随着无线业务需求的快速增长，网络资源(比如频谱、能量、空间等)成为影响系统性能的一个重要因素。而现有的通信网络存在频谱利用率低，功率消耗大，系统时延较长等问题，不能很好地适应未来海量流量、绿色网络、物联网等发展的需要。如何充分发挥这些有限资源的效用与价值一直都是学术研究的焦点。通过密集部署低功耗基站是降低能耗、提高资源利用率的一种有效途径。这些密集部署的基站与现有的通信网络组成超密集异构网络，这给网络资源管理以及网络间的切换带来巨大的挑战。本文以MIMO技术与超密集异构网络为研究基础，设计适合新型网络下的资源分配和切换方案，对于推动5G技术的应用具有重要意义。本文主要工作如下： (1)针对5G通信系统的组网问题，论文基于网络自组(SON)和IP方式，构建了包括宏基站、微基站、微微基站、WiFi、WLAN、3G以及LTE等多种接入网的超密集异构网络架构。该架构通过采用智能管理(AIM)、MIMO以及全频正交多址接入等技术，有效提高频谱资源利用率，降低系统能耗和时延，且降低组网成本和提高网络的开放性、灵活性及可扩展性。 (2)针对通信网络中频谱资源利用率低和能耗高的问题，提出了改进的模式搜索和多目标粒子群算法的网络资源分配算法。首先，在全面考虑超密集异构网络特点的基础上，分析了影响频谱利用和能量消耗的因素，并对需要优化的问题进行建模。其次，为使系统吞吐量最大，根据凸优化方法，利用拉格朗日对偶中的KKT条件分析导出了用户的动态注水功率。接着，为求解有约束的优化目标，对模式搜索和多目标粒子群算法进行了改进。最后，在迭代运算中，采用次梯度的方式处理优化目标方程中不可微的拉格朗日乘子问题。通过仿真，结果表明了所提算法能够明显地提高频谱利用率并降低系统能耗。 (3)针对授权频谱利用率不高的问题，提出了基于频谱感知与空间信道控制的联合资源分配算法。首先，在超密集异构网络环境下，建立优化目标的数学模型。为使系统容量最大，利用拉格朗日对偶以及波束功率约束的松弛条件解得非凸问题的最优控制波束矢量集。为简化求解，给出了一种基于干扰消除的控制波束矢量的求解方式。最后，根据资源调度为感知用户分配资源，并把传输数据加载在控制波束矢量上。由于该方案有效地降低主用户和次级用户间的干扰，因而提高了空闲频谱的利用率。仿真结果表明，联合控制算法较现有方法具有更大的系统容量。 (4)针对超密集异构网络切换中网络选择的问题，提出了一种基于区域感知贝叶斯决策的策略。首先通过区域感知，分析用户在各基站中的覆盖概率。其次根据最大似然估计，计算在各基站覆盖条件下用户需要切换的条件概率。接着，根据贝叶斯原理计算各基站的贝叶斯概率。最后选择贝叶斯概率最大的作为用户要切换的目标网络。同时，提出了结合用户服务要求(速率，RSS)形成超密集异构网络环境下联合切换策略，以实现用户的无缝接入。实验结果显示本文提出的切换策略能够有效地解决超密集异构网络切换时的网络选择问题。 关键字：超密集异构网络，MIMO，空间信道控制，贝叶斯决策

 As the demand for wireless services grows rapidly, network resources (such as spectrum, energy, space, etc.) become an important factor affecting system performance. However, there are some drawbacks in existing system, such as lower throughput, higher energy consumption, higher delay etc. Therefore, it cannot meet the future demand of application in massive traffic, green network and internet of things. It is focus to make fully use of utilization and value of these limited resources (such as frequency spectrum, energy, space), which has ever been concerned by many scholars and researchers. Dense deployment of low-power base stations is an effective way to reduce energy consumption and improve resource utilization. These densely deployed base stations and existing communication networks form an ultra-dense heterogeneous network, and this brings a great challenge to network resource management and handover between networks. Based on the research of MIMO technology and ultra-dense heterogeneous network, this dissertation designs a resource allocation and switching scheme suitable for new networks, which is of great significance for promoting the application of 5G technology. The summarizations of the work are as following: Aiming at the networking problem of 5G communication system, based on the network self-organizing (SON) and IP methods, the dissertation constructs ultra-dense heterogeneous network architecture including macro base stations, micro base stations, pico base stations, WiFi, WLAN, 3G and LTE etc. By adopting technologies such as intelligent management, MIMO, and full-frequency orthogonal multiple access, the architecture effectively improves spectrum resource utilization, reduces system energy consumption and delay, reduces networking costs, and improves network openness, flexibility and scalability. To solve the problem of spectrum resource's low utilization ratio in and high energy consumption wireless communication networks, an improved pattern search and multi-objective particle swarm optimization algorithm for network resource allocation is proposed. First, fully consider the characteristics of ultra-intensive heterogeneous networks, the factors affecting spectrum utilization and energy consumption are analyzed and the problems that need to be optimized are modeled.Then, in order to maximize system throughput, according to the convex optimization method, the user's dynamic water injection power is derived by using the KKT condition analysis in the Lagrangian duality. The improved pattern search and the multi-objective particle swarm algorithm are used to solve the constrained optimization target, and in the iterative operation, the subgradient is used to solve the non-differentiable Lagrangian multiplier problem in the optimization objective equation. The result shows that the proposed algorithm can significantly improve the spectral utilization and reduce the energy consumption. For the problem of low utilization of authorized spectrum, a joint resource allocation algorithm based on spectrum sensing and spatial channel control is proposed.Firstly, a optimization mathematical model is established in an ultra-dense heterogeneous network environment.Then, in order to maximize the system capacity, the Lagrange dual and beam power constrained relaxation conditions are used to solve the optimal beam vector set of non-convex problems. In order to simplify the solution, a solution method of beam control vector based on interference cancellation is also given. Next, according to the resource scheduling, the spectrum resource is allocated to the sensing user, and the transmission data is loaded on the controlling beam vector. Since the scheme effectively reduces interference between the primary user and the secondary user, the utilization of the idle spectrum is improved. The simulation results show joint control algorithms have greater system capacity than existing methods. For solving the problem of network selection in the ultra-dense heterogeneous network handover, the scheme of region sensoring based on Bayesian decision is proposed. Firstly, through area perception is adopted, the coverage probability of users in each base station is analyzed. Then, according to the maximum likelihood estimation, the conditional probability is calculated that the user needs to switch under the coverage condition of each base station. Next, according to the Bayesian principle, the Bayesian probability of each base station is calculated. The network with the largest Bayesian probability is selected as the user's handover target network. At the same time, it is proposed to combine the user service requirements (rate, RSS) to form a joint handover strategy to achieve seamless access for users. Experiment results show the proposed scheme can deal with effectively network selection problem in the handoff of UDHN. Keywords: Ultra dense heterogeneous networks, MIMO, Spatial channel controlling, Bayesian decision