Spectrum Resource Sharing Algorithms For Tiered Heterogeneous Networks

With the popularity of smart phones, mobile Internet and all kinds of wireless service, there has an explo-sive growth of wireless data and wireless devices. Although the fourth generation (4G) mobile communication system, represented by LTE, has been commercially deployed, its transmission rate, spectrum efficiency, time delay, access density and other indicators are still unable to meet the growing demand for wireless access in the future. The research and development of the next generation mobile communication system, i.e.,5G, has been put on the agenda. Deploying small cell or D2D (Device-to-Device) equipment in the traditional cel-lular system thus forming tiered heterogeneous networks, is a core technology of 5G mobile communication system, since it can improve the system capacity, increase information coverage, improve the transmission rate, and improve the spectrum utilization.The deployment of small cell or D2D brings new challenges. For one thing, there exist both inter-tier interference and cross-tier interference as a result of spectrum sharing for improving spectrum efficiency, which severely restricts the performance of the system. For another, the traditional centralized control is not applicable to tiered heterogeneous networks due to the fact that the density of small cell and D2D is much higher than that of the traditional cellular base stations(BSs), and the centralized control will cause a huge amount of feedback and large delays. In order to effectively deal with the complex interference in tiered heterogeneous networks and the challenge of high-density deployment, in this paper, we study distributed resource allocation interference management for two main types of tiered heterogeneous networks, namely the Small Cell/Macro Cell and the D2D/Macro Cell networks.Firstly, for Small Cell/Macro Cell tiered heterogeneous networks, we develop a distributed algorithm based on the generalized Nash equilibrium problem (GNEP) to optimize the achievable rates of macro base station (MBS) and small base station (SBS). Goal is to ensure that all the base stations can actively maximize their transmission rates under the premise of meeting the quality of service (QoS) requirements of macro cellular users (MUEs). For this purpose, the Small Cell/Macro Cell network is modeled as a GNEP, and then we decompose the formulated GNEP into a traditional Nash equilibrium problem (NEP) and a variational inequalities (VI). Then, two distributed power optimization and price update algorithms are proposed, and the convergence properties of the algorithms and the existence and uniqueness of the equilibrium solution are provided. The simulation results show that the proposed algorithms only need the MBS to broadcast a small amount of price information and achieve the dual purpose of protecting macro cell communication and optimizing small cell communication in a distributed way. Compared with the distributed algorithm based on the traditional NEP, the proposed algorithms shows a greater improvement in performance.Secondly, we further study optimization of the utility of the entire Small Cell/Macro Cell network. Due to the interference between the macro cell and small cells as well as interference among small cells, the network utility maximization (NUM) is a highly non-convex problem that has brought great challenges for distributed optimization. Nevertheless, we establish a relation between the NUM and a GNEP. By introducing appropriate signaling exchanges, we propose a GNEP-based distributed algorithm to achieve a KKT solution of the NUM problem and prove its convergence. The simulation results show that the distributed algorithm can, upon guaranteeing the QoS of macro cell's communication, maximize the sum rate of the entire and achieve the same performance with the centralized method.Finally, we study the resource sharing of D2D/Macro Cell tiered heterogeneous network. Two typical questions have been considered. The first is to maximize the utility of all D2D links while MUEs have QoS requirements. The second is to maximum the utility of all D2D links while MUEs and part of D2D links have QoS requirements. As for the first question, we propose an approximation method, develop the corresponding distributed algorithm, and prove the convergence of the algorithm. As for the second question, since the constraints are neither convex nor linear, we then propose a logarithmic approximation method, and devise the distributed algorithm with its convergence. The simulation results show that these two distributed algorithms can reach the performance of the centralized optimization method and effectively improve the system spectrum efficiency.