Optimization of Data Fitting Algorithm for Tissue Dielectric Properties in THz-band using Genetic Algorithm

The non-linear least squares method (NLLS) is often used to extract the double Debye parameters. However, to solve NLLS, the traditional numerical iteration method depends heavily on the initial values. This paper provides a method to optimize the algorithm, using genetic algorithm (GA) to acquire suitable initial values for the numerical iteration method. Computer simulations show that the optimized algorithm extracts more accurate double Debye parameters and fits the data better.