Research On Cognitive Radio's Decision Maker Based On Boosting Algorithm And Its Applications

In 1999, Joseph Mitola III first mentioned the terminology, Cognitive Radio; and just 4 years later, the Federal Communications Commission (FCC) of the US government issued to authorize the techniques of cognitive radio for spectrum sharing/reusing and approved the applications of fixed and mobile services in TV bands in December 2004. Cognitive radio has already been a new technology hot-spot. It might cause the next climax in wireless communications.However, there are still tremendous challenges in getting an intelligent communication system. The Decision Maker of Cognitive Radio is one of the key issues, and it is also the hot-spot in academia. There are various solutions, but all of them still have some limitation. In the thesis, a new decision maker model of Cognitive Radio which based on Boosting is proposed. This new decision maker model is called new AdaBoost.M1-RF algorithm which improves the performance of Adaboost.M1 is proposed. The final soft system for bridge monitor system is implemented by CR based on AdaBoost.M1-RF, and the evaluation results confirm the effectiveness of it. The main work to complete and innovation are as follows:1. After a broad study on current research of CR, for the verifying and analysis of the model's effectiveness and feasibilities, different cases are studied. The limitations of existing models, which based on Game Theory, GA and NN, are analyzed.2. Proposing a new CR's decision maker model based on Boosting algorithm and a new solution of Cognitive Radio.3. A new approach of AdaBoost.M1-RF algorithm is proposed. In connection with the problems of Adaboost.M1 has worse performance on the multi-classes, Random Forest is computationally effective and offer good prediction performance. The AdaBoost.M1-RF frame, which using Random Forest as weak learner is proposed. To evaluate the performance of AdaBoost.M1-RF algorithm, the algorithm is compared with other machine learning algorithms, such as BPNN, SMO, RBF Network, Bagging, Random Forest, Decisionstump in AdaBoost.M1, ADTree in AdaBoost.M1 J48 in AdaBoost.M1, etc, in different experiments.4. To resolve the issue of the communication in bridge monitor system, SMS,GPRS and CDMA technologies are analyzed and network structure of bridge monitor system is designed. A new solution for communication of bridge monitor system, using CR based on AdaBoost.M1-RF algorithm, is proposed. The information model of CR's Decision Maker is designed. With actual application data of bridge monitor system, the CR decision maker based AdaBoost.M1-RF algorithm is evaluated. The model is used to program a soft for bridge monitor system communication and it reduces the cost of Bridge Health Monitoring System and improves reliability of the communication in monitor system, meanwhile it also extends the scopes of Cognitive Radio's application.