應用深度學習與蜂巢式網路資訊於5G行動網路之交通工具類別辨識

Transportation Type Identification using Deep Learning with Mobile Cellular Information for 5G Mobile Networks

林逸豪(Yi-Hao Lin)

指導教授 ： 陳志成

國立交通大學/資訊學院/資訊科學與工程研究所/博士(2020年)

摘要

網路切片是第五代(5G)行動網路系統的關鍵功能之一。在5G 行動網路中，能智慧 地了解用戶的移動類型是非常重要的，系統可為不同移動 型的用戶配置客製化的網 路切片。先前的研究中，使用者的移動類型可以通過利用全球定位系統（GPS）和智慧 型手機上配備的多個慣性感測器 辨識。但是，這些感測器具有一些限制，例如GPS 的信號不穩定以及慣性感測器易受干擾。在本論文中，我們提出了基於深度學習的交通 工具類型辨識（DeepTTI），它只能通過終端裝置的蜂巢式網路資訊來辨識使 者搭乘的交通工具類型。我們使用將近700 多小時的真實數據集進行效能評估。實驗結果顯示DeepTTI 的可用性，且可達到約95％的準確度。實驗結果還顯示，智慧手機可以減少8％-16％的電池消耗。

關鍵字

5G ； 移動類型辨識 ； 深度學習 ； Android 應用程式 ； 蜂巢式網路資訊

並列摘要

Network slicing is one of the critical features for 5th generation (5G) systems. In 5G, it is vital to intelligently understand the mobility patterns of users so that the system can allocate customized network slices for different types of users. In prior studies, user mobility type could be identified by exploiting Global Positioning System (GPS) and several inertial sensors
equipped on smartphones. However, they have some limitations, such as unstable signals on GPS and interference of inertial sensor data. In this dissertation, we propose Deep Transportation Type Identification (DeepTTI) which can identify transportation types by using cellular network information only. Performance evaluation is done with real-world dataset that contains more than 700-hour measurements. The experimental results confirm the effectiveness of the proposed DeepTTI, which achieves approximately 95% accuracy. The results also show that 8%-16% battery consumption can be reduced for a smartphone.

並列關鍵字

5G ； Mobility type identification ； Deep learning ； Android app ； Cellular information

參考文獻

[1] S. Chen, Y. Shi, B. Hu, and M. Ai, “Mobility-driven networks (mdn): from evolution to

visions of mobility management,” IEEE Network, vol. 28, no. 4, pp. 66–73, 2014.

[2] M. Series, “Imt vision–framework and overall objectives of the future development of imt

for 2020 and beyond,” Recommendation ITU, pp. 2083–0, 2015.

[3] 3GPP, “Telecommunication management, Study on management and orchestration of network