Research On Dynamic Clustering Algorithm For 5G Ultra-Dense Network

With the rapid development of mobile communication technology in recent years,the large-scale commercialization of 5G communication systems has become more urgent.Currently,as the key technology of 5G,Ultra-Dense Network(UDN)has been proven to effectively extend coverage in signal blind areas and offload traffic in hot spots.However,with the high transmission capacity provided by UDN,the problem of signal interference has become more serious.At the same time,the anti-interference research on 5G UDN has focused on clustering and cooperative processing,but most of the proposed schemes are just reexploring the clustering strategies from traditional cellular networks to ultradense networks,which is not well suitable and compatible with 5G ultra-dense networks.Therefore,the study of clustering strategies for 5G ultra-dense networks is of great significance.Based on the Coordinated Multiple Point(Co MP)technology,in this paper,the receiving model in ultra-dense networks is derived and the dynamic clustering techniques are studied.For the first time,Density-Based Spatial Clustering of Applications with Noise(DBSCAN)clustering algorithm is introduced into the clustering strategy,and a network-centric clustering strategy is provided for signal blind areas and general hot spots.At the same time,based on the theory of multi-objective optimization,use-centric clustering strategy is investigated on hotspots with heavy traffic load.The research contents of this thesis include the following:(1)The receiving model of the current 5G ultra-dense network is analyzed.Based on the analysis of deployment principles and interference components within 5G ultradense networks,the transmission rates in both non-clustered and clustered cases are derived,which provide theoretical basis and evaluation indicators for subsequent research.(2)The classical clustering strategy is studied.Based on the design principles,a dynamic clustering strategy based on the improved K-means algorithm and a user-centric clustering strategy based on interference channels are investigated.The advantages and disadvantages of the two classical clustering strategies are mainly studied.(3)Aiming at the signal blind areas and general hot spots,the DBSCAN clustering algorithm is introduced into the clustering strategy,and a network-centric clustering strategy is proposed to improve the system transmission rate.The principle and process of DBSCAN algorithm are explained first.Secondly,for the scenario of a fixed number of cooperative clusters,combining the DBSCAN algorithm and the particle swarm optimization algorithm,K-DBSCAN(K-limited DBSCAN Strategy)clustering strategy is proposed.Then for the scenario of limiting the number of base stations in the cluster,the DBSCAN algorithm is improved,and P-DBSCAN(P-limited DBSCAN Strategy)clustering strategy is proposed.Finally,the two strategies are simulated separately.Compared with the classic clustering strategy based on the improved K-means algorithm,the two strategies proposed in this section can increase the system transmission rate by about 10% in their respective scenarios.(4)To improve the communications in hotspots with heavy traffic load,based on the multi-objective optimization theory,user-centric clustering strategies are proposed to reduce the number of outage users.First,a multi-objective optimization equation is established for the purpose to minimize the scale of cluster size,and a user-centric clustering strategy with limited cluster size is given.Then,for the scenario with limited cooperative distance between base stations,the multi-objective optimization equation is improved,and a user-centric clustering strategy with limited cooperative distance between LPBSs is proposed.Finally,compared with the classic user-centric clustering strategy based on interference channels,simulations have been conducted to verify that the two strategies proposed in this section can reduce the number of outage users by about 30% and the collaborative resources consumed by about 10%.