Research On Spectrum Prediction Technology In Cognitive Radio Networks

The spectrum resource becoming increasingly scarce with the development of wireless communication technology. It is an effective way to solve the problem of current spectrum resource scarcity by detecting spectrum holes based on spectrum sensing technology. However, there are some defects in the realization of spectrum sensing, spectrum decision,spectrum sharing and spectrum switching in cognitive radio networks, such as transmission delay, energy consumption and transmission collisions. In order to solve these problems, spectrum prediction technology has been widely studied. In this thesis,referred to a large number of domestic and foreign relevant references, a new spectrum model was proposed and the prediction scheme had also been improved. The work and innovation of this thesis are as follows:1. The key technologies in cognitive radio networks were investigated and summarized. Four key technologies in congnative radio networks,spectrum models of primery users and widely studied spectrum predcition methods were introduced and analyzed.2. A more practical spectrum model was proposed based on the queuing theory and the Markov model. Two kinds of channel assignment schemes were simulated: random distribution method and greedy distribution method. Moreover, the second user's perception framework was redesigned. Detailed theoretical derivation and simulation were given to calculate the throughput and spectrum correlation, which verified the correctness of the theoretical analysis and the practicability of the spectrum model. The simulation results showed that the normalized throughput of second users was improved significantly by using the spectrum prediction method.3. The improved method of the cooperative spectrum prediction was proposed based on the coalition game theory. To solve the problem caused by the threshold of false alarm probability and the merge-split algorithm principle in coalition formation algorithm which was proposed by W. Saad,the maximum value of second user number and the individual adjustment of coalition member were introduced in this study. By comparing improved cooperative method, merge-split method and local method in spectrum prediction, the rationality and validity of the improved algorithm were verified. The simulation results showed that the cooperative spectrum prediction method improved in this thesis effectively increased the prediction accuracy.