Research On Resource Management In Transformium Network

In the past several decades,mobile communication system experienced full-scale development.Network scale expanded continuously and system performance obtained a big promotion.Moreover,the service types provided by the system also become diver-sified.The initial mobile communication system only provide voice service.However,in the current system,multimedia services like data,video and other services like on-line game,telemedicine all are supported.In the near future,with the development of machine type communication(MTC),mobile communication can also be applied to automated driving,factory automation controlling,sensor data collecting,etc.Ac-cordingly,the associated terminals of mobile communication system,which contains not only smartphone,tablet but also various MTC terminals,also become diversified.However,with the high development of mobile communication system,it also faces a lot of challenges and problems.Specifically,we summarize them into three points:(1)functional diversity increases complexity of the network architecture which leads to the lack of flexibility;(2)network scale becomes large which result in high energy consumption.In addition,network cannot dynamic adjust its state which leads to low energy efficiency;(3)as the user demand is increasing and the service types become diversified,we need to improve the system capacity and devise multi-service resource allocation scheme.In this thesis,we focus on solving the problems described above and thus improve the flexibility,energy efficiency,spectrum efficiency of the system.First,as the net-work service rely on the underlying proprietary devices,the mobile communication network becomes static and difficult to update and change.Therefore,we propose a new network architecture called transformium network which can transform into other forms flexibly.This architecture can dynamically optimize the network and solve the network ossification problem.Second,as current network cannot dynamically adjust its state according to environmental changes,we devise the environment sensing based dy-namic adjustment scheme of base station(BS).This scheme realizes the environmental changes based transformation of Radio Remote Unit(RRU)and improve the energy ef-ficiency of mobile network.Third,aiming at the flaw of multi-service resource manage-ment in existing works,we devise a wireless network virtualization based multi-service resource allocation scheme.This scheme realizes the multi-service based transforma-tion of building base band Unit(BBU)and improve the spectrum efficiency of mobile network.We integrate multiple modules together organically to address the challenges of environment changes and multi-service.In the first work of this thesis,as existing architecture devising works only solve part of problems in current network,we devise a comprehensive network architecture to solve all the problems.We first investigate some emerging technologies and net-work architecture related works.Then,we devise the control core based on software defined network(SDN),cloud-radio access network(C-RAN)and network virtualiza-tion(NV)technologies.Environment sensing module is also introduced to collecting and processing environmental information.Specifically,transformium network archi-tecture consists of transformium core,transformium body and transformium cloud.We expound the functions of these parts,the relationships among them and the operation logic of the whole network.Moreover,we describe a example in which the transformiun network architecture is applied to present the concept of the proposed architecture bet-ter.We also analyse the problems and challenges need to be addressed to realize the deployment of transformium network.The existing energy efficiency management works have the following problems:(1)the scenario considered is relatively simple;(2)some emerging technologies are not considered;(3)not all the adjustment schemes,i.e.,BS sleeping,antenna switch and power control are considered;(4)they just consider the traffic fluctuations without other environment changes.Moreover,traffic volume is not obtained in real-time but as-sumed in advance.Therefore,in the second work of this thesis,we study the environ-ment sensing based dynamic management scheme of BS to address this disadvantages.In the transformium network architecture,we consider a multiple-input multiple-output based(MIMO-based)heterogeneous C-RAN and environment information can be ob-tained through environment sensing module.The BS management problem is formu-lated as a mixed interger programming which is NP-hard.As a result,we devise a heuristic algorithm to obtain a sub-optimal solution.In this algorithm,we jointly con-sider BS sleeping,antenna switch and power control to dynamically adjust the network according to the environmental changes.Simulation results show that the proposed al-gorithm outperforms other baseline algorithms in system energy consumption.Besides,the impact of some parameter settings,BS sleeping strategy and the impact of rainfall are also analyzed.This work realizes the RRU transformation based on environmental changes.Moreover,it improves the energy effciency and flexibility of the network.Traditional resource scheduling algorithms are not suitable for MTC services which are one of the fundamental parts of future networks.Moreover,in the exist virtualized resource allocation related works,how to satisfy the various Quality od service(QoS)requirements of different services is not considered.Therefore,in the third work of this thesis,we study the multi-service resource allocation scheme in the network with wire-less virtualization.We first propose a service-centric wireless virtualization model to slice network according to service types.Based on this model,we devise a virtual BS model in which the multi-service resource allocation problem is decoupled into inter-slice resource allocation and intra-slice resource scheduling problem.We formulate the inter-slice resource allocation as a discrete optimization problem and propose a heuristic algorithm to get sub-optimal solution of this NP-hard problem.In addition,we devise several scheduling algorithms based on existing ones for scheduling users of several specific services to solve the resource scheduling problem.Numerical results show the superiority of the proposed scheduling algorithms over the existing ones when applied to schedule specific services.Moreover,proposed resource allocation scheme is veri-fied to meet the properties of virtualization and solves the multi-service resource alloca-tion problem well.This work realizes the BBU transformation based on multi-service.Moreover,it improves the spectral effciency and flexibility of the network.