| As one of the important networking structure for the fifth generation mo-bile network,small cell networks(SCNs)have the advantages of low cost,com-prehensive coverage,and spectrum and energy efficiency improvement.How-ever,with the explosive growth of network data traffic and the rapid spread of new applications such as virtual reality and augmented reality,SCNs are challenged by the problems of limited backhaul links,insufficient computing capability and high content transmission delay.To encounter with these chal-lenges,the integration of caching and edge computing technologies with SCNs is considered.And this integration can highly enhance the storage and comput-ing capabilities for the network,which has received extensive attention from academia and industry.In the SCNs with caching and edge computing technologies,the network resources contain the traditional wireless resources as well as storage resources and computing resources.The heterogeneous resources and the different per-formance metrics make resource management problem more complicated.It is well known that reasonable and efficient resource management solutions can not only improve network performance,but also support various new ser-vices and applications and improve users’ quality of services(QoS).So,how to jointly manage and optimize heterogeneous resources,such as wireless re-sources,caching and computing resources in the SCNs is worthy of further study.Therefore,we study the heterogeneous resource management problem in the SCNs from three aspects:jointly considering caching and wireless re-sources,jointly considering computing and wireless resources as well as jointly considering computing,caching and wireless resources.Accordingly,three in-novative resource management schemes are proposed including a content pair-ing and resource allocation mechanism for capacity enhancement,a two-stage computing offloading and resource allocation scheme for marco-small cell base stations(SBSs)as well as a distributed heterogeneous resource allocation strat-egy based on network virtualization.Moreover,the performance of the pro-posed schemes is evaluated by simulation.The main contributions and innova-tion points of this dissertation are concluded as follows:1.To maximize the network throughput and save the backhaul resource while ensuring users’ data rate requirements,a content pairing and resource al-location mechanism for capacity enhancement is proposed where the caching and full-duplex self-backhaul between SBSs are considered.Firstly,by considering the caching and the full-duplex self-backhaul tech-nologies between SBSs,the content pairing and resource allocation problem is studied to maximize the system throughput under the guarantee of the user’s QoS requirements.Secondly,in order to solve the problem effectively,a con-tent pairing mechanism based on Kuhn-Munkres algorithm is proposed accord-ing to the user requests and cached contents.Furthermore,in order to reduce the computational complexity,the spectrum and power allocation problem is decoupled,a power allocation method based on the successive convex approxi-mation(SCA)and constrained concave convex procedure(CCCP)is proposed.And the spectrum allocation is further optimized based on the power allocation results.Moreover,an iterative algorithm is proposed to obtain a sub-optimal spectrum and power allocation strategy.Finally,the simulation shows that our proposed mechanism can make full use of the caching resource and full-duplex self-backhaul technology,save the backhaul resource and effectively improve the network throughput with low complexity.2.To minimize the system energy consumption while guaranteeing the de-lay requirements of users,a two-stage computing offloading and resource allo-cation scheme for macro-small cell base stations is proposed where the various computing requirements of users and different computing capabilities between macro base station(MBS)and SBS are taken into account.Firstly,by introducing mobile edge computing(MEC)and full-duplex self-backhaul technologies,a network model with two-stage computing offloading among MB S and SBS is constructed.Secondly,under the guarantee of the delay of users,the two-stage computing offloading and resource allocation problem is modeled as a mixed integer non-convex optimization problem to minimize the system energy consumption.Thirdly,in order to reduce the computational complexity,the formulated problem is divided into two subproblems.For the first subproblem,the j oint computing offloading and power allocation problem,as a fractional optimization problem,is solved by the parametric convex opti-mization method.For the second subproblem,the spectrum allocation strategy is optimized by proving the convexity of the subproblem.Based on the two subproblems,an iterative optimization algorithm is proposed to obtain a sub-optimal solution to the computing offloading and resource allocation problem.Finally,the simulation shows that our proposed scheme can make full use of the computing resource between different base stations according to the require-ments of users,optimize user offloading decisions and resource allocation,re-duce the system energy consumption and save the backhaul resource with low complexity.3.To optimize the system benefit while providing the high-data-rate ser-vices in the downlink and the computing-sensitive services in the uplink,a dis-tributed heterogeneous resource allocation strategy based on network virtual-ization is proposed.Firstly,the network virtualization technology is introduced to make all physical infrastructure and wireless,caching and computing resources into vir-tual resources.And a virtual resource management model is established for the high-data-rate services and the computing-sensitive services where the virtual resources can be allocated dynamically.Secondly,by pricing the revenue of dif-ferent services and the corresponding resource consumption,the system benefit is defined as the objective function.And the user association,computing of-floading decisions,caching decisions,and virtual resource allocation problem are modeled as a mixed integer non-convex optimization problem.Thirdly,in order to solve the problem efficiently,variables substitution is conducted to transform the original problem irnto a convex optimization problem.Then a distributed virtual resource allocation algorithm based on alternating direc-tion method of multiplier(ADMM)is proposed.Finally,the simulation shows that our proposed strategy can optimize the resource allocation according to the users’requirements,and improve the system benefit with low complexity. |
