Resource Allocation Theory And Method In Cloud Radio Access Networks

With the rapid development of wireless communication networks and the Internet of things,new system architectures and advanced signal processing technologies need to be introduced.A novel network architecture based on cloud computing technique,named cloud radio access networks(CRANs),integrates computing resource into a central baseband unit(BBU)pool to realize baseband processing functions of traditional base stations(BSs),while the radio frequency functions remain at remote radio heads(RRHs).The RRHs are deployed close to the users,thus the transmit power is reduced and the spectral and energy efficiency is improved.Furthermore,the virtualization technique can improve hardware unitization and the centralized signal processing can achieve cooperative gain in C-RANs.Because the users' demands become more diverse,the network performance metrics also show diversity.In addition to high data rate,low latency,low power consumption,and other key metrics have been the focus of attention.This thesis studies how to allocate power,time,spectrum,space,and other forms of resources to realize the network performance optimization objectives,such as maximizing network sum-rate,minimizing network power consumption,minimizing network delay,and so on.Firstly,for a two-tier heterogeneous C-RAN consisting of a macro-cell BS and multiple smallcell RRHs,the influence of time-frequency resource allocation on the system sum-rate is studied.A macro-cell BS applying massive multiple-input-multiple-output technique serves macro-cell users and wireless fronthaul links share the spectrum resource with radio access networks.A transmission protocol named reverse time-division duplexing is designed to reduce the inter-tier interference.Then,a scheme of interference cancellation is further proposed: The regularized zero-forcing precoding combined with a projection technique is used at the BS in downlink to avoid interference to the RRHs in uplink.Meanwhile,the joint linear minimum mean square error detection is applied in uplink to mitigate the inter-tier interference.The deterministic expressions for ergodic uplink and downlink sum-rates are derived by leveraging the large-dimensional random matrix theory,which then are used to optimize the bandwidth and time allocation to maximize the system sum-rate.Numerical results show that the deterministic sum-rate equivalents are accurate and that the proposed resource allocation method can significantly improve the system sum-rate.Secondly,in order to further study the impact of capacity-constrained wireless fronthaul links on the system sum-rate,an uplink heterogeneous C-RAN is considered,where a macro BS coexists with many RRHs to serve users together and wireless fronthaul links share the spectrum resource with radio access networks.Due to the limited capacity of fronthaul links,a joint optimization strategy of the dynamic bandwidth resource allocation and the compress-and-forward scheme design is proposed to maximize the system sum-rate.The compress-and-forward scheme includes point-to-point compression and Wyner-Ziv coding.In addition,the effects of different decoding schemes on the system sum-rate are also considered.Since the joint optimization problem is a mixed time-scale issue and in order to reduce computational complexity and communication overhead,an approximation problem of the joint optimization problem based on large system analysis is introduced,which is a slow time-scale issue because it only depends on statistical channel information.Furthermore,an algorithm based on Dinkelbach's algorithm and another algorithm based on uniform compression are proposed to find the optimal and suboptimal solutions to the approximate problem,respectively.Simulation results demonstrate that the bandwidth allocation factor derived from the approximation problem,as well as the compression noise matrix,is near-optimal.And the performance of the uniform compression strategy is close to that of the optimal compression strategy when the bandwidth allocation factor is small.Then,the power consumption in downlink C-RANs is considered,which includes the power consumed at the BBU for computation and the power consumed by RRHs and fronthaul links for transmission.The power minimization problem for transmission is a fast time-scale issue whereas the power minimization problem for computation is a slow time-scale issue.Therefore,the joint network power minimization problem is a mixed time-scale problem.To tackle the time-scale challenge,we introduce large system analysis to turn the original fast time-scale problem into a slow time-scale one that only depends on statistical channel information.A bound improving branch-and-bound algorithm and a combinational algorithm are further proposed to find the optimal and suboptimal solutions to the power minimization problem for computation,respectively.And an iterative coordinate descent algorithm is also proposed to find the solution to the power minimization problem for transmission.In addition,a distributed algorithm based on hierarchical decomposition is proposed to solve the joint network power minimization problem.Simulation results show that the network power consumption decreases with the increase of the delay constraint,and the performance of the compression-based transmission scheme is better than that of the data-sharing transmission scheme.Finally,this thesis study how to jointly optimize task scheduling,virtual machine(VM)allocation,and RRH assignment under the total power constraint to minimize task execution time and data transmission time in a clustered C-RAN.A hierarchical structure of virtual controllers is introduced in the clustered C-RAN,where a high-level controller is responsible for coordinating local controllers and each local controller is in charge of a cluster of RRHs.Moreover,each local controller is equipped with one server for creating VMs to execute users' baseband processing tasks.Then the joint optimization problem proves submodular and it is translated into a matroid constrained submodular maximization problem.Some heuristic algorithms are proposed to find solutions with 0.5-approximation.Besides,both centralized and distributed control schemes are proposed: the high-level controller in the centralized control scheme makes all decisions whereas the high-level controller in the distributed control scheme is only in charge of task scheduling based on graph theory and the local controllers are responsible for VM allocation and RRH assignment in their respective clusters.Simulation results show that the proposed algorithms can achieve better performance than the separate optimization algorithms of VM allocation and RRH assignment.