Abstract:
Opportunistic spectrum access is a proposed solution to the problem of increasing
scarcity of radio resources. In certain bands, spectrum is utilized extremely inefficiently
by the licensed, or primary, users. Opportunistic spectrum access would allow a secondary
user to utilize spectrum when the primary user is idle while not causing harmful interference
when the primary user is active. Spectrum sensing techniques determine portions of the
spectrum that are occupied by primary user signals at a given time and location. Temporal
sensing of a known narrowband channel involves modeling the temporal dynamics of the
primary user signal and performing estimation and prediction of the primary user state.
Wideband sensing involves determining which parts of a given wide spectrum are occupied
or unoccupied at a given point in time. Both temporal and wideband sensing have been
studied extensively in the literature. There has been relatively little work on temporal
sensing over a wide spectrum band with either well-defined or unknown channels.
In this dissertation, novel approaches to wideband and multiband temporal sensing are
developed. A class of hidden Markov models is proposed to jointly model time dynamics of
the primary system and channel impairments between the primary user and the secondary
user over a wide spectrum band. Methods to segment a wide spectrum band into individual
channels and to optimize parameter estimation over the channels are proposed. Simulation
results are presented to evaluate the effectiveness of the proposed wideband and multiband
temporal sensing schemes. Some comparisons to performance bounds are provided.
