Resource Allocation For D2D Communicatios In Cellular Networks

Recently, as the tremendous popularity of powerful mobile devices and the rapid advance of wireless communictions, there is a growing trend for proximity-based applications, such as peer-to-peer(P2P) file sharing, local multicasting and advertising. To satisfy the increasing demand of high-data-rate local services and simultaneously to provide better user experience as well as to alleviate the huge infrastructure investment of operators, device-to-device(D2D) communications have being considered as one of the key techniques in future cellular networks.With D2 D communications, proximity users in a cellular network can communicate directly to each other without going through the base station(BS). Since the physical proximity, it can bring proximity gain, reusing gain and hop gain, and thus can potentially increase system spectral-efficiency(SE) and device energy-efficiency(EE). However, D2 D communications may generate harmful interference into the existing cellular network if not designed properly. Therefore, interference management is one of the most challenging and important issues for D2 D communications to harvest the potential benefits. In this dissertation, we mainly focus on D2 D interference management and resource allocation issues including the following parts:(1) Cross-layer optimization for D2 D communications with perfect channel state information(CSI);(2) Qualityof-service(QoS) aware resource allocation for D2 D communications with channel uncertainty;(3) SE optimization based mode selection;(4) EE optimization based mode selection.In the research of cross-layer optimization for D2 D communications with perfect CSI, a framework of resource allocation for D2 D communications underlaying a fully loaded cellular network is presented, where the objective is to maximize the overall network throughput of existing cellular users(CUs) and admissible D2 D pairs while guaranteeing the QoS requirements for both CUs and D2 D pairs. The framework includes three parts. First, a minimum distance metric is proposed for BS to decide whether a D2 D pair can be accessed or not under QoS requirements for both CUs and D2 D pairs. Then, an optimal power control scheme is investigated for each D2 D pair and its possible CU partners to maximize the overall throughput. At last, a maximum weight bipartite matching based scheme is developed to determine a specific CU partner for each admissible D2 D pair. Numerical results show that compared with the Fixed margin scheme and MINLP based heuristic scheme, the proposed scheme can significantly improve the performance of the hybrid system in terms of D2 D access rate and the overall network throughput.The performance of D2 D communications depends on D2 D user locations, cell radius,the numbers of active CUs and D2 D pairs, and the maximum power constraint for the D2 D pairs.In the research of QoS aware resource allocation for D2 D communications with channel uncertainty, we investigate resource allocation of D2 D communications underlying cellular networks in fading channels and also analyze the tradeoff between feedback overhead and performance, where the BS only knows partial CSI of the links between regular CUs and D2 D receivers(CU-D link). We propose two different ways to reduce CSI feedback overhead: probabilistic and partial feedback schemes. We utilize the channel statistical characteristics to avoid the instantaneous CSI feedback and increase the access distance between CUs and D2 D pairs to counteract the channel fading. Since the exact signal-to-interference-plus-noise ratio(SINR) of D2 D users can not be determined, we use the maximum outage probability to guarantee QoS for D2 D pairs. We derive a closed-form expression of the distance factor for different channel models to satisfy the access conditions of both CUs and D2 D pairs and develop a probability-based resource allocation framework to maximize the overall network throughput of CUs and D2 D pairs. To avoid user outage and reduce CSI feedback, we then propose an efficient and simple partial feedback scheme as an alternative to deal with channel uncertainty,where maximum-distance ratio(MDR) metric is derived to select the best CUs to send feedback. In addition, to exploit benefits of both the probabilistic and the partial feedback schemes, we develop a combined scheme, which is less sensitive to the change of channel statistics and D2 D cluster radius. It is shown that the combined scheme significantly improves the performance of probabilistic scheme or the partial feedback scheme in terms of the overall network throughput and outage probability of D2 D users.In the research of SE optimization based mode selection, we investigate the optimal D2 D mode slection problem to maximize the system SE while considering the QoS requirements and power constraints for both D2 D pairs and CUs as well as prioritizing different classes of services. We develop frameworks to solve the optimization problems in the three D2 D transmission modes including dedicated mode、reusing mode and cellular mode. Specifically, the bisection algorithm is used to solve the quasiconvex optimization problems in the dedicated mode and the cellular mode while concave-convex procedure(CCCP) is adopted to deal with the difference of convex(D.C.) optimization problem in the reusing mode. Simulation results show that system SE can be significantly improved with the proposed mode selection algorithm compared with the single mode transmission without mode selection.In the research of EE optimization based mode selection, we study energy-efficient mode selection to reduce the energy consumption of user terminals, where both transmit and circuit power consumptions are considered. As in SE optimization based mode selection, we develop novel frameworks to solve the optimization problems in the three transmission modes. In particular, parametric Dinkelbach method is first adopted to remove the factional formal of the original nonlinear fractional optimization problems for better tractability. Then, CCCP is exploited to approximate the problems with D.C. structure. Finally, the classical interior point method with logarithmic barrier function is used to obtain the optimal solution. In addition we analyze the SE-EE tradeoff between SEand EE-based mode selection. Simulation results show that system EE can be improved significantly with the proposed mode switching algorithm compared with the single mode transmission. Besides, it is also shown that the reusing mode is more preferred in the EEbased mode switching while it is the dedicated mode in the SE-based mode switching in most situations.The research results in this thesis can provide a theoretical base for investigating more complex D2 D communications systems and also some useful guidelines for designing heuristic algorithms.