An advanced hardware design based on ensemble empirical mode decomposition algorithm for heart sound signal processing

Abstract

In this study, an advanced hardware design for heart sound signal processing based on ensemble empirical mode decomposition (EEMD) is developed and implemented. The EEMD method [1] is developed to alleviate a key drawback in the original empirical mode decomposition (EMD) algorithm. In a previous research, Huang et al. [2] developed an adaptive and efficient EMD method for nonlinear and nonstationary signal analysis. The physical meaning of a single intrinsic mode function (IMF) is obscure, and the original EMD algorithm cannot separate signals with different scales into appropriate IMFs. To overcome this major drawback, a noise-assisted data analysis (NADA) method called EEMD is developed. Heart sound signals are fed into the proposed system to simulate the EEMD-fixed-point performance. A comparison of the floating-point and fixed-point results exhibits satisfactory consistency and demonstrates that our design can accommodate wide variations of dynamic ranges and complicated calculations.