可调重构误差的传感数据时间相关压缩算法

ADJUSTABLE RECONSTRUCTION ERROR SENSOR DATA TIME CORRELATION COMPRESSION ALGORITHM

无线传感器网络(WSN)节点能量与带宽均非常有限,难以适应大量数据长时间传输的需求,所以非常有必要对原始采集的数据进行聚合或压缩处理。利用传感数据间存在的时间相关性,提出分段常量近似与Haar小波压缩相结合的二级压缩算法,在误差可调的情况下压缩该类时间相关的传感数据。通过真实数据集上的实验,分析该算法的数据重构误差、数据压缩比与压缩耗时情况,并与其他压缩算法进行对比。实验结果表明,该算法能够有效地利用传感数据中存在的时间相关性,显著减少冗余数据,有较高的压缩比并保证数据精度。

Wireless sensor network （WSN） lacks both node energy and bandwidth, so that it hardly meets the long time transmission needs ; hence it is essential to aggregate or compress originally gathered data. The thesis utilizes time correlations among sensor data to propose a two level compression algorithm that is a combination of piecewise constant approximation and Haar wavelet compression so that under adjustable error conditions the time correlation sensor data are compressed. Then through experiments on real data sets, the reconstruction error, data compression ratio and consumed compression time of the algorithm are analyzed and compared with that of other algorithms. Experiment results demonstrate that the algorithm can efficiently take advantage of time correlations that exist among sensor data, significantly reduce redundant data, boasts for higher compression ration and promises data precision.