Abstract: A data compression algorithm for wireless sensor networks based on DTW and multi-wavelet transform is proposed. Firstly, the correlation and warping path of asynchronous data point pairs is introduced, and then an iterative algorithm for solving DTW warping path with maximal similarity is designed. Secondly, a best match point algorithm is presented, which employs prediction to estimate the relationship of synchronous data vectors and then find out the best match points. A sensory data matrix with maximal correlation is thus obtained. Afterwards, a multi-wavelet transform is introduced, which is capable of utilizing the spatial correlation of sensory data matrix and solving the row-column asymmetry problem. Experiment results show that proposed method has higher energy concentration ratio, better reconstruction accuracy and lower running time compared to the traditional distributed wavelet compression algorithm.