Research On Base-station Energy Saving And User Computation Offloading Strategy In Heterogeneous Cellular Network

With the diversified development of Internet of Things applications,the number of mobile terminals and communication services continue to rise,and people have put forward higher requirements for data transmission rate and service quality.Network communication requires more abundant spectrum resources and higher spectrum efficiency.The heterogeneous cellular network(HCN)formed by the combination of 5G ultra-dense micro base station and heterogeneous network deployment effectively improves the regional spectrum efficiency and expands the network capacity,but it also brings the problems of extremely unbalanced load of base station and low energy utilization.In addition,the diversity of communication services of cellular terminal users in HCN brings about larger computing tasks and higher latency.Mobile edge computing(MEC)technology can effectively meet the requirements of low latency and large broadband services.In fact,the coverage of the MEC server is limited.If the user moves to a new location and cannot receive the calculation result in time,the user will repeatedly request calculation offloading,resulting in a waste of computing resources.Meanwhile,each user is also self-interested that causes excessive consumption of computing resources in the process of preempting resources.Therefore,how to balance the load of the base stations in the HCN to improve the energy utilization rate of the base stations,and how to reduce the mobile client computing task offloading delay to realize the efficient allocation of server computing resources are worth further study.Based on the above considerations,the main contents of this dissertation can be summarized as follows:First,the relevant theoretical basis of HCN is summarized,including HCN architecture,base station energy saving technology and MEC computing task offloading strategy.Second,aiming at the problems of uneven load distribution and low utilization rate of renewable energy caused by intensive deployment of small cell base stations in HCN,an energy cost optimization scheme based on social D2 D communication and energy adaptive sensing is proposed to minimize the energy cost of base stations.Considering the social relationship between users in real scenarios,this strategy first enables users who meet the D2 D communication requirements to establish D2 D communication connections.Next,combining the temporal and spatial diversity of renewable energy sources,the remaining cellular users are associated with renewable energy adaptively aware base stations.Finally,the ratio coefficient of bandwidth allocation is introduced,and the total transmitting power of each base station is optimized by Lagrange duality method to minimize the total energy cost of the system base station.The numerical simulation results verify that the algorithm can effectively reduce the energy consumption of the power grid.Compared with the maximum reference signal receiving power algorithm and the base station preference bias factor algorithm,the proposed algorithm reduces the total energy cost of the system base station by 54.8% and 25.8%,respectively.Third,in multi-user MEC system,the mobility of users during task offloading brings about waste of computing resources and increased computing delay,a software-defined network(SDN)assisted MEC network architecture is constructed.The SDN controller is introduced to collect the information of mobile users and base stations in the macro base station from the global perspective of the network state,and optimize the network configuration as needed,thereby improving the efficiency and flexibility of computing resource management in the HCN.A computing task offloading scheme with user location prediction and cognitive learning is proposed to further reduce the delay.In the scheme,the extended Kalman filter is firstly employed to predict user mobility and estimate connectivity.And then,according to the predicted location information,reinforcement learning algorithm is adopted to dynamic strategies selection in the process of task request and task collection in order to minimize the delay and save computing resources.Numerical simulation results show that the proposed scheme has lower duration of task than that of other existing schemes.