| Summary: | 碩士 === 國立臺北科技大學 === 電子工程系 === 107 === Weak signal coverage of the mobile communication network is a kind of the common reasons that greatly affect network performance. In Taiwan, the major cause is the radio jamming from the unknown wireless signal source near base stations. Another cause is dense and complex high-rise buildings of urban environment, that makes wireless signal of base stations suffer from multipath effects. Therefore, the impact of the multipath effects makes many unknown three-dimensional weak signal areas indoors and outdoors. The network users near the adjacent floor of the area experience high handover rate between different base stations, frequent handovers result in low communication quality and increase burden of base stations resource allocation. For the complex environmental conditions around the base stations, these weak signal areas are difficult to predict by simulation.
In order to overcome problems of these weak signal areas, we collect various types of sensing data of the user equipment and the base station for inputting into appropriate convolutional neural network (CNN). In the data collecting, an unmanned aerial vehicle (UAV) is used to measure various data at different altitudes, and using deep learning method to train model and online track orientation of the weak signal area center relative to moving UAV. In the future, the proposed method can be used for field trials of Internet Service Providers (ISP) and can reduce the density of urban field trials’ route, speed up where to find the weak signal areas. ISP could refer to the information of the weak signal areas, and then optimize deployment parameters and resource allocation of base stations or deploy a new base station for improving communications of quality of client side. In mobile communication system side, the serving base stations near weak signal areas will have the benefits of saving unnecessary transmit power and carbon reduction as well.
|
|---|
