Data Reduction in Body Sensor Networks Using Wavelet Principal Components Analysis

Today transmitting biomedical data via networks for monitoring physiological situation of patients has an important role in healthcare systems. This paper represents a combinational compression method to optimize the energy usage of these networks by reducing transmitted data rate. In this method, physiologic signals are compressed in two stages. In first stage each signal is compressed by neural network thresholding of its discrete wavelet coefficients and only the coefficients which include main characteristics of their signals are selected. In the next stage the number of physiologic signals is reduced using principal component analysis according to their natural correlation. It should be noticed that original signals are reconstructed by considering trade-off between accuracy and compression rate.