Compressive Sensing Based Dynamic Spectrum Access For Interference Networks

Future wireless information and communications technologies shall address manifold scenarios and applications. Key drivers behind this diversification are especially the vertical industries such as automotive, manufacturing, and energy. In the course of the fifth generation of cellular mobile communications (5G) the manifold scenarios and applications shall be represented by a single communication infrastructure. Therefore, we need more flexibility in resource allocation and reduce signaling overhead especially in device-to-device applications. This thesis introduces dynamic spectrum access schemes, which are enabled by compressed sensing. In particular, we propose a gray space detection scheme for cognitive radio system, which detects temporary small fraction of unused resources within an already occupied primary user spectrum band. There, gray space detection performs a uniformly most powerful one-sided composite hypothesis test exploiting noise and channel statistics. Furthermore, we extend the gray space detection scheme for multiple active transmitters simultaneously transmitting on exclusive resources. There, we propose an objective function to solve the combinatorial problem of allocation map retrieval for frequency-division multiple access signals and show that based on the objective function a receiver is able to identify the non-adjacent resources belonging to the same transmitter. Furthermore, we propose a breadth-first search approach for decision trees, which solves the allocation map retrieval problem efficient but sub optimal. Moreover, a measurement campaign is described in which practical channel models were obtained for the evaluation of dynamic spectrum access methods.