Smartphone-Based Parameter Estimation of the Cole-Impedance Model for Assessment of Agricultural Goods

The Cole-impedance model is extensively utilized for modelling the electrical impedance of biological samples, including agricultural goods (e.g. fruits and vegetables). The conventional methods for estimating parameters of the Cole-impedance model rely on processing multi-frequency impedance datasets using non-linear least squares methods. The quality of the initial value used in these methods has a direct impact on the convergence and estimation accuracy, while requirement for complex processing units lowers portability and in-situ applications. This paper introduces method not dependent on a particular platform to estimate parameters of the Cole-impedance model that best represent an impedance dataset, eliminating the need for the specific toolbox within the software package, and it does not necessitate the user to supply initial values. The proposed method is validated using synthetic datasets (with and without noise) and experimental bioimpedance of carrot, potato, and pear samples. Further, it is implemented on a low-cost embedded hardware with execution time <7 seconds (for an impedance dataset with 256 datapoints) and estimation accuracy comparable to PC-based estimations. The embedded hardware is interfaced wirelessly to a smartphone application to demonstrate the in-situ graphical evaluation and reporting available using the proposed system.