現今隨著線上直播與遊戲的興起，人們對於無線網路WIFI的需求量變得越來越高。而一旦某個區域有大量的人群聚集時，就會產生大的干擾，這時現有的macro cell網路通常很難滿足用戶的服務質量(QOS)要求，甚至會造成某些用戶無法獲得通訊的服務。 在本篇論文裡，我們透過派遣無人機基地台(UABS)從空中進行快速支援，為macro cell提供分流服務，從而減輕macro base station（MBS）的負擔。無人機的優點就在於它能夠動態的對服務區域進行調整，可以以機器學習的方式根據使用者的位置來進行配置。我們討論無人機基地台網路之間的換手情形，以及對於如何有效的讓用戶平均分攤到每個基地台以增加傳輸，透過使用3GPP所提出的CRE和ABSF機制來減少干擾並且增加吞吐量。 模擬結果顯示，所測量到的負載平衡和換手成功次數在使用機器學習配置無人機基地台的方法下，都得到了顯著的提升。而吞吐量的部分在透過動態CRE和所提出的動態ABSF機制下也能觀察出有所增加。 ;Nowadays, with the rise of online live and games, people′s demand for wireless network has become higher and higher. Once a large number of people gather in a certain area, it will cause large path loss and interference. At this time, the existing macro cell network is often difficult to meet the user′s quality of service (QOS) requirements, and even some users may not be able to obtain communication service. In this thesis, we dispatch unmanned aerial vehicle base station (UABS) to provide rapid support from the air, and provide the diversion service for the macro cell, thereby reduce the burden of the macro base station (MBS). The advantage of the UAV is that it can dynamically adjust the service area, and can be configured according to the user′s location by machine learning manner. We discuss the handover situation between UAV base station networks and how to effectively distribute users to each base station to increase transmission. By using the CRE and ABSF mechanisms proposed by 3GPP, we can reduce interference and increase throughput. The simulation results show that the measured load balancing and the number of successful handover have been significantly improved by using machine learning to configure the UAV base station. The throughput part can also be observed to increase through the dynamic CRE and the proposed dynamic ABSF mechanism.
