Theories And Methods Of Resource Allocation In Non-Orthogonal Multiple A Ccess Enabled Fog Radio Access Networks

Integrating non-orthogonal multiple access(NOMA)technology into the physical layer of fog radio access network(F-RAN)can increase the number of admitted user equiment(UE),enhance capacity,and reduce the average service delay of UEs.However,in order to realize these performance goals,it is essential to overcome the critical issue of efficient multi-dimensional resource management in non-orthogonal multiple access enabled fog radio access network(NOMA-FRAN).This problem involves the following challenges:(1)NOMA needs to adopt successive interference cancellation technology,which has a large computational overhead.The quantitative impact of computational complexity cost on system performance is not yet clear.Meanwhile,NOMA adopts a non-orthogonal resource multiplexing method,which introduces cochannel interference,leading to complex interference relationships between users.Existing resource management techniques do not consider the tradeoff between spectral efficiency and computational overhead in NOMA-FRAN.(2)In NOMA-FRAN,multiple services coexist,and different users have diverse quality of service requirements,leading to resource contention that degrades network performance.Therefore,it is necessary to design corresponding resource management mechanisms to achieve a balance between meeting the performance requirements of different services and achieving performance equilibrium.(3)In NOMA-FRAN,communication,caching and computing resources are coupled in multiple dimensions.The number of users and access nodes is large,and their capabilities vary.NOMA uses power-domain resource multiplexing to further increase the dimension and complexity of resource management.Existing resource management approaches have not considered joint allocation and optimization of multi-dimensional resources in NOMA-FRAN.In view of the above problems and challenges,the main research content and innovation points of this dissertation are summarized as follows:1.To address the challenge of balancing spectrum efficiency and computational overhead in NOMA-FRAN,this dissertation first proposes a network utility function metric weighted by spectral efficiency and computational complexity cost.With this metric,a comprehensive evaluation of the impact of different multiple access schemes on the performance of F-RAN is achieved.Meanwhile,taking into account practical constraints such as decoding order,power allocation,and quality of service,an optimization problem is constructed for maximizing network utility.To solve this non-convex problem,a method is proposed to decouple the original problem into user association and subchannel allocation subproblems,and a power allocation subproblem,and then iteratively solve them using algorithms based on matching game theory and sequential convex programming,respectively.Simulation results show that,compared to F-RAN based on orthogonal multiple access,the proposed method can improve network utility by nearly two times.2.To address the challenges of multi-service coexistence and differentiated requirements in NOMA-FRAN,we consider a scenario where high-capacity and high-connectivity services coexist in the downlink network.We design network modes corresponding to different services and propose a joint resource allocation and admission control scheme that maximizes the number of admitted UEs.Specifically,for high-connectivity services,the NOMA access mode of edge nodes is designed,and the wireless resource allocation problem of maximizing the cross-layer interference threshold is constructed.We decompose the original problem into subproblems of NOMA user pairing and power allocation and solve them iteratively using matching game and interior-point methods,respectively.For high-capacity services,we design a centralized cloud processing-based orthogonal access mode and construct a beamforming vector optimization problem that maximizes the number of admitted UEs.We solve the problem using semi-definite relaxation algorithm and lp-norm minimization method.The simulation results show that the proposed scheme can achieve efficient collaborative adaptation in different service scenarios.Compared with the baseline scheme based on minimizing total power,the number of admitted UEs increases by 25%on average.3.To address the challenges of multi-dimensional resource coupling and complex optimization in NOMA-FRAN,this thesis proposes a cache strategy based on NOMA,taking into account the cache capacity and power allocation limitations of edge access nodes in NOMA-FRAN.The impact of communication resource and cache placement under different access modes on transmission delay is studied,and a joint allocation problem of communication and caching resources is constructed to minimize the average transmission delay.The problem is decomposed into a joint user association and caching placement subproblem and a power allocation subproblem,and solved iteratively.A low-complexity joint user association and caching placement algorithm is proposed based on McCormick envelope theory and Lagrangian relaxation,and a low-complexity power allocation algorithm is proposed based on continuous convex approximation algorithm.Simulation results show that the proposed scheme can reduce the average transmission delay by 52%compared with the most popular maximum signal-to-noise ratio access baseline scheme when the cache capacity of edge nodes is limited.