Research Of Wideband Spectrum Detection Based On Compressed Sensing  Posted on:20160828  Degree:Doctor  Type:Dissertation  Country:China  Candidate:W G Wang  Full Text:PDF  GTID:1108330482973190  Subject:Signal and Information Processing  Abstract/Summary:  PDF Full Text Request  Nowadays, demands for the application of radio spectrum are constantly increasing, leading to scarce spectrum resources. Cognitive radio technology is one of the effective solutions to solve the problem of spectrum scarcity, and the main idea of this technology is to offer the idle spectrum to share with second user temporarily under the premise of ensuring the primary user normal communication. The second user needs to make quick and accurate judgment on the spectrum hole in CR system and it is a prerequisite for cognitive radio communication. The spectrum hole judgment is a process needs sampling firstly and analyzing secondly. Because of the limited ability of analog to digital devices for processing wideband signal, the traditional method cannot make quick and direct judgment for the wideband spectrum holes. Compressed sensing technology is a new kind of signal processing technology with simultaneous data acquiring and compressing. This technology can break through the Nyquist sampling rate on the wideband spectrum quick detection and it has attracted increasing attention in the field of cognitive radio technology in recent years.Based on the existing study of wideband compressed spectrum sensing technology, we mainly completed the following innovative work in this thesis:(1) Cooperative compressed spectrum sensing is based on second users clustering. Compressed sensing realize a direct data acquisition process from which analog signal is converted to information. Compressed spectrum sensing means the compressed sensing technology is used in spectrum detection process. In the multiple cognitive users system, user’s collaboration can improve the detection probability. However, seriously signal fading occurred for users with longer distance to the fusion center, leading to the deterioration of the overall sensing performance. In this thesis,An algorithm of collaborative clustering simultaneously orthogonal matching pursuit is proposed by combing with the distributed compressed sensing technology and the model of collaborative clustering compressed spectrum sensing is constructed. Each cluster header collects the compressed data from all user’s in this cluster, and the signal’s energy value calculated by the header is sent to the fusion center and the signal’s energy value will be weighted processing. The problem of remote users bringing about negative effects is solved in this innovation job, and it improves the reconstruction probability of multiple user’s compressed data. All the subbands can be detected by using the collaborative clustering compressed spectrum sensing algorithm in this thesis, and it gives out the result of spared subbands quickly and accurately.(2) The compressed spectrum sensing technology is realized by combing the data in spatial and frequency domain based on the wireless sensor network. Wideband spectrum is detected in this thesis by using wireless sensor network. Because the wireless sensor network node’s data has spatial correlation, these data exhibited a characteristic of sparse on the spatial wavelet basis by designing the latter. We design projection matrix corresponding to the spectrum detection based on wireless sensor network, and investigate the division of labor and cooperation of multiple nodes based on different band detection and the data correlation between nodes is extracted by lifting wavelet technology. By designing reasonable and effective wavelet basis and reducing the computation complexity of single node, a sparsely representing method of spectrum data in spatial wavelet is proposed by using the data correlation between nodes. In this thesis, the model of data compressed in frequency and spatial domain is established and the spatialfrequency compressed spectrum sensing joint reconstruction algorithm is proposed and the problem of poor reconstruction effect in one dimension is solved. Results of simulation indicated that, based on the same total compression ratio, the proposed algorithm can improve the reconstruction probability, and it can make an accurate judgment on spared subbands through the wireless sensor network in a short time.(3) Joint optimization of projected matrix and the adaptive process is used in the compressed spectrum sensing. The projected matrix affects the quality of the compressed sensing reconstruction effect directly and the reconstruction probability can be improved by reducing the relation of column vectors in projected matrix. When the similarity between column vectors is higher, the pursuit ability of reconstruction algorithm is weaker. In order to improve the probability of compressed sensing spectrum detection, it needs to design appropriate projected matrix and do optimization. We investigate how to reduce the relation of projected matrix column vectors, and design an optimization algorithm for the projected matrix in compressed spectrum sensing. In addition, the observation number of compressed sensing also affects the reconstruction accuracy. In order to select the appropriate number of observations, sequential measurement is used to adjust the projected matrix size so as to obtain the estimation error without reconstruction, and the adaptive method of projection matrix in compressed spectrum sensing is proposed. Combined with the projected matrix optimization and adaptive number of observation,the achieving model of the joint optimization based on the adaptive projected matrix is constructed,and the optimized algorithm of the sparse adaptive matching pursuit in wideband detection is proposed. The problem of seeking the optimal projected matrix is also solved. Results of simulation showed that the performance of this algorithm is better than that of the traditional method of matrix without optimization.(4) Distributed Bayes compressed spectrum sensing is proposed based on improved relevance vectors machine. Although the spectrum of signal to be measured is unknown usually, the situation of subbands occupied by PU is usually affected by experience and has a characteristic of certain prior probability distribution. Bayesian CS can effectively use the prior knowledge in subbands occupation. So, by using the spectral prior information, the Bayesian CS for judging the spectrum hole has certain advantages. In this thesis, the Bayesian classification and regression model is studied by analyzing the compression and reconstruction process of Bayesian compressed spectrum sensing. Besides, the classical Bayes reconstruction algorithm and the hierarchical prior reconstruction of Laplasse algorithm is discussed. The spectrum reconstruction technology of distributed Bayes compressed sensing is explored according to the characteristics of multiple cognitive distributed users. By modifying the target function of multinodes joint Bayes reconstruction, a distributed Bayes improved compressed spectrum sensing model is set up, and an algorithm of DRRVM is proposed in this thesis. It solves the problem of week robustness by single user. Results of simulation indicated that this algorithm can effectively discard the outliers and improve spectrum detection probability significantly.  Keywords/Search Tags:  Cognitive Radio, Wideband Spectrum Detection, Compressed Sensing, Signal Reconstruction, Wireless Sensor Network, Distributed Compressed Sensing, Bayes Compressed Sensing, Enegy Detection, Information fusion  PDF Full Text Request  Related items 
 
