論文ID: 19-00183
This study proposes the efficient error rate estimation method for Single Event Effects (SEE) considering statistical uncertainty. SEE is known to make significant damage to a semiconductor device used for a spacecraft on orbit by exposed to cosmic rays or high energy protons. Semiconductor devices for the spacecrafts must conduct various ground experiments to evaluate the error rate using several environment simulators. Since the ground experiment takes much cost and time, the number of experimental samples is limited. That makes it difficult to estimate the error rate with high accuracy. That is, effect of statistical uncertainty due to the small number of samples on the error rate estimation should be considered. Therefore, this study proposes the highly accurate estimation method by introducing a prior distribution update into Hamiltonian Monte Carlo method (HMC) that is one of Malkov Chain Monte Carlo (MCMC) methods considering effects of statistical uncertainty. Through numerical calculations, the validity of the proposed method is illustrated.
