Numerical simulation of microring resonator biosensor with FDTD algorithm based on GPU and CPU architectures

In this paper we present numerical simulations of microring resonator biosensor with FDTD algorithm based on GPU and CPU architectures. High quality factor and high sensitivity microring resonator with implemented hallow core defects is analysed. Since defects are small compared to overall microring dimensions, small grid size is required to describe them accurately using FDTD scheme. This in turn greatly increases simulation time. One way to reduce computational time is to perform calculations on GPU. Here we compare two freely available FDTD implementations: one based on CPU (MEEP) and one - on GPU (B-CALM). Comparisons between these two implementations in computational time and accuracy are made.