Date of Award
Master of Science (MS)
Dr. Mohammed Abdel-Hafez
Dr. Atef Abdrabou
Dr. Pavel Loskot
The efficiency of cognitive radio networks mainly depends on the spectrum sensing stage, in which spectrum opportunities are exploited. However, one of the challenges facing spectrum sensing is the presence of fading and log-normal shadowing. Moreover, when the spectrum utilization is high and details regarding primary user activity are not available, a need to sense the whole spectrum arises. Hence, developing wideband spectrum sensing technique is a fundamental concern.
In this thesis a narrowband spectrum sensing in a log-normal shadowing environment is addressed, a closed-form expression for the probability of detection under shadowing is derived. The accuracy of the expression is tested using a MATLAB simulation. Collaborative spectrum sensing is addressed, and expressions for the probability of detection and false alarm in both AWGN channels and lognormal channels are derived for different fusion rules namely; soft fusion using square-law selection (SLS), square-law combining (SLC), hard fusion using OR, AND and Majority combining. The detection performance of these fusion rules is tested and compared. Simulation results showed that sensing performance is enhanced due to collaboration and better detection is achieved with more collaborative secondary users. Moreover, SLC outperforms SLS in terms of the probability of detection. OR-combining is found to outperform both AND-combining and Majority-combining from the primary user's point of view by providing higher protection for the primary user from any secondary user interference; while AND-combining is found to outperform the other two techniques, from the secondary user perspective, as it results in higher spectrum utilization and more spectrum opportunities.
Wide band spectrum sensing using wavelet-based detection is investigated. The performance of this method and the effect of parameters such as the scale factor of the wavelet smoothing function, collaboration between secondary users in edge detection and the presence of log-normal shadowing is investigated and analyzed using MATLAB simulation. Simulation results indicate that better edge detection was achieved at higher scale factor values. Log-normal shadowing affected the accuracy of edge detection since it attenuates the average power received at the secondary user, and adds random variations at the same time as detecting false edges.
Two approaches to wideband spectrum detection are investigated and compared. The first approach is the tunable bandpass filter (TBPF) filterbank. The second approach is a proposed model using wavelet-based detection. Simulation results indicate that the proposed approach performed better in terms of spectrum occupancy and utilization as it accurately detected the primary user signal. While the TBPF filterbank approach failed to detect the primary user at low probabilities of false alarm when it partially occupied the subbands, leading to more interference for the primary user.
Mohammad Al Hussien, Nedaa Yousef, "Narrowband and Wideband Spectrum Sensing for Cognitive Radio Networks in a Log-Normal Shadowing Environment" (2014). Theses. 119.