Performance Analysis Of Clustering In Cloud Radio Access Network Based On Stochastic Geometry

With the proliferation of laptops,smartphones and smart wearables,wireless data traffic has experienced an unprecedented explosive growth.Cloud Radio Access Network(C-RAN),characterized by centralized processing,cooperative radios,and real-time cloud architectures,has become one of the most promising network architectures in the 5th Generation Mobile Networks(5G).Based on stochastic geometry,this thesis studies the network modeling and performance analysis of different clustering schemes in C-RAN under the composite fading channel.For the network-centric clustering network,a theoretical analysis method under the composite fading channel is proposed,focusing on the outage probability as a performance index to study the influence of channel propagation characteristics on network performance and cluster size.First,the locations of the Remote Radio Heads(RRHs)and the users in the cluster are modeled as the Binomial Point Process(BPP)and the Homogeneous Poisson Point Process(HPPP)respectively,and the channel is modeled as a composite fading channel that experiences path loss,Rayleigh fading,and shadowing simultaneously.Second,the closed expression of the Laplace transform of the user received signal power is given by Gauss-Hermite integration,and on this basis,the expressions of the outage probability when the user is associated with a single RRH and multiple RRHs are deduced.Finally,the accuracy of the theoretical derivation is verified by simulation,and the influence of network parameters,channel parameters and user locations on outage probability under the above two association schemes are analyzed.Because the analysis method of the network-centric clustering is not suitable for the user-centric clustering.Therefore,this thesis further studies the user-centric clustering based on stochastic geometry,and proposes a theoretical analysis method under the composite fading channel.First,the locations of RRHs and users using the same resource blocks are modeled as the HPPP and the Matérn Hard-Core Point Process of type II(MHCPP),respectively.Second,the difficult-to-analyze MHCPP is approximated to the easy-to-analyze HPPP by the point process approximation.Based on the approximated model,the distance distribution function of node positions in the network is deduced,and on this basis,the expression of outage probability of the user-centric clustering is given.Finally,the validity of the point process approximation and the accuracy of the theoretical derivation are proved by simulation,and the effects of network parameters,channel parameters and void cells on network performance are analyzed.