Research On Sleep Awareness Resource Allocation Based On TDM-PON And C-RAN Architecture

The rapid growth of data services and the proliferation of smart devices have contributed to the evolution of wireless communications,but brought many new challenges at the same time.The convergence network based on Passive Optical Network(PON)and Cloud Radio Access Network(C-RAN)combines the advantages of optical networks and wireless networks to meet the growing demand of mobile wireless network.The C-RAN geographically separates the baseband processing unit and the front-end wireless transmitting unit in the traditional base station,and connects the two parts through the PON.Therefore,an architecture in which a distributed radio remote head is combined with a centralized baseband processing unit is formed.In this context,this thesis starts from the overall architecture of C-RAN and studies the related issues of resource management and device dormancy for the purpose of improving network energy efficiency.Finally,the corresponding strategies are proposed.Firstly,this thesis introduces PON and C-RAN network in detail,including research background,network structure,technical characteristics,etc.Then,it introduces the research hotspots from the aspects of content caching,collaborative resource management,and network energy saving.Finally,the focus of this thesis is drawn by introducing the fronthaul resource management strategy of C-RAN network.Secondly,a C-RAN energy saving mechanism based on load aggregation is proposed.Considering user quality of service(QoS)and network resources,a network optimization model is established.Based on the independence of BBU and RRH,the original problem is decomposed into two sub-problems: BBU resource allocation problem and RRH resource allocation problem.And then,a heuristic algorithm is used to solve network resource allocation problem.The results show that the proposed mechanism can effectively improve the utilization of network resources and the energy efficiency of the network while ensuring the QoS of user.Thirdly,a resource allocation mechanism based on machine learning for hybrid power supply C-RAN is proposed.According to the characteristics of energy collected by the network,a regression analysis model is established.And,the energy arrival rate at different times are calculated by using the model.Based on the time-varying and energy arrival rate of the wireless channel,the network and the environment are interactively learned through reinforcement learning,and the optimal allocation of network resources is realized.The results show that the proposed mechanism guarantees the QoS of user and improves resource utilization and energy efficiency of network.Finally,the main work content and innovation points of this thesis are summarized,and the objectives of the follow-up research work are also proposed.