System Throughput Optimization For D2D Communications Underlaying Cellular Networks

Device-to-Device(D2D) communication has been proposed as a promising technique to improve the spectrum utilization. In D2 D communications, two nearby users are allowed to communicate directly with each other over the D2 D link by sharing the cellular resources instead of through the base station. The cellular resources are allowed to be occupied by both D2 D users(DUs) and cellular users(CUs) simultaneously in hybrid cellular-D2 D networks. Therefore, the joint DU-CU association and resource allocation is a promising solution to improve the system spectral efficiency while keeping satisfactory quality of services for CUs. Optimizing the shared resource between cellular users(CUs) and DUs is meaningful to improve utility of spectral. Despite its importance, shared resource optimization in hybrid D2D-cellular networks is complex and hard to be solved. With different shared modes, system feature are different. In particular, the DU-CU association is to determine which DU and CU use the same cellular resource, and the resource allocation is then to determine how to allocate the cellular resource between these two DU and CU. The joint DU-CU association and resource allocation problem, although its significance, has not been thoroughly addressed for hybrid cellular-D2 D networks.This thesis addresses the joint DU-CU association and resource allocation problem in hybrid cellular-D2 D networks. The contributions of this thesis are summarized here. The joint DU-CU association and resource allocation problem in this paper is defined to maximize the system throughput while meeting the minimum data rate requirements of CUs and DUs in service through allowing DUs to share uplink resource blocks(RBs) of CUs. The defined optimization problem in this paper is solved in two phases: DU-CU association phase followed by the optimal resource allocation of RB between any two involved DU and CU. Due to the convex nature, the optimal solution to the resource allocation optimization problem can be obtained in the closed-form expression. On the contrary, the DU-CU association problem is an integer linear programming, and thus we propose a hybrid-network Resource optimization Method(hnRoM) based on linear programming and bipartite matching. Further, we extend this work to minimize power consumption meanwhile maximizing served DU number based the information of user positions.