ENHANCED MOBILITY MANAGEMENT ALGORITHMS FOR CELLULAR NETWORKS

Abstract

The use of mobile phones and other portable devices is continuously increasing the demand for high quality of experience (QoE) in wireless networks, such as huge data rate, and extremely low latency. To satisfy the heavily growing QoE demands, the ultra-dense network is considered a promising technique for fifth generation (5G) and beyond 5G cellular networks. Therefore, to support the data demand, as well as to increase network capacity small cells are densely deployed in present cellular networks. However, due to the low service area, the small-cell network is vulnerable to the mobility of user equipment units (UEs). During the course of movement, the wireless connections between UEs and small cells can fail frequently, such as handover failure or call drop, thus disturbing user experience. The problem will be exacerbated under ultra-dense small cell network if mobility-related parameters are not optimized. Hence, appropriate configuration and management of the network is required to enhance user quality of experience, and this thesis we studied two topics: handover optimization issues and resource management problems in cellular networks. The first part of the thesis focuses on handover optimization for seamless mobility under ultra-dense small-cell networks. In Chapter I, in order to overcome handover failure in ultra-dense small-cell networks, we propose a low-complexity distributed mobility robustness optimization framework for small-cell networks to optimize handover parameters, such as time-to-trigger (TTT), handover offsets (A3O_set), and cell-individual offsets (CIO). The framework performs handover failure classification by exchanging message between cells in the system. The failure includes too-late handover, too-early handover, and wrong-cell handover. Due to a trade-off between too-late handover with too-early handover and wrong-cell handover, handover parameters are optimized according to the reasons for failure. Results show that the proposed algorithm improved handover performance more than baseline algorithms. In chapter II, we analyzed handover problems to clarify when and how optimal handover parameters can be obtained. The study utilized geometry to model handover problems (such as too-late, too-early, and wrong-cell handover) and derived mathematical condition for handover failures. After that, optimal settings to avoid undesirable handover was introduced, and the trade-off between too-late handover and too-early handover was deeply investigated. We perform analyses for various aspects of wireless networks, such as impacts of interference, heterogeneous environments, mobility models, and network topology. In chapter III, utilizing the results of chapter II, we propose a machine learning based mobility robustness optimization framework for dynamic small-cell networks. Due to energy saving or traffic demands, small cell can be activated or deactivated, thus making the topology of wireless network become dynamic. Also, user mobility affects handover performance in a dense deployment of small cells. Taking into account the dynamics of network topology and user mobility, we apply transfer learning and reinforcement learning to optimize handover parameters. The transfer learning-based algorithm utilized the handover analysis in chapter II to adapt the varying topology, and reinforcement learning used the transferred knowledge to optimize handover parameters with a fast convergence. The results show that the proposed framework provide significant improvement in handover performance while achieving short convergence rate under dynamic small-cell networks. Part II of the thesis focuses on power allocation and beamforming design for multi connectivity 5G wireless network. In chapter IV, we propose a cooperating scheme to maximize network throughput while guaranteeing user quality of experience (QoE) demands in multiple-input-multiple-output (MIMO) systems. One of the aspired-to targets of the fifth generation (5G) network is to guarantee QoE everywhere in the network. However, UEs in the edge areas are vulnerable to QoE violations, and they need dual connectivity from two nearby transmission points. Hence, with the motivation to utilize multi connectivity to satisfy the demanded QoE, our algorithm categorized UEs into two sets: single-connectivity and dual-connectivity. After classification, transmission power is allocated to maximize the network capacity while guaranteeing the minimum QoE. We show that our proposed algorithm not only satisfies all the UEs in the system but also maximizes the network capacity and outperforms benchmark algorithms. In chapter V, we proposed a resource allocation algorithm for multi-connectivity wireless networks considering the minimum required QoE and the impact of CSI error. In practice, CSI error caused by hardware impairments or quantization of channel estimation scheme can affect the power allocation mechanism. Therefore, first, we investigate the impact of CSI error on the received signal quality at UEs in terms of signal-to-noise-plus-interference (SINR). Then, UEs are classified into two set: single-connectivity and dual-connectivity. After that, power is allocated to each set with the aims at minimizing transmit power and satisfying QoE requirement together. The results show that our algorithm can achieve 100 % satisfaction rate while minimize the transmit power.