Research On Base Station Sleeping Strategy In Heterogeneous Cellular Networks

With the increasingly growing volume of mobile traffic nowadays due to date explosion,telecommunication devices' capacity is facing enormous challenges.Dense heterogeneous cellular network,as important part of the fifth generation(5G)telecommunication technology,is an important way to solve the above problem.However,with the development of telecommunication system and the increasing scale of heterogeneous network,the energy efficiency in the network is declining seriously.Researching on how to improve the system's energy efficiency and increase the spectrum efficiency of the system is urgent and imperial.In heterogeneous cellular networks,base station dormancy technology had been widely used to improve system energy efficiency and spectrum efficiency due to it is easy to be implemented and low cost.Nowadays,there are still some problems in the academic filed of base station dormancy technology in heterogeneous cellular networks,such as the problem of how to reduce the interference between base stations,problems of how to scientifically analyze the dynamic traffic of base stations and how to reduce the interruption probability between base stations and users.Therefore,this thesis had proposed two different base station dormancy strategies to improve the system's energy efficiency and spectrum efficiency in the two-tier heterogeneous cellular network environment composed of Macro-Femto.Firstly,in order to improve the energy efficiency of the system and reduce the interference between the base stations,a novel dormancy strategy for heterogeneous cellular network clustering SMDP base stations is proposed and a clustering algorithm is proposed at the beginning.According to the different interference values between the home base stations,the home base stations with larger interference values are divided into the same cluster,and then the SMDP dormancy strategy is applied to each family base station within the cluster.Then,the switching problem of home base stations in heterogeneous cellular networks is modeled by using version Markov decision process.We define the traffic and working status of the home base station as the state space of the system,the switching action of the home base station as the action space,and the return function as the energy efficiency of the system.Then we derive the long-term energy efficiency benefit model of the system to improve the energy efficiency of the system.Finally,the optimal dormancy strategy is obtained by the proposed hybrid genetic particle swarm optimization.The simulation results verified that the dormancy strategy reduces the interference between the home base stations and improves the energy efficiency of the system effectively.Secondly,in order to improve the spectral efficiency of the system and reduce the probability of interruption between the home base station and the user,machine learning technology is introduced into the dormancy strategy of the home base station.Using the wavelet neural network algorithm,the home base station in each cluster is trained,and the traffic volume of the home base station in one week is predicted,and the reasonable dormancy strategy of the home base station is formulated through the predicted value.At the same time,considering the interruption probability of users and base stations,a user re-association algorithm is designed to optimize the quality of service for users in home base stations.The simulation results verified that the base station dormancy strategy based on wavelet neural network were capable of not only reducing the probability of user interruption,but also effectively improved the spectral efficiency of the system.