Research On Adaptive Measurement Matrix And Its Application In Wideband Spectrum Sensing

In a cognitive radio system, spectrum sensing is the core technology. However, in wideband spectrum sensing application, the Nyquist sampling rate is too high to be realized in hardware equipment. To deal with this problem, the current common practice is to introduce compressed sampling. By doing so, we can sample the signal at a lower rate and reduce the requirements for hardware. This article intensively analyzes the theory of wideband compressed spectrum sensing as the basis, and on account of the unknown sparsity order and uncertain noise power in spectrum sensing, the adaptive method of the measurement matrix has been proposed to improve the accuracy of the reconstruction signal. The main contributions of this dissertation are summarized as follows:1. A two-stage adaptive compressive sensing procedure is proposed. In the first stage, i.e. initial stage, the initial matrix is optimized via decreasing its mutual coherence. In the second stage, i.e.adaptive stage, each ensuing row of the projection matrix is adaptively optimized by minimizing the Cramer-Rao bound of recovery errors. This approach combines the projection matrix optimization with the adaptive process of compressed sampling, and it is indicated in Simulation results that the proposed method has a better performance than many traditional compressive sensing methods in the aspect of anti-noise ability and recovery accuracy.2. An adaptive compressive sensing algorithm based on sequential detection is presented. As the observations is obtained, the subsequent sensing vectors is designed according to previous estimates.In the follow-up observations, the information of the original signal can be better maintained, thus improving the SNR of the measurements and decreasing the reconstruction error. In the simulation, it is demonstrated that the algorithm can further reduce the number of observations under the same requirements of detective performance.3. The Fast Bayesian Matching Pursuit algorithm based on optimized measurement matrix is proposed. It introduces the projection matrix optimization into the Fast Bayesian Matching Pursuit algorithm, and combines the signal estimation and projection matrix construction. Here, measurement matrix is optimized with less cross correlation which effectively improves the estimation accuracy of the sparse signal. Simulation results show that compared with other algorithms, the proposed algorithm can improve the estimation accuracy of the signal under the condition of low SNR.