The Unlicensed Frequencies: A Matter of Peaceful Coexistence

Beginning in the late 1980s a number of nations, including the United States, began to explore
a new concept, sometimes known as Open Spectrum, that called into question the whole
rationale behind the licensing of spectrum, or, in effect, making it the property of the license
holder. The concept of Open Spectrum led more or less directly to the creation of the unlicensed
bands.
This concept is that spectrum, rather than being private property, should be a commons—
a shared resource available to all. To avoid the tragedy of the commons—that is, the mutual and simultaneous overexploitation of a scarce resource to the ultimate benefit of no one—the
Open Spectrum will be open only to users of various kinds of spread spectrum radios. The
users must have the following attributes:
• They have the ability to tolerate high values of interference.
• They are frequency agile within the band, distributing the signal among various coding
sequences, subbands, and time slots so that no slice of spectrum is exclusively occupied
by any individual user.
• They utilize various mechanisms for ensuring fairness such as mutual power control,
network polling for controlling access, and the exchange of information among intelligent
radios via control channels.
The first commercial spread spectrum radios appeared in the early 1990s at just about the
same time that the FCC began to allocate unlicensed spectrum for miscellaneous uses. The
first band so designated was the ISM band situated at 902MHz to 928MHz. This was shortly
followed by the band extending from 2.4GHz to 2.4835GHz, the unlicensed band that is most
subject to use today in transmitting data. This in turn was followed by two contiguous bands
extending from 5.15GHz to 5.23GHz and another band between 5.725GHz and 5.825GHz. The
5GHz unlicensed frequencies are often referred to as the Unlicensed National Information
Infrastructure (U-NII) bands and are subject to more usage restrictions than the lower
frequencies, being explicitly designated for data transmissions only, not for the operation
of remote control devices. In November 2003, the FCC allocated 250MHz of new spectrum
extending from 5.470GHz to 5.725GHz, a band that is already in use elsewhere in the world. By
this action the FCC has harmonized U.S. spectral allocations with those in many other nations,
which should stimulate product development and bring down equipment prices.
Abroad, both 2.4GHz and 5.8GHz as well as 5.4GHz are commonly available as unlicensed
spectrum, though not on a worldwide basis.
A further unlicensed band occupies spectrum between 24.0 and 24.25GHz in the United
States, and another is situated between 59GHz and 64GHz. The 24GHz band is newly approved
and has seen little or no use as yet. The 60GHz band has been commercially exploited on a
small scale thus far, and the principal user is the now-defunct CAVU-eXpedient network in
Florida. Rumor has it that the FCC will probably allocate additional spectrum for unlicensed
usage in the midterm. Incidentally, the 60GHz band is in use in some other areas of the world
as well.
The general rule for operation within the unlicensed bands in the United States is that
users must be prepared to tolerate interference but must not generate undue interference
themselves that would prevent others from utilizing the bands. All users are essentially
enjoined to live and let live, and what is supposed to ensure that this will happen are rather
severe output power limitations placed upon individual radios.
Since for many prospective broadband wireless operators licensed spectrum is simply
unavailable, and unlicensed spectrum is the only option, such rather vague injunctions cannot
but be disturbing. Won’t interference inevitably increase as more and more entities use the
bands? What is to prevent everyone from interfering with everyone else at a certain point,
thereby rendering the bands useless and undermining the whole concept of Open Spectrum?
These are valid concerns, but in my experience of covering wireless broadband as a trade
journalist, which dates from the earliest days of the industry, I have encountered relatively few
instances where a network operator was severely hampered by high levels of interference. This is not to say that interference is not an issue; it is in fact a matter of some concern, particularly
within the increasingly crowded 2.4GHz band. Still, for a variety of reasons, it is unlikely to be
such as to cripple a broadband public network operating in the unlicensed bands.
Transmission Control Protocol/Internet Protocol (TCP/IP), the networking protocol specified
for 802.16, is best effort in its basic form, which means that the nodes will retransmit in the
presence of contention and dropped packets. If the traffic is suited to best effort, then the only
result of a moderate increase in interference will be a slowing of throughput, which may or may
not be noticeable to the user. If the particular application depends on maintaining a certain bit
or error rate, however, then interference can pose a serious problem, and that would be the
case for voice, multimedia, or real-time interactive applications. With the expected increase in
such applications, concerns over interference and crowding in the unlicensed bands are likely
to grow as well, though at the same time, new technologies for mitigating interference will be
entering the marketplace.
The worst possible state of affairs is where users of unlicensed spectrum attempt to solve
the problem by operating at maximum or above-maximum power levels in an attempt to rise
above the interference. The situation then becomes akin to a rock band where every player is
continually cranking up his amplifier in an effort to be heard above his band mates. The end
result is a deafening cacophony where no one is heard distinctly. Unfortunately, inexperienced
network operators often set their output levels at the maximum as a matter of course. The irony
is that even when other operators are not present, the single operator may actually be
degrading the signal at the receiver by increasing the level of self-interference from reflected
signals.
Network operators faced with severe and intractable interference and with the resultant
difficulty of honoring service-level agreements in its presence have a number of options. They
can utilize high-gain, highly directional antennas that focus energy at the receiving node and
establish a high carrier-to-interference ratio at that point in space. They can seek to identify
interferers and see if a means of coexistence can be established through negotiation. They
can seek injunctive relief by complaining to the regulating body if they can determine that an
interferer or interferers are exceeding power limitations. Or they can try to confine their transmissions
to less highly trafficked unlicensed bands, such as the 5GHz bands; however, by
moving up in frequency to less-crowded regions of the spectrum, they may also be giving up
effective distance and thus are somewhat restricting their potential markets.
In the future, smart radios with adaptive antennas may be able to distinguish interference
from the desired signal and cancel it out at the antenna. This may not completely solve the
problem because radio front ends may still be subject to overload if the amount of energy in
the band rises to a certain level, but the intelligent radio will certainly provide an effective
means of coping with interference in most cases. Currently, however, the cost of radios
utilizing this technology is still high, and to date these techniques have been primarily
employed in military radios and radar to thwart jamming and spoofing attempts.
In the future, you may also see mandatory coordination of all radios in an area utilizing
unlicensed frequencies. Such coordination would serve to control the power levels of individual
transceivers and maintain energy levels below certain stated limits. Power control of
this type has long been a characteristic of licensed CDMA mobile phone networks and has
been proven to work well. Extending it over a motley assemblage of unaffiliated users may pose
some policy problems, though.