Research On Cognitive Networks Intelligent Decision Oriented To Resources Allocation

Traditional networks existing many problems such as low resources utilization ratio, withoutservice QoS assurance and low users satisfaction, Cognitive Networks （CNs） provide theopportunity to solve the above problems. CNs are intelligent networks, which can aware thewhole network situation, make plan and decision independently through reasoning study, and thenimplement the corresponding action, meanwhile continuously improve the service strategies byfeedback control, so as to ensure the end-to-end quality of service.This paper mainly focus on theresources allocation, according to the results of each service flow prediction based on theimproved grey prediction model. The main research content is summarized as follows.Firstly, this paper proposed an intelligent decision-making structure for resources allocation,and introduced the function and characteristics of each component in the architecture. Using thedecision-making structure to solve resources problems in CNs, based on the prediction of the nextmoment service flow state and the experience learned by source nodes dynamically regulate thestrategies of resource allocation, queue scheduling, congestion control.Secondly, this paper proposed a service flow prediction method based on improved greyprediction model. It preprocessed and properly clustered those data into the available data used bythe improved grey prediction model thus obtaining the flow development trend in continuous timeslice of cognitive networks, provided data for the resource allocation and scheduling.Thirdly, an intelligent resource configuration method based on service flow prediction inCNs proposed by this paper. Considering service flow prediction data, we allocate traffic flows ofresources adaptively, and sent the affection of the configuration strategies to source end throughthe feedback learning mechanism based on Q-learning algorithm, then dynamically regulatebuffer resource utilization. This method also can reduce routing computation burden, as when asimilar service comes, which can directly use the above routing strategies. On the basis of therelated simulation analysis we can see that the intelligent decision-making method not only canimprove the network performance and the users satisfaction.Finally, conclusions are made based on above work and the future work to the existentproblem is pointed out.