一种完备数据流的不确定数据择优算法

Algorithm for Choosing Optimal Uncertain Data of Complete Data Streams

针对射频识别(RFID)数据与上层应用需求之间存在的信息鸿沟及其需要实时处理的特征,提出了一种完备数据流的不确定数据择优算法。分析了常规粒子滤波方法存在的不足之处,采用基于熵的方法推导属性最优权重,并利用可能度矩阵选择最佳粒子,从不确定RFID数据流上有效捕获对象的当前状态。算法的优化结果使得采样集向后验概率密度分布取值较大的区域运动,从而提高了算法计算效率并且显著地减少了精确定位所需的粒子数。最后,通过实例表明了该方法能够有效度量RFID数据中蕴含的不确定性。

To address the information gap between RFID data and the requirements of upstream applications, the chara- cter of real time of sensor data, an algorithm for choosing the optimal uncertain data of complete data streams was pro- posed. The drawbacks of generic particle filter were analyzed. Then an entropy-based method was adopted to estimate the most likely attribute weight for each object, by using possibility degree matrix to select optimal particles~ to effi- ciently capture the possible locations and containment for tagged objects. The performance of the generic particle filter is improved. In this method, though particle optimization, particles are moved to the regions where they have larger values of posterior density function. The experimental results show the accuracy and efficiency and the number of particles nee- ded for accurate location are reduced dramatically. Finally,a numerical example was given to show the feasibility and ef- fectiveness in terms of measurement of underlying uncertainties over RFID data.