Real-Time Lossless Compression for Biomedical Signal Processing
Luca Notarianni, Anna Sabatini, Giulia Di Tomaso, Luca Vollero · IEEE Access · 2026
Wearable healthcare devices enable continuous monitoring of physiological signals, generating vast amounts of data that challenge energy, computational, and storage resources. Lossless data compression can address these issues by reducing data volume while preserving signal integrity. This study evaluates the “Quite OK Image Format” (QOI) lossless compression algorithm for photoplethysmographic (PPG) signals, which are essential for measuring vital parameters like heart rate and blood oxygen saturation. QOI was compared to libpng and stb_image_write (stbi) on a Raspberry Pi 3 and an STM32 microcontroller. Key metrics included encode time, decode time, compression rate, throughput, and samples compressed per second. On the Raspberry Pi 3B, QOI achieved an encoding time of 16.10 ms, an encoding speed of 6.80 million samples per second, and a throughput of 64.45 MB/s, outperforming libpng and stbi in encoding speed and throughput. On the STM32 platform, QOI processed more than 567 000 samples/s without reshaping and more than 650 000 samples/s with reshaping, while increasing the compression ratio from 3.37 to 4.16. These findings establish QOI and reshape optimization as effective tools for real-time PPG compression in resource-constrained healthcare applications.