Refined Lightweight Temporal Compression for Energy-Efficient Sensor Data Streaming
O. Sarbishei · 2019
Lightweight Temporal Compression (LTC) is an energy-efficient lossy compression algorithm that maintains a memory usage and per-sample computational cost in O(1). The method provides a trade-off between compression ratio and accuracy using an error bound. In this paper, we present the Refined LTC (RLTC) algorithm, which uses a binning approach to widen the search space and increase the LTC's compression ratio and reduce its dynamic energy consumption, which is characterized by CPU computations and radio transmissions, without compromising the error bound. The proposed RLTC algorithm adds negligible overhead to the memory usage and latency of LTC. Experimental results on an environmental sensor dataset have shown that the LTC's compressed byte stream can be further reduced in size by up to 18%, while the dynamic energy consumption is reduced by 9.5% on average.