Cognitive Radio Dynamic Intelligent Spectrum Access Algorithm Baesed On Markov Model

In the recent years, with th e rapid development of wireless broadband technology, the scarce spectrum resources are becoming a bottleneck and increasingly constraining the improvement of wireless communications systems. Cognitive radio(CR)technique is considered as one of the solutions of current spectrum resource scarcity.In order to improve the spectrum efficiency, the core idea of cognitive radio is to exploit the licensed but unused spectrum and carry through dynamic spectrum access.Based on avoiding interference to primary users, dynamic spectrum access plays a key role in cognitive radio. This thesis is mainly about cognitive radio dynamic spectrum access based on the theory of Partially Observable Markov Decision Process Model.First, the defination of cognitive radio is elaborated and discuss the research of cognitive radio at home and abroad. Machine learning in the Artificial Intelligence is defined and classified.Then, an approach of Markov Decision Process(MDP) and Partially Observable Markov Decision Process(POMDP) about intelligent dynamic spetrum access is illustrated.Meanwhile three solutions to the MDP is discussed and two methods of solving POMDP is also represented.In third chapter, the POMDP algorithm based on accurate model is adopted to access spetrum holes dynamically. For the secondary throughput per slot, the paper proposes an error-free perception using the optimal algorithm to access spectrum intelligently. Meanwhile the sub-optimal algorithm is presented based on the complicated environment model. The comparison between two algorithms is given by the simulation results. At last, frequent changes among primary users in the channel states are also illustrated by the simulation results.In the last chapter, the intelligent spectrum access based on non-accurate model is discussed. This chapter is divided into two parts. First, when sensing errors exists, how the physical layer and MAC layer of the joint spectrum sensing is given. Second, when the secondary users do not have a priori knowledge of the environment, how to apply statistical learning methods to build the environment model is illustrated. And then dynamic spectrum Intelligent Access is complished using POMDP algorithm too.