Lightweight Data Compression in Wireless Sensor Networks Using Huffman Coding

This paper presents a lightweight data compression method for wireless sensor networks monitoring environmental parameters with low resolution sensors. Instead of attempting to devise novel ad hoc algorithms, we show that, given general knowledge of the parameters that must be monitored, it is possible to efficiently employ conventional Huffman coding to represent the same parameter when measured at different locations and time periods. When the data collected by the sensor nodes consists of integer measurements, the Huffman dictionary computed using statistics inferred from public datasets often approaches the entropy of the data. Results using temperature and relative humidity measurements show that even when the proposed method does not approach the theoretical limit, it outperforms popular compression mechanisms designed specifically for wireless sensor networks.