The increasing popularity of wireless communication and multimedia applications yields a growing demand for radio spectrum. However, most of the commercially interesting spectrum has been assigned to various services and allocating further spectrum is a cumbersome process. Some authors claim that current policies are inefficient and cause an artificial spectrum scarcity. Recent measurements support these statements by showing that significant amount of spectrum is underutilized. Secondary use of licensed but vacant spectrum has been proposed as solution to this problem and several architectures for Dynamic Spectrum Access (DSA) have been studied. The most popular one assumes prioritized Primary Users (PUs), who often possess the spectrum license, and Secondary Users (SUs) that opportunistically access spectrum bands left idle by the PUs. The SUs search for vacant spectrum bands, so-called spectrum sensing, and may exploit these as long as the secondary emissions do not cause harmful interference to the primary system. Most measurement activities have been carried out in the USA and there is a lack of comparable results from Europe. Additionally, the scope of most campaigns has been limited to selected spectrum bands or the sole estimation of the duty cycle. Similarly, only simplified models focused on single aspects of spectrum use are available that prevent a more complete study. We address these shortcomings in this thesis and present an integrated approach to spectrum modelling and evaluation of DSA. The empirical work spans from the careful design of two flexible measurement setups to the execution of extensive spectrum measurements in various scenarios and two European countries. We confirm that considerable amount of spectrum is underutilized and verify further analytical results on cooperative sensing. Based on the gathered measurement data, we identify main properties of spectrum use and present models that reproduce almost all determined characteristics. In this process, special care is taken to develop a generic modelling methodology not specific to our measurement scenarios but still provide realistic model parameters. As example for model applications we study adaptive spectrum sensing using a renewal theory based problem formulation. Exploiting knowledge on spectrum use statistics enables performance enhancements although we also point out that the gains computed in previous analytical studies can hardly be accomplished under more realistic conditions. This result shows the importance of accurate and comprehensive spectrum modelling as put forward in this thesis. Recent spectrum regulatory decisions to allow for secondary access in selected spectrum bands and the increasing number of standardization activities demonstrate the wider interest in DSA. However, most existing work has been limited in scope and focused on single system components. We believe that the presented modelling methodology and the listed parameters provide a valuable basis for further DSA research under well justifiable assumptions.
展开▼
