Research On Traffic Offloading Technology In Heterogeneous Cellular Network

In order to meet the huge demand for network capacity of the explosive growth of mobile communication services,heterogeneous cellular networks appeared.To improve the throughput and energy efficiency of the network,The heterogeneous cellular network superimposes several types of small power base stations on the macro base stations,such as pico base stations,femto base stations,relays and etc.However,deploying a large number of small power base stations in a heterogeneous cellular network will also make the network structure and interference environment very complicated.Aiming at the downlink system of a heterogeneous cellular network,this thesis comprehensively considers the minimum rate requirements of different users,interference between different base stations and available system resources and etc.,and proposes Two kinds of traffic offloading algorithms from the perspectives of maximizing user utility and system energy priority.Aiming at the differences of different type of services in heterogeneous cellular networks and the data rate requirements of each user,This thesis start from increasing the throughput of the network,and propose a traffic offloading algorithm with the goal of maximizing user utility.This algorithm decouples the traffic offloading problem into the problem of access selection between the base station and the user and the power control problem of the base station.Aiming at the problem of access selection,a Qo S(Qualify of Service)-aware user association algorithm is proposed in consideration of the user's minimum rate requirement and service priority.Aiming at the problem of power control,a power control algorithm based on symbol planning is proposed by approximating the optimization model to a symbol planning problem.The simulation results show that the algorithm proposed in this thesis can obtain higher system utility while meeting the user's Qo S requirements.Aiming at the problem of high power consumption of each base station in the heterogeneous cellular network,a joint optimization model of access selection and power control with the goal of minimizing the total power consumption of the system is established by using the advantages of deep reinforcement learning in solving complex nonlinear optimization problems.Based on the DDPG(Deep Deterministic Policy Gradient)method in the deep reinforcement learning method,an energy-first intelligent traffic offloading algorithm is proposed.The simulation results show that,compared with the traditional optimization algorithm,the algorithm proposed in this thesis obtains lower system power consumption,improves the energy efficiency of the system,and meets the minimum speed requirements of users.