Solution of an inverse scattering problem using optimization with a variable number of dimensions

The inverse scattering problem applies in many areas of engineering such as biomedicine, civil engineering, electromagnetic compatibility or geophysics. In this paper, one-dimensional inverse scattering problem is formulated as an optimization task with variable number of dimensions. Goal is to reconstruct material properties of a layered medium from its reflection coefficient. Then, this problem is solved using PSO-VND algorithm. Its solution is compared with solutions by conventional approaches: PSO, GA and DE optimizers. Results show that VND formulation significantly reduces computational time.