| The rapid development of wireless communication technology has caused increasing scarcity of spectrum resources available for allocation. Cognitive Radio(CR) theory provides a strong technical support to solve the problem of scarce spectrum resources, improve spectrum utilization, as well as allocate and manage dynamic spectrum. Spectrum sensing technology is the primary task of cognitive radio. Only by rapid, accurate and reliable detection of spectrum holes can we effectively redistribute the frequency resources to complete the follow-up communication links. As people’s demand for wireless services increases, communication displays the trait of broadband. The increasing sampling rate and the limited processing speed become the major bottleneck that limits the development of broadband spectrum sensing theory. Compressed Sensing(CS) theory projects high-dimensional sparse signal into low-dimensional space for storage, processing and transmission, providing an effective solutions for high-speed sampling in broadband spectrum sensing techniques. In this thesis, based on the above background, we have worked to explore new techniques for broadband spectrum sensing based on compressed sensing theory. The main works and innovation points are as follows.(1) We have proposed a weighted algorithm for broadband spectrum sensing based on Differential Signal Distributed Compressed Sensing(DS-DCS) by utilizing distributed compressed sensing to achieve broadband spectrum sensing. This algorithm aims to solve the problem of high sampling rate in broadband spectrum sampling. It uses Compressed Sensing Technology to reduce sampling rate and differential processing techniques to lower algorithm complexity. To circumvent problems, such as deep fading, hidden terminals and poor noise immunity caused by single node detection, it applies distributed compressed sensing system to conduct cooperative multi-node detection and estimated SNR to weigh signals. Corroborating simulation results show that this algorithm can effectively reduce sampling rates at each node, substantially increase system detection probability and saliently improve system robustness against noise.(2) Based on the algorithm proposed above, we have presented an optimum weighted approach for wideband spectrum sensing. Distributed compressive sensing technology is exploited to obtain dramatic rate reductions while differential procedure is deduced to extremely enhance the detection sensitivity. The measurements are collected from each SU at a fusion center, where a C-out-of-J method is proposed to dramatically heighten the detection performance. From reality, we add SNR to estimate error, and use the reliable SNR that I get by taking certainty factor times the estimated SNR to improve the detection performance reduction resulting from estimation error. SCSMP recovery algorithm is utilized to reconstruct the signals, which are then weighted by the reliable SNRs. Simulation results show that the raised algorithm can effectively reduce sampling rates at each SU, substantially raise the detection performance and saliently improve system robustness against noise.(3)We have brought forward two broadband spectrum sensing algorithms exploiting compressed sensing theory as well as Support Vector Machines(SVM), which could satisfy the demand of real time applications. In the two algorithms, signal reconstruction process is substituted with spectrum sensing classifier utilizing support vector machines. The multi-level binary classifier algorithm is fit for spectrum detection environment with low real-time requirements while the single-level multi-classifier algorithm is suitable for real-time sensing with decreased system complexity. Compared to the traditional broadband spectrum sensing algorithm based on CS, the raised two algorithms could remarkably shorten the sensing delay, evidently lower the computational complexity, extremely enhance the detection performance and significantly strengthen system robustness against noise. |
PDF Full Text Request