Compressed Spectrum Sensing For Congnitive Radio Based On USRP And GNU Radio

In recent years, more and more wireless communication equipments access spectrum, whichleads limited spectrum resources to be increasingly scarce. Unlicensed users cannot use the licensedband because of traditional static spectrum allocation strategy, which makes unlicensed spectrumresources are greatly wasted. Cognitive users can make fully use of idle spectrum by spectrumsensing and spectrum sharing which can improve spectrum utilization. Compressed spectrumsensing becomes a hot research topic of spectrum sensing technology based on sparsity of signalthat can be reconstructed with a little number of observed data.Firstly, the concept of compressed sensing is introduced in detail, including three keycomponents of compressed sensing such as the sparsity of signal, the selection of observationmatrix and the signal recovery algorithms, and then varieties of applications with compressedsensing are summarized.Secondly, a two-step compressed spectrum sensing algorithm based on sparsity estimation isstudied. By Monte Carlo simulation and curve fitting methods, sparsity of signal can be obtainedwith limited priori information, then the final results are achieved by the aid of sample signal withthe sparsity estimation. Analysis and simulation results show that the proposed algorithm canachieve the similar performance of the traditional algorithm, requiring less sample observationpoints, and the rate of spectrum sense is increased and the cost of signal processing is reduced.Lastly, spectrum sensing algorithm based on energy detecting is implemented on the platformof USRP and GNU Radios. By sensing the local signals, the GSM signals and wifi signals, we canobtain occupation of different frequency channels. Furthermore, the compressed sensing algorithmbased on this platform is described with the flow diagram.