The common-source digital upsets analysis by MDHMM for control system performance

The paper analyzes the common-source digital upsets process characterized by using multi-dimensional hidden Markov model (MDHMM) which is used to simulate digital upsets. The digital upsets are generated from the artificial environments operating on the distributed redundant processing controllers. The main contribution of this paper is to characterize the correlated digital upsets relationship generated by the common-source environment by MDHMM. Furthermore, the paper applies them to performance degradation of a control system analysis. In this paper, different methods are used to model the correlated digital upset processes under the same condition. The MDHMM model can better characterize the correlated digital upset processes than hidden Markov model (HMM) through comparing the distributions of the correlated digital upset processes. Finally, the performance degradation simulations of a certain control system are conducted when subject to common-source digital upsets, and the corresponding results are analyzed in details. These analyses also demonstrate the importance of the optimal redundancy design for control systems, which can provide a good analysis result on the best balance among performance, reliability, and costs in practice.