Improving the performance of electrocardiogram sub-band coder by extensive Markov system

Abstract

The paper reports on a continuation of previous work on the digital coding of 500 samples s-1 ECG at 240 bits s-1. Focus is on six-band sub-band coding (SBC) with extensive Markov system as post-processors for each sub-band signal. An extensive Markov process, which considers both the coding redundancy and the intersample redundancy, utilises the redundancies by predicting the incoming n samples based on the previous m samples that constitute an extensive Markov process state. Both the previous m samples and the incoming n samples are considered as extension codes of the quantisation levels. Simulation results show that a reproduction with an average of signals ratio (SNR) of 27.21 dB or peak SNR of 58.6 dB was achieved at an average bit rate of 0.48 bit per sample, which corresponds to an average compression ratio of 25, whereas under the same signal fidelity, the previous proposed SBC system achieved an average bit rate of 0.714 bit per sample.