An Adaptive Multiple-Modality Sensor Network Data Compression Algorithm Based on Lifting Wavelet and Polynomial Fitting

Suisheng CHEN,Jiangang LU

DOI: https://doi.org/10.3969/j.issn.1004-1699.2013.04.022

2013-01-01

Abstract:In order to improve the accuracy of the sensed data in wireless sensor networks, we propose an adaptive multiple-modality data compression algorithm based on lifting wavelet and adaptive polynomial fitting(AMLP)in this paper. The AMLP algorithm clustes all sensed data by using grey clustering technique, and approximates the relevant data in each cluster by using adaptive polynomial fitting. After that, the un-fitted characteristic data is abstracted as a matrix and then the AMLP algorithm uses lifting wavelet to remove the temporal and spatial redundancy. Finally, it compresses the data by RLE(Run Length Encoding)coding method. Simulation results demonstrate that the AMLP algorithm can effectively remove the redundancy both within the values of a single measurement as well as among values of different measurements, thus to decrease the energy consumption of sensor node. Although the proposed algorithm does not have advantage in compression ratio and energy consumption when compared to the adaptive multiple-modality data compression algorithm based on wavelet(AMMC), but it significantly outperforms the AMMC algorithm in data accuracy, so the AMLP algorithm is adaptive to the sensor network applications which require high accuracy data, such as geological hazard monitoring, medical care and military standard requirement.