北京邮电大学学报

  • EI核心期刊

北京邮电大学学报 ›› 2013, Vol. 36 ›› Issue (4): 110-115.doi: 10.13190/jbupt.201304.116.zhengjp

• 研究报告 • 上一篇    下一篇

基于高斯和粒子滤波的传感器数据处理技术

郑吉平1,2, 韩秋廷1, 张慧1   

  1. 1. 南京航空航天大学 计算机科学与技术学院, 南京 210016;
    2. 南京大学 计算机软件新技术国家重点实验室, 南京 210093
  • 收稿日期:2012-09-12 出版日期:2013-08-31 发布日期:2013-05-22
  • 作者简介:郑吉平(1979—),男,副教授,博士,E-mail:zhjcs@nuaa.edu.cn.
  • 基金资助:

    教育部博士点基金项目(20103218110017);江苏高校优势学科建设工程资助项目;南京航空航天大学青年科技创新基金项目(NN2012102,NS2013089);南京航空航天大学研究生开放基金项目(KFJJ120222)

Sensor Data Processings Based on Gaussian Sum Particle Filters

ZHENG Ji-ping1,2, HAN Qiu-ting1, ZHANG Hui1   

  1. 1. Department of Computer Science and Technology, Nanjing University of Aeronautics and Astronautics, Nanjing 210016, China;
    2. State Key Laboratory for Novel Software Technology, Nanjing University, Nanjing 210093, China
  • Received:2012-09-12 Online:2013-08-31 Published:2013-05-22

摘要:

对不确定传感器数据进行建模,利用高斯和粒子滤波技术进行概率推理,以达到节约能量的目的. 首先,根据传感器不同节点之间的时空相关性,采用历史数据建立概率模型;然后,在建立的概率模型上利用基本粒子滤波技术进行概率推理;最后,根据传感器数据符合高斯分布的特征,分别采用高斯粒子滤波、高斯和粒子滤波进行概率推理. 实验结果表明,高斯和粒子滤波在准确率和运行效率两方面均能达到良好效果.

关键词: 传感器网络, 动态概率模型, 粒子滤波, 高斯粒子滤波, 高斯和粒子滤波

Abstract:

Sensor data is modeled and Gaussian sum particle filter method is used for probabilistic inference based on model-driven methods for saving energy. First, the correlations on different sensor nodes and constructed probabilistic models from historical data are exploited. Then, the particle filter method is adopted to infer values for one sensor node from values acquired from real world by other sensor nodes, which saves energy of sensors efficiently. Finally, according to the fact that the sensor data generally satisfies Gaussian distribution, Gaussian particle filters and Gaussian sum particle filters are utilized for probabilistic inference respectively. Experiments show that the proposed Gaussian sum particle filter method is of high accuracy and efficiency.

Key words: wireless sensor networks, dynamic probabilistic models, particle filter, Gaussian particle filter, Gaussian sum particle filter

中图分类号: