Research On Dynamic Resource Management Algorithms For Spectrum Sharing Networks

The emergence of various new wireless services has resulted in the explosive growth of data traffic for wireless communication,facilitating the development of The Fifth Generation Mobile Communication System(5G).Due to the scarcity of spectrum resource,spectrum sharing becomes one of the key technologies of 5G,which can meet the performance indicators of 5G,such as high capacity and high spectrum utilization.However,there are many problems about radio resource management in the typical ultra-dense 5G application scenario,such as the interference problem and the problem of effective channel allocation.The application of spectrum sharing network still has many challenges.Therefore,this paper has carried out a series of studies on the dynamic resource management of spectrum sharing networks.For the aggregate interference problem caused by multiple secondary users(SUs)to the primary user(PU)in the spectrum sharing network,in order to protect the primary user,an aggregate interference prediction algorithm based on the neural network is proposed in this paper.This algorithm is applicable to complex wireless network environments,and can model the functional relationship between the users'state parameter information and the aggregate interference by training historical data without knowing the wireless environment parameters(such as the path loss coefficient).Furthermore,key factors affecting the prediction accuracy of aggregate interference are analyzed in this paper,including the number of input parameters in the input layer,the representation of input parameters,and the number of nodes in the hidden layer.For the spectrum resource allocation problem among secondary users in the spectrum sharing network,firstly,a primary channel allocation scheme based on the interference overlap graph is proposed.In this scheme,the actual state of end user devices(EUDs)is considered in the process of constructing the interference overlap graph.And in order to ensure efficiency of the primary channel allocation,the method of adjusting the edges in the interference overlap graph is proposed,which is aimed at minimizing the variation of the interference to the whole network.In this edge adjusting method,four new parameters are proposed as the basis for edge adjusting,which are the signal to interference plus noise ratio(SINR)of EUDs and the margin of SINR in the overlapping area between vertices,the number or density of EUDs in the overlapping area between vertices,the number of radio access base stations(BSs)in the vertex,and the proportion of the affected BSs in the vertex.In addition,for the spectrum resource allocation problem among secondary users,this paper also proposes a machine learning based extended channel allocation algorithm,which is suitable for the scenario where the load of BSs changes dynamically.The Multi-Armed Bandits(MAB)algorithm can be executed by the BS for bandwidth expansion according to spectrum requirement,which can reduce the measurement overhead of the BS and the information interaction overhead between the systems.In this paper,the conflict adjustment factor is also introduced in the MAB algorithm to reduce the conflict probability of the multiple extended channel requests.Finally,the MATLAB simulation tool is used in this paper to verify the feasibility and efficiency of the proposed solutions for the spectrum shar:ing network.The performance indicators in the proposed algorithms are analyzed in depth using various simulation results.