| In recent years,with the development of mobile communications into a new era of B5 G and6G standards,Low Earth Orbit(LEO)satellites and space Internet have also become research hotspots.The continuous increase in the number of satellites and the development of various new services will lead to more demand for spectrum resources.However,as a non-renewable resource,the shortage of spectrum will seriously limit the future development of satellite communication systems.With the realization of many LEO satellite constellations programs in the Ku/Ka band,it is of great significance to study and analyze the coexistence technology of Geostationary Earth Orbit(GEO)satellites and LEO satellite systems.In this thesis,in the scenario of cognitive LEO satellites sharing GEO satellite spectrum resources,a dynamic spectrum sharing scheme is proposed,which can increase the spectrum availability of cognitive LEO satellite networks without affecting the operation of existing GEO satellite services.The main research contents of this thesis include:Aiming at the long-distance transmission of satellite communication links and the problem of poor spectrum sensing performance due to complex spatial environment,this paper proposes a spectrum cognitive scheme based on Radio Environment Mapping(REM).In this paper,a dualthreshold energy detection algorithm combined with a soft-hard fusion strategy is used to achieve a good trade-off between the complexity of the cognitive LEO satellite system and the perception accuracy.Meanwhile,the spatial detection probability and spatial false alarm probability are included in the study to show the reliability of REM for spatial description.The simulation results show that,under the premise of alleviating the burden of cognitive links,the test performance of the proposed hybrid spectrum cognitive scheme can still approach the performance of using only soft fusion approximately,and REM facilitates a reliable spatial description of the spectrum occupancy of GEO satellites.Aiming at the problem that the spectrum sensing process of cognitive satellite system has a non-negligible time lag,this paper proposes a spectrum prediction scheme based on Cascaded Neural Network(CNN)to improve the spectrum sensing process.Considering the uncertainty of spatial noise,an adaptive threshold judgment method is introduced to preprocess the spectral data;At the same time,a prediction algorithm based on CNN is adopted,which can integrate the historical spectrum occupancy information of GEO satellites and the current spectrum sensing results to predict spectrum holes,which can not only improve the accuracy of spectrum sensing,but also improve the efficiency of spectrum sensing.The simulation results show that,compared with the prediction algorithm based on Feedforward Neural Network(FNN),the prediction algorithm of CNN can effectively alleviate the spectrum conflict probability between the GEO satellite and the LEO satellite system.Aiming at the joint spectrum sensing and power allocation of multiple and multi-channel cognitive LEO satellites,this paper proposes Muti-Objective Memetic Algorithms(MOMA)to optimize the spectrum sensing time,decision threshold and power allocation of cognitive LEO satellites,so as to maximize the throughput of cognitive LEO satellites and minimize the interference to GEO satellites.Simulation results show that the proposed scheme achieves good performance and dynamic parameters in the cooperative spectrum sensing and power allocation of cognitive satellite system. |
