Multistatic electromagnetic imaging of dielectric targets with LSTM cells

Electromagnetic imaging of dielectric targets at microwave frequencies needs dealing with an inverse problem, whose solution process is further complicated by the unavoidable presence of model error. In this work, the adoption of a neural network based on long short-term memory cells is introduced for the compensation of such an issue. In particular, multistatic measurement settings are considered here for the first time. The proposed network performs a preliminary processing of the scattered field, which is then passed to a multifrequency nonlinear inverse-scattering algorithm formulated in non-Hilbertian Lebesgue spaces with non-constant exponents. The approach is initially validated in a simulated environment.