Characterizing Empirical Mode Decomposition Algorithm Using Signal Processing Techniques

In this paper, we propose a model for empirical mode decomposition algorithm to represent nonlinear and nonstationary data. Two threshold operators (Signum and Relu) and the set of fundamental operators of a linear time invariant system (viz. delay, summer, and scalar multiplier) are used to completely characterize the proposed model. Models for finding number of zero crossings and number of local extrema of residual intrinsic mode function are also discussed. These representations are also based on the same block of elements, n-bit asynchronous up-counter and binary to decimal conversion. We obtain a closed-form expression for residual intrinsic mode functions to decompose the input signal using the proposed model. Performance of the proposed model is analyzed and discussed in terms of orthogonality index and percentage error in energy. Also, linear-in-the-parameter model for the two threshold operators is discussed in this paper.