A scattered-field one-step leapfrog ADI-FDTD algorithm is proposed in the convolutional perfectly matched layer (CPML). The total field is split into the incident and scattered fields where the incident term is calculated analytically in the whole computational domain. A fundamental scheme is exploited to reduce the number of matrix operations. Also, a technique for the reuse of variables is introduced to reduce the number of memory variables. The proposed algorithm is verified through simulating the scattering of a plane wave by a metallic sphere and comparing the results with the conventional FDTD method and the Mie series solution. Simulations show that compared to the total-field formulation, the proposed method has a better performance in the shadow region of a scatterer.
