The Research On Wirless Resources Management In Cloud Radio Access Networks

In recent years, with the wide using of smart devices and multimedia ap-plications, the mobile users and wireless traffic have grown explosively, which leads to the increase of cost and power consumption of mobile operators. How-ever, the revenue just stays flat or even decreases over time. As a result, the profit is decreased constantly. To address this problem, a novel architecture named cloud radio access network (Cloud-RAN) is proposed and attracts wide attention. In this architecture, the function of conventional Baseband Units(BBUs) are shifted to a BBU pool by using cloud computing and network virtu-alization technologies. In addition, the remote radio heads (RRHs) with simple structure are adopted as the access points to implement coordinated multi-point for users. As a result, Cloud-RAN can significantly reduce the overall cost and energy consumption of operators while effectively improving the system performance and resource utilization efficiency.It is well known that the resources of the wireless network include spec-trum, time slot, power, antenna and so on. With the growth of the number of mobile users and the enhancement of the users' requirements, the wireless re-sources become increasingly scarce. Accordingly, efficient and reasonable re-source management algorithm plays an important role in enhancing the spectral and energy efficiency, mitigating interference as well as ensuring the quality of service (QoS) of users. In general, resource allocation, resource scheduling,user association, handover control, interference management and service qual-ity guarantee are all the functionalities of the resource management. According to the characteristics of Cloud-RAN, we focus on the study of resource alloca-tion, user association and service quality guarantee in this dissertation. Corre-spondingly, several innovative schemes are proposed and the performance of these schemes are verified by computer simulation methods. The main contri-butions of this dissertation are concluded as follows:1. A joint dynamic time slot and power allocation strategy is proposed to deal with the asymmetry of uplink and downlink traffic.Firstly, we present a clustering algorithm to classify the RRHs into dif-ferent sets, and allocate resource for each cluster, which can significantly re-duce the computational complexity. Furthermore, zero forcing beamforming is adopted to eliminate the interference in each cluster. Secondly, we define a normalized mismatching degree as the performance metric, and set the met-ric as the target to formulate the dynamic time slot and power joint allocation problem as a mixed strategy non-cooperative game. Since the strategy space of each player in the game is so large that the Nash Equilibrium of the game is hard to be achieved through Best Response method, we employ the reinforcement learning algorithm to obtain a Logit Equilibrium. Finally, the simulation results show that the proposed strategy with low complexity can efficiently enhance the user rate and match the bidirectional traffic.2. Considering fronthaul capacity limitation and the requirements of users,we propose a user association scheme for Cloud-RAN to reduce the network power consumption.Firstly, we study the problem of joint user association and downlink beam-forming, with the objective to minimize the network power consumption while considering the fronthaul capacity constraint, the maximal power constraint and the quality-of-service requirements of users. Since the problem is formulated as a mixed integer nonlinear program (MINLP) and hard to be solved, we trans-form the problem into a mixed integer second-order cone program (MI-SOCP),and then present a low complexity algorithm to solve the MI-SOCP. Secondly,a joint admission control and user association scheme is put forward to deal with the problem that the network can not serve all the users. We prove that the proposed scheme can achieve the maximum number of MUs admitted by the network and the associated minimum network power consumption. Further-more, a low complexity algorithm based on semi-definite program is presented to obtain the suboptimal solution. Finally, simulation results show that the per-formance of the proposed algorithm is very close to the optimal solution.3. We propose a delay guarantee scheme which takes the discrete rate adaptation into account.Firstly, the user delay in Cloud-RAN is modeled as a double-layer queuing network, and then we derive the expression of user average delay through queu-ing theory. Secondly, we study the power consumption minimization which considering the minimal delay requirements for users, the maximal power con-straint and the discrete rate adaptation. Since this problem is also a MINLP,we reformulate it based on convex optimization. Then, we present an optimal algorithm based on branch and bound (BnB) method and a suboptimal algo-rithm with low complexity, respectively. Finally, simulation results show that the performance of the suboptimal algorithm is very close to that of BnB based algorithm, while the complexity of the former algorithm is much lower than the latter one.