Date of Award

January 2015

Document Type


Degree Name

Master of Science (MS)


Electrical Engineering

First Advisor

Naima Kaabouch


The proliferation of mobile devices led to an exponential demand for wireless radio spectrum resources. The current fixed spectrum assignment has caused some portions of the radio spectrum to be heavily used whereas others to be scarcely used. This has resulted in underutilization of spectrum resources, and, hence has demanded the need for solutions to address the spectrum scarcity problem. Cognitive radio was proposed as one of the solutions. One of the techniques involved in cognitive radio is the dynamic spectrum access technique. This technique requires the identification of free channels in order to allow secondary users to exploit the spectrum resources. The process of identification of free channels is known as radio spectrum scanning, which is performed by sensing a particular channel in the radio spectrum to determine the presence or absence of a signal. In most of existing studies, the frequentist technique using energy detection with fixed threshold was used to scan the radio spectrum. However, this method comes with a major drawbacks. First, energy detection is unable to distinguish between signals and noise and suffer for high false detection rates. Second, energy detection has high false alarm probability. Finally, frequentist techniques are subject to uncertainty and do not provide real time monitoring/sensing. Therefore, the goal of this thesis is to develop a more efficient scanning technique that deals with uncertainty and scans the radio spectrum in real time and determines its occupancy levels.

An enhanced spectrum scanning approach is developed using an efficient spectrum sensing technique: an uncertainty handling Bayesian model along with a Bayesian inferential approach. Two Bayesian models are developed: 1) a simplified model, and 2)

an improved model to incorporate the Bayesian inferential approach to estimate the

spectrum occupancy level.

The performance evaluation of the proposed technique has been done using simulations as well as real experiments. For this purpose, two metrics were used:

probability of detection and probability of false alarm. Furthermore, the efficiency of the

proposed technique was compared to the efficiency of the frequentist technique, which uses only a spectrum sensing technique to identify the occupancy of the spectrum channels. As expected significant improvements in the spectrum occupancy measurements have been observed with the proposed Bayesian inference method.