Lightweight temporal compression of microclimate datasets [wireless sensor networks]

Abstract

Since the inception of sensor networks, in-network processing has been touted as the enabling technology for long-lived deployments. Radio communication is the overriding consumer of energy in such networks. Therefore, data reduction before transmission, either by compression or feature extraction, will directly and significantly increase network lifetime. This paper evaluates a simple temporal compression scheme designed specifically to be used by mica motes for the compaction of microclimate data. The algorithm makes use of the observation that over a small enough window of time, samples of microclimate data are linear. It finds such windows and generates a series of line segments that accurately represent the data. It compresses data up to 20-to-1 while introducing errors in the order of the sensor hardware's specified margin of error. Furthermore, it is simple, consumes little CPU and requires very little storage when compared to other compression techniques. This paper describes the technique and results using a dataset from a one-year microclimate deployment.