Using MVPCA: An uncertain sensor data estimation method

作者：Yan, Xiaozhen^*; Xie, Hong; Wang, Tong

来源：Journal of Computational Information Systems, 2012, 8(10): 4185-4192.

• Data estimation
• Data processing applications
• Estimation methods
• Feature data
• Intrinsic features
• Local fields
• Local sensors
• MVPCA
• Principle component analysis
• Sensor data
• Sensor errors
• Sensor noise
• Transmission error
• Uncertain datas
• WSN (wireless sensor networks)

As the adverse impact of sensor noise, sensor error, transmission error and other factors in WSN (Wireless Sensor Networks), the sensor data of different nodes in the same local sensor field are different from each other, which lead to the uncertain sensor data of the local sensor field. This has created a new challenge for further data processing applications and decision making of the local field sensor data. To the best of our knowledge, there is no related research works on the uncertain local sensor filed data estimation problem. So in this paper, an uncertain sensor data estimation method based on MVPCA (Multiple Variable Principle Component Analysis) was proposed. Firstly the uncertain sensor data are pretreated by MVPCA, so the intrinsic feature data of the uncertain data are obtained. Secondly a multiple-regression estimation method is applied to the feature data to model and estimate. At the stage of evaluation, the sensor data of real wireless sensor network are used to test this method, different parameters is analyzed and discussed. The experiment results show that the scheme can estimate efficiently the uncertain sensor data. So it can be used in wireless sensor network practically.

• 出版日期2012
• 单位哈尔滨工程大学