Improved Cole-Cole parameter extraction from frequency response using least squares fitting

In this paper we propose using a method of non-linear least squares to extract the Cole-Cole impedance parameters (R_o, R_∞, α, τ) from collected frequency response datasets. We show how this method reduces the error of the extracted parameters when compared to direct extraction methods whose accuracy are limited by the measured low and high frequency gains. Parameters extracted using the non-linear least squares show up to 4 orders of magnitude lower relative error from ideal simulated datasets compared to direct methods. Experimentally extracted parameters from fruit tissue frequency responses verify the accuracy of the proposed non-linear least squares method over the direct extraction method.