ISSN 1000-1239 CN 11-1777/TP

• 网络技术 •

### 一种面向4G-LTE网络的丢包区分算法

1. (西安交通大学计算机科学与技术系 西安 710049) (陕西省天地网技术重点实验室(西安交通大学) 西安 710049) (zhangwzh@mail.xjtu.edu.cn)
• 出版日期: 2015-12-01
• 基金资助:
国家自然科学基金项目(61472317,61428206,61103239,91118005,91218301);教育部创新团队基金项目(IRT13035);国家科技支撑计划基金项目(2013BAK09B01);国家自然科学基金创新群体项目(61221063);国家公派高级研究学者及访问学者项目(201406285028)

### A Packet Loss Differentiation Algorithm for 4G-LTE Network

Du Haipeng, Zheng Qinghua, Zhang Weizhan, Yan Jifeng

1. (Department of Computer Science and Technology, Xi’an Jiaotong University, Xi’an 710049) (Shaanxi Province Key Laboratory of Satellite and Terrestrial Network Technology (Xi’an Jiaotong University), Xi’an 710049)
• Online: 2015-12-01

Abstract: On wireless networks, packet losses are usually caused by congestion or wireless fading. Loss differentiation is a fundamental problem of how to identify the cause of packet losses based on the transmission characteristics of wireless network. When moving on to LTE network, this remains an issue worthy of study due to the possible changes in packet loss patterns. In light of this, this paper proposes a loss differentiation algorism LDA-LTE for LTE network. We firstly conduct a detailed measurement study of the transmission characteristics of LTE network in configurable congestion and wireless channel fading scenarios. The test bed is based on an emulated LTE network in the laboratory by adopting a LTE base station emulator. Then, we find that the packet loss patterns that previous works used as the principal basis for loss differentiation do not hold when moving on to LTE network. This makes these works fail to work on LTE network. Finally, we design a loss differentiation algorism LDA-LTE, which collectively uses observed loss patterns to infer the cause of the packet losses on the LTE network in the hybrid scenario where congestion loss and wireless loss occur simultaneously. Evaluations on the test bed show that the accuracy of loss differentiation with LDA-LTE is significantly improved compared with previous work.