Enabling Sensing-based Opportunistic Spectrum Re-usage with Secondary QoS Support

Within the last years a discrepancy between the spectrum licenses and the actual usage of spectrum has been observed. While the vast majority of frequency bands attractive for wireless communication are licensed, measurement campaigns have shown that large portions of the spectrum are temporarily unused in many locations. Sensing-based opportunistic spectrum re-usage has been identified as an attractive approach to overcome this discrepancy. In this approach Cognitive Radio (CR) based Secondary Users (SUs) sense the licensed frequency bands owned by Primary Users (PUs) for available spectrum and use the temporarily available spectrum on an opportunistic basis with the constraint to vacate the spectrum as soon as the license holder returns. An apparent challenge for such secondary spectrum usage is the reliable detection of the PU communication. The SUs have to ensure that the licensed spectrum is always vacated in a timely manner and that no harmful interference is created to the PUs. However, not only the protection of the PU communication is a challenging task for such CR networks but also the maintenance of a proper Quality of Service (QoS) for the secondary communication. Due to the strict access priority of the PUs, the secondary communication potentially has to be often relocated to new, temporarily available frequency bands. In this thesis we present and evaluate a CR system design, which is able to cope with these two challenges. We show that the proposed system design can achieve both reliable PU protection and secondary QoS support even for small secondary networks consisting of simple, low complexity CRs using energy detection-based spectrum sensing for the PU protection. Using a proper amount of spectral overhead for spectrum sensing and for secondary QoS support, reliable PU protection and secondary QoS support can also be maintained in environments with very high variability of temporarily available spectrum. We evaluate the tradeoff between both overheads and show that there exists an optimal joint spectral overhead which maximizes the spectral efficiency. We further show that, while the spectral efficiency of initially small network deployments based on the proposed system design might be low, it is still significantly greater than zero. Furthermore, the spectral efficiency is improved, as the network size increases. This makes the proposed CR design an ideal approach for initial deployments of very small and cheap CR networks, which are perfectly scalable: with an increase of network size (which usually goes hand in hand with an increase in spectrum demand), also the spectral efficiency is increased.