Abstract: Wireless sensor networks usually have limited energy and transmission bandwidth, and restricted computing and storage capacity, which pose new challenges in data dissemination and data analyzing and mining. This paper proposes a data approximation method, discusses data compression and approximate recovery, and analyzes its corresponding similarity measure. Theoretically and experimentally, under the circumstance of guaranteeing data precision, this method can compress data efficiently to save storage space, and reduce the amount of data transmission. Moreover, it can also analyze the similarity measure based on the compressed data, which can improve the efficiency of cluster analysis of data.

Key words: wireless sensor networks;data approximation;similarity measure;data mining