Efficient Communication Overhead Reduction using Polygonal Approximation-based ECG Signal Compression

ECG signal requires a high sampling frequency of 100 to 1000 Hz, as well as long measurement times of longer than 24 hours. Therefore, efficient data compression for storage and transmission of data is required. ECG signal can be represented by a fiducial point composed of the onset, offset, and peak, which are essential for ECG signal analysis. Detecting the onset and offset are ambiguous because the feature values are similar to those of the surrounding samples. In this paper, we represent ECG signal as vertices by polygonal approximation, and suggest an auxiliary signal generated by the amplitude change rate between vertices. The proposed method can compress the number of data bits to about 89.26% and preserve the fiducial points as vertices. Also, we analyze the features of each vertices and determine the fiducial points. The clustering results of QRS complex were stable with the QT-DB provided by Physionet.