (LOS) technology (but, that's changing). This can present issues of reliability similar to those of satellite or broadcast TV.

Distance is probably the top design consideration. Wireless cable signals can typically only be received within a 30-mile radius of the transmitter, and that's under optimal conditions.

Impediments to signal path such as dense tree cover, hilly terrain, buildings and heavy precipitation can degrade reception. Multipath distortion (signal reflections off of buildings or other structures) also causes problems. However, because most of these issues have been around for a long time, the industry has a pretty mature database from which to model.

On the other hand, fixed wireless has one obvious advantage. In terms of performance, fixed wireless allows for downstream connections ranging from 500 Kb/s to 155 Mb/s — faster than digital subscriber lines (DSL), cable, or T1 lines. The only significant disadvantage at this point is that the upstream data is much slower. This is because the upstream data is usually sent via a telephone line limited to about 53 kb/s.

This works because today about 80% of Internet traffic is downstream. But that's changing as large files (multimedia in particular) are finding their way upstream as well. Therefore, to accommodate this shift, some wireless channels (particularly the ITFS) are being allocated to become the upstream data port for fixed wireless. This will dramatically improve the uplink data rate and enhance two-way file and image transmissions.

The most prolific applications for potential fixed wireless systems today are for broadband access to the Internet or establishing local/wide/metropolitan-area networks. Basically, the infrastructure is similar for both.

The network infrastructure requires three main ingredients: protocols, interfaces and hardware. Software, in terms of middleware and end user also play a role, but because this article focuses on wireless, those topics will be left for discussion at a later time.

Essentially, the focal point of the wireless access point has become the mobile computer. The lines of distinction have blurred when it comes to using it to link corporate businesses, consumer services or Internet connectivity. All can use fixed wireless in one fashion or another and require much the same wireless infrastructure. As one lowers the level, one starts to see particular technologies or layers used.

One argument for employing fixed broadband is that if the wireless network can use the same networking protocols as the fixed-end network, connections are more streamlined and interface issues become more manageable. Furthermore, standard protocols and routers can be used to interconnect the wireless network to the fixed-end networks (see Figure 1).

Wireless links for the above generally use the following protocols. Obviously, there are variations on the protocols, but the platform has to be relatively standard.

Four main layers exist in such networks. The lowest is the physical layer where the RF carrier signal is digitally modulated to create a bit stream. It contains all of the modulation and transmission protocols like forward error correction (FEC), transmission rate determination, interleaving and other techniques to maximize the data transmission and minimize errors.

Above that sits the link layer. This is where things usually go wrong because it demands the most proprietary and specialized protocol optimized for the radio environment. The process involves link protocols that interact between the wireless modems and base stations. Given the fact that so much equipment already exists, much of which is legacy, coming up with a form of medium access that is acceptable to all is daunting at best.

Layer three is the network layer. Some (not all) wireless WANs, such as macro-level RAM mobile data and ARDIS have been around for a while. This is another area where one finds a fly in the ointment. Inherently, these networks are highly proprietary. They use network-layer protocols designed specifically for that network, so often trying to interface to them is another daunting challenge. It's not that it can't be done, it's just that the more interfacing one has to do, the slower and more complex the system gets.

Fortunately, as of late, and thanks to the Internet, the trend is being forced toward Internet protocol (IP) stacks. This may be the saving grace for ubiquitous interconnect because, like it or not, it has become the defacto standard for public interconnect protocol. And everyone knows what happens when one tries to buck a widely accepted trend. Witness that such is the case with cellular digital packet data (CDPD), as well as the packet services being developed for personal communications services (PCS) networks (global system for mobile communications — GSM, CDMA and time-division multiple access — TDMA).

The fourth communal layer is the transport layer. Above this layer are things like applications, session initialization protocols and other various, industry protocols. These layers usually are not part of the wireless network, but are implemented as part of the application solution. Some transports have been designed specifically for wireless networks. But it is also possible to use tried and proven transports such as TCP, though some optimization of TCP's timing parameters and algorithms tends to yield better results.

Once the protocol is in place, access to the network is necessary. This is accomplished by the interface. Because the interface is largely software, it is only given a cursory overview.

Interfaces determine access points to the network, both at the mobile and at the fixed-end. At the mobile, the interfaces of interest are between the application and the protocol stack and between the mobile and the wireless modem. At the fixed end, the area of interest is the interfaces the wireless network presents.

Given that the protocol stack and interface are in place, hardware is the final plug-in. Hardware is determined by the above, but does have some variables.

Mobile computers, Internet appliances, personal digital assistants (PDAs), 3G phones and whatever comes down the line next are all going to have to be wireless, multifrequency, multiprotocol and broadband. The biggest issue here will be hardware compatibility. The next generation of wireless devices must be system-agile and protocol-independent — meaning that they must configure on-the-fly and be pay-and-go. End users aren't interested in reading directions or manually configuring wireless devices — the cellular industry taught us that. It will be interesting to see how this issue shakes out.

As we are all painfully aware, reliability is the bane of wireless communications and current wireless infrastructures fall short. Assuming the technological and political issues previously discussed are resolved (and, without a doubt, they eventually will be), the last major hurdle to over come is the LOS roadblock.

Ultimately, broadband wireless must provide subscribers with high data rates over non-line-of-sight (NLOS) fading channels at wireline reliability — no insignificant challenge. And the key here is non-line-of-site.

NLOS has been addressed in the IEEE 802.16a channel models, which address NLOS characteristics such as Rician K factor and delay spread.

Putting aside the normal technospeak that accompanies such standards or specifications, NLOS is designed to find a way to receive signals based on dynamic conditions (read: addressing fading).

While there are several levels of “near-LOS,” let's focus on true NLOS links.

Such links address deep fading. For single-input, single-output channels under NLOS, the random received-signal level is said to be Rayleigh-distributed. Rayleigh-defined links are defined by the 10/10 rule, which states that 10% of the time, a signal is 10 dB below, or effectively one-tenth of its nominal level. That means that, to guarantee that a 10 dB fade can be overcome, the transmit power sent from a transmitter has to be 10 times that of LOS — or, 10 dB margins.

For even better reliability (99%), the Rayleigh statistics tell us that 1% of the time a signal is 20 dB below, or 1/100 of, its nominal level. That means a 1% down time must be accepted, with power cranked to 100X that of LOS. Finally, to ensure 99.9% reliability, a 30 dB power boost (1000X) would be required.

Because this amount of extra power is unworkable, the ideal of spatial diversity is being given serious consideration. This means that, when receiving, a signal is simultaneously sampled and combined at multiple locations in space. Transmitted signals are simultaneously launched out of multiple locations in space. Wow…this looks like it just might work.

Diversity isn't new. But looking at it for broadband wireless is certainly forward thinking. Antenna diversity is a strong contender for a reliable and viable (the IEEE thinks so as well) next-generation wireless broadband infrastructure. To date, wireless broadband has promised much and delivered little. While NLOS issues are paltry when compared to the political and technological issues, they may just be the final crank that starts the engine. If reliability can be used as a marketing tool, the customer might just listen.