A Research On Data Regularity And Interpolation Algorithms For Wireless Sensor Networks

Wireless sensor networks have been the targets of active research in the recent past due to their military and civil applications. The low-cost of sensor nodes and the lack of real sensor network deployment in wireless sensor networks result in a wide range of inaccurate data, and thus the data need standardize. Meanwhile, a limit number of sensor nodes make monitoring blind spots exist in wireless sensor networks, and the monitor data need be interpolated. This paper deals with the data standardization and spatial interpolation issues in wireless sensor networks based on wavelet techniques and triangulation. The main works are as follows:1. The characteristics of wireless sensors technology and wireless sensor networks, along with the typical application of wireless sensor networks are summarized.2. The basic theory of Wavelet Transform was summarized from four aspects: Fourier transform, wavelet function character, multi-resolution analysis and Mallat algorithm. Lagrange interpolation, Elgin interpolation, Neville interpolation, Newton interpolation, Hermite interpolation and some other major data interpolation algorithms are also introduced in detail.3. A data regularity algorithm for wireless sensor networks based on wavelet is proposed. First, we set up an effective range of physical value to make a thresholding pretreatment to the sensor data. Then, a linear regression parameter adjustment algorithm and the abnormal data detection algorithm are designed based on wavelet transform. Finally, we make simulation experiments using the data from the real sensor network deployed by Berkeley-Intel research lab. The results show that our proposed algorithm can effectively retain abnormal data and remove erroneous data.4. A spatial data interpolation algorithm based on triangulation is proposed. First, we establish a triangulated graph of wireless sensor network based on the well deployed nodes. Then, an interpolation algorithm is derived based on triangulation in detail. Next, we provide a two-value environment map building methods. Finally, we make simulation experiments using the data from Berkeley-Intel research lab, compared with the existing IDW interpolation algorithm. The results show that our proposed algorithm has a good accuracy of data interpolation.