Handover Optimization In 5G Network Based On User Mobility Data Analysis

5G is considered as a promising field among the research of mobile communication.Cloud radio access network(C-RAN),a new type of access network architecture in 5G network,aims to reduce the cost of hardware devices and improve the flexibility of network deployment.The baseband processing unit(BBU)responsible for receiving and transmitting signals and the remote radio head(RRH)responsible for processing signals of each base station are coupled to each other in conventional radio access network,while the base station in C-RAN only retains RRH,and BBUs are aggregated in the baseband processing unit pool at the cloud.Each BBU takes control of several RRHs.How to accomplish RRH grouping and BBU mapping to optimize BBU-RRH resource allocation has become a research hotspot in C-RAN.Most of the existing BBU-RRH resource allocation studies manage to decrease numbers of BBU,enhance transmission rate or economize on power assumption of the network from the perspective where only the network is concerned,ignoring the quality of users' experience.In fact,handovers of base station in users' moving process have a significant influence on the quality of users' experience.Whereas there is not enough work concentrating on the optimization,not to mention the involvement of mobility of real users.In this paper,we analyze the mobility characteristics of users with real communication data collected from mobile phones.Based on this,we analyze the BBU-RRH resource allocation problem aiming at reducing handovers and improving the quality of users'experience.The content and contributions of this paper are as follows.(1)Based on the mobile phone data of users collected from a mobile operator,we analyze the mobile mode of users from the following four aspects.First of all,we analyze the active period of mobile phone users determining by the number of users and traffic volume in weekdays and weekends respectively;Secondly,we analyze users' moving range based on the identification of users' address.Meanwhile,the distribution characteristics of the moving range are analyzed.In addition,we analyze users' moving frequency.It is found that the distribution of users' moving frequency is subject to the exponential distribution,and the moving frequency in weekdays is higher than that in weekends.Last but not least,the concept of transition probability in Markov chain is introduced into users' mobility analysis.And the transition probability of users is found to be non-uniform,where different base stations provide service.(2)Based on the mobility characteristics of real users,we optimize the BBU-RRH allocation scheme in C-RAN to reduce the number of handovers.In the existing research,users' mobility characteristics are simulated with random walk,which is quite different from the actual situation.Moreover,in the existing research the network established only considers the geographical proximity among RRHs,which is not sufficient to reduce the number of handovers.In response to these problems,we propose a dynamic spatial-temporal RRH network model that takes both temporal and spatial proximity into account using mobility data of real users.On this basis,RRH community discovery and BBU allocation are implemented.Experiments show that compared with traditional access network,the number of handovers in our network model is reduced by 39%,and the number of BBUs is reduced by 59%.Compared with the existing scheme which only takes spatial proximity into account,the number of handovers in the network is reduced by 10.29%.In summary,this paper analyze characteristics of users' moving pattern based on users' communication data,and after that we provide a solution towards the lack of consideration of users' mobility,which is an improvement for existing C-RAN research.Furthermore,the work can also be applied to deal with the allocation of BBU resources under C-RAN architecture,which actually manifests obvious practical value.