Lossy Compression of Multi-Channel EEG and PPG Signals Based on Golomb-Rice Coding with Parameter Estimation

The integration of electroencephalography (EEG) and photoplethysmography (PPG) in consumer-grade wearable devices is revolutionizing healthcare by enabling continuous health monitoring. However, these devices face significant challenges in terms of power consumption, computational complexity, and data storage, necessitating efficient data compression techniques. This paper proposes a modified Golomb-Rice coding method tailored for lossy compression of multi-channel EEG and PPG signals. By leveraging the statistical properties of filtered data and employing an optimized parameter estimation strategy, the proposed method achieves high compression ratios while minimizing computational overhead and preserving critical signal information. The proposed algorithm shifts data to handle negative values, scales and converts it into an appropriate format for encoding, and uses a novel parameter estimation technique to optimize the Golomb-Rice parameter. Extensive simulations on EEG and PPG datasets demonstrate that this parameter estimation method yields significantly better performance, improving both efficiency and processing speed. Results underscore the importance of preprocessing in data compression and provide a robust framework for future advancements in biomedical signal processing.