Wideband Spectrum Sensing Based On Sparse Group Lasso

With the continuous improvement of science and technology, more and more mobile devices come into the life of people gradually. Mobile devices can only work within a certain range of the radio spectrum, and under the available spectrum resources, with the expansion of the data business, has become very congested. In order to improve the utilization rate of spectrum, the federal communications commission(FCC) show that in the case of authorized spectrum have spare parts, unauthorized users can opportunity to conduct dynamic spectrum access type. The cognitive radio technology can improve the spectrum utilization, avoid the waste of spectrum. Spectrum sensing is the key technology of cognitive radio, now the study of wide band spectrum perception has become a hot topic. How under the low signal-to-noise ratio can reach a high detection performance is a difficulty. Aiming at this problem, this paper proposes a wideband spectrum sensing method based on sparse group Lasso.Firstly, this paper puts forward a method of sparse group Lasso(SGLasso) for wideband spectrum sensing. This method need a model that take advantage of the sparsity of primary users(PU) and primary user transmission power spectral density(PSD), this converted wideband spectrum sensing problem to estimate of the PU transmitting PSD, reconstructed PSD vector and realized the spectrum detection.Secondly, the spectrum sensing method of sparse group Lasso if the channel gain matrix are known, but in general the channel gain matrix is not certain, this will be affected by some interference, then the detection probability of the cognitive users will be affected. In order to solve this problem, sparse group Lasso can be combined with total least squares(TLS). This paper proposes a spectrum sensing method based on sparse group total least squares. thereby solved the problem of the channel gain matrix uncertainty.Finally, using the multivalued model of sparse group Lasso combined with cognitive users cooperative detection in spectrum sensing, create a new detection method of cooperation, the cooperative spectrum sensing model based on sparse group Lasso. With the help of the sparse group Lasso parameter model, it solved the unknown parameters by block coordinate descent. Comparing with the sparse group Lasso spectrum sensing algorithm, in the case of low SNR sparse group Lasso cooperative sensing algorithm can achieve high detection performance.