An ECG compression algorithm based on gradient difference for body sensor networks

We propose a novel scheme for ECG compression based on Gradient difference used in BSN. As a newly emerged application, BSN can continuously monitor the patients' physiological signal without hindering their everyday life, however, it has some constraints, namely, size, power, capability of storage and computation. ECG is a nontrivial signal collected by BSN, continuous collection will bring huge data flow which is a burden both to sensor nodes and central node, therefore we have to do some compression to relieve the burden. As the constraints exist, most of the traditional algorithms are not adaptive enough, we need a simple, real-time and effective approach. This paper introduces an approach to meet those requirements, by conducting various experiments, we found that this approach can effectively compress ECG signal and meanwhile keep a good fidelity of the reconstructed signal, to prove that, we did some compare between our approach and traditional ones. Yet, this algorithm has not been tested on the clinical ECG signals and there is still room for the performance improvement.