Abstract:
The scarcity of radio spectrum poses a significant challenge to the sustained growth
of wireless communications, since most of the useful radio spectrum is already allocated
for licensed users. However, recent spectrum measurement studies have shown that there
are plenty of white spaces or spectrum holes that could be utilized opportunistically by
"secondary" users, provided that they do not cause harmful interference to the primary
users.
In this dissertation, we develop efficient methods by which a group of secondary users
equipped with cognitive radios can determine and access spectrum holes opportunistically
based on signal measurements. The cognitive radios are frequency-agile in that they can
dynamically tune to different frequency channels for transmission and reception. By ex-
changing signal strength measurements, a group of cognitive radios can calculate maximum
likelihood estimates of the location and transmit powers of the primary transmitters in the
system. We apply the Cramer-Rao bound (CRB) to characterize the error in the primary
system parameter estimates. The parameter and error estimates are then used to derive an
approximation to the Maximum Interference-Free Transmit Power (MIFTP), which is the
maximum allowable power that a given cognitive radio can use on a given frequency channel subject to an interference constraint. To mitigate interference from multiple cochannel
primary transmitters, secondary nodes maintain a distributed database that records the
location, power, and error estimates of cochannel nodes for each frequency channel.
The proposed MIFTP approximation takes into account errors in spectrum sensing such
that the approximation becomes more conservative when measurement errors accrue, and
conversely, it becomes more accurate when better measurement data is available. The property of being conservative is important, since secondary users must avoid causing harmful
interference to primary users. We also propose two model identification and measurement
clustering criteria to identify the number of cochannel primary transmitters and to cluster
the measurements appropriately for more accurate estimation of the primary system parameters. Finally, we extend the proposed opportunistic spectrum techniques to incorporate
angle-of-arrival information for improved localization accuracy and hence tighter estimates
for the MIFTP.
