AMUR:一种RFID数据不确定性的自适应度量算法  

AMUR:An Adaptive Measuring Algorithm of Underlying Uncertainty for RFID Data

作　　者：王永利[1] 钱江波[2] 孙淑荣[3] 张功萱[1] 刘冬梅[1] 

机构地区：[1]南京理工大学计算机科学与技术学院,江苏南京210094 [2]宁波大学信息科学与工程学院,浙江宁波315211 [3]国电南京自动化股份有限公司技术管控部,江苏南京211100

出　　处：《电子学报》2011年第3期579-584,共6页Acta Electronica Sinica

基　　金：国家自然科学基金项目(No.60803001;No.60803021;No.60850002);中国博士后科学基金特别资助项目(No.200902517);江苏省博士后基金(No.0801043B);南京市科技计划项目(No.2010软资02014)

摘　　要：为适应基于RFID(无线射频识别)位置跟踪过程中传感数据的连续变化和需要实时处理的特征,本文提出一种度量RFID数据不确定性的自适应进化粒子滤波算法,根据K-L距离改变重采样粒子个数,并引入粒子群寻优方法PSO改变传统粒子滤波(SIRPF)的重采样效率,采用常规赋权聚集(CWA)定义适应度函数,以均衡先验密度与似然密度的重要性,在采样粒子空间探寻最优粒子,为概率数据库上的初始元组提供可靠的置信度度量.实验证明,与已有的算法相比,AMUR算法能够有效地度量RFID数据中蕴含的不确定性,可进一步改善粒子退化现象和粒子贫化问题.To adapt the character of evolving over time and real-time of sensor data in location tracing service based on RFID,we present an adaptive evolving particle filtering algorithm-AMUR（an adaptive measuring algorithm of underlying uncertainty for RFID data）.AMUR adaptively changes the number of samples on the basis of K-L distance,introduces an improved PSO（particle swarm optimization） method to enhance the efficiency of resampling phase of conventional particle filter（SIRPF）.Meanwhile,to detect the most optimal samples among candidate sample set,AMUR defines a fitness function based on CWA（conventional weighted aggregation） for PSO which balances the importance between priori density and likelihood densitys.It provides a reliable measure of confidence for initial tuple in the probability RFID database.Experimental comparison of current algorithms shows,AMUR outpreforms current methods in terms of measurement of underlying uncertainties over RFID data,particle degradation and particle depletion.

关 键 词：无线射频识别 物联网 不确定性 粒子滤波 自适应 粒子群优化 

分 类 号：TP311.2[自动化与计算机技术—计算机软件与理论]