Research On Dynamic Spectrum Access Technology In Cognitive Satellite Communication Networks

Satellite communication technology, with its advantage of large coverage and long communication distance, has been playing a more and more important role in various domains. The increase of satellite communication business requirement and application types gradually makes the spectrum resource become more and more scarce. However, the current fixed spectrum allocation strategy is the basic cause of low frequency spectrum resource utilization. In order to solve the contradiction between the shortage of spectrum resources and low frequency spectrum utilization in satellite communication network, the research on dynamic spectrum access technology in cognitive in satellite wireless networks is of the essence.At first, the main challenges in the process of dynamic spectrum access in cognitive satellite wireless networks are analyzed according to the research states at home and abroad. And after investigating the basic theory of dynamic spectrum access in cognitive satellite wireless networks, the key technologies in this process are clarified. Moreover, three recently proposed typical application scenarios of dynamic spectrum access in cognitive satellite wireless networks are analyzed and compared, and based on that, one new application scenario with centralized network architecture is designed in this dissertation. In this scenario, the limitation of load and energy consumption of satellite terminals are fully considered, so it’s easier to construct, implement spectrum sensing and suppress interference in this scenario, due to its centralized network architecture.Next, in order to deal with the difficulty of spectrum sensing in cognitive satellite wireless networks, the big data analysis based spectrum sensing algorithms for large-scale cognitive networks are investigated under the above scenario, and one new approach using mean exponential eigenvalue is proposed based on big data analysis theory. This new approach fully uses all the eigenvalues in sample covariance matrix of the sensing results to make a decision, and compared to the conventional big data analysis based spectrum sensing algorithms it can effectively improve the detection performance without requiring the prior information of licensed users.Finally, in order to decrease the licensed users’ risk of being interrupted in the cognitive satellite wireless networks, the users’ behavior prediction algorithms are investigated under the scenario designed in this dissertation, and a new multistep behavioral predicting approach based on cascade forward artificial neural network is proposed. With the powerful learning capability of cascade forward network, this approach studies and analyzes the licensed users’ historical spectrum occupation records, and then predicts licensed users’ behavior in several future sensing slots. And these prediction results can help the base stations in the cognitive network to dispatch the dynamic access process of cognitive users, in order to decrease the interference caused by cognitive users to licensed users.