Hardware-Accelerated Data Compression in Low-Power Wireless Sensor Networks

摘要

In wireless sensor networks, the actual transmission of collected data is often the most energy-consuming operation. Frequently, it is worthwhile to spend energy aggregating the raw sensor data on the node to reduce the transmission effort. For many cases, lossless data compression can be employed as a general data aggregation method, as incompressible data (noise) generally does not carry any information worth transmitting. Nevertheless, the energy spent for data compression must be traded-off against the energy saved for transmitting the compressed data. In this work, sensor data of two real-life applications is compressed using a hardware-accelerator of the heterogeneous HaLOEWEn sensor node. The benefits of providing the node with a reconfigurable compute unit is demonstrated by comparing its energy consumption with that of of a purely software-based implementation.