期刊
COMPUTER NETWORKS
卷 129, 期 -, 页码 481-492出版社
ELSEVIER
DOI: 10.1016/j.comnet.2017.05.011
关键词
Healthcare IoT sensor networks; Time synchronization; Self-recovery; Two-points least-squares
类别
资金
- National Natural Science Foundation of China [61374154, 61572095, 61672131]
- Fundamental Research Funds for the Central Universities [DUT16QY27]
Sensor networks for healthcare IoT (Internet of Things) have advanced rapidly in recent years, which has made it possible to integrate real-time health data by connecting bodies and sensors. Body sensors require accurate time synchronization in order to collaboratively monitor health conditions and medication usage. Self-recovery and high accuracy are crucial for time synchronization protocols in sensor networks for healthcare IoT. Because body sensors are generally deployed with unstable energy sources, nodes can fail because of inadequate power supply. This influences the efficiency and robustness of time synchronization protocols. Tree-based protocols require stable root nodes as time references. The time synchronization process cannot be completed if a root node fails. To address this problem, we present a Self-Recoverable Time Synchronization (SRTS) scheme for healthcare IoT sensor networks. A recovery timer is set up for candidate nodes, which are dynamically elected. The candidate node whose timer expires first takes charge of selecting a new root node. Meanwhile, SRTS combines the two-points least squares method and the MAC layer timestamp to significantly improve the accuracy of PBS. Furthermore, SRP and RRP models are used in SRTS. Thus, our approach provides higher accuracy than PBS, while consuming a similar amount of energy. We use NS2 network tools to evaluate our approach. The simulation results show that SRTS exhibits better self-recovery than time synchronization protocols STETS and GPA under different network scales. Moreover, accuracy and clock drift compensation are better than those of PBS and TPSN. (C) 2017 Elsevier B.V. All rights reserved.
作者
我是这篇论文的作者
点击您的名字以认领此论文并将其添加到您的个人资料中。
推荐
暂无数据