EMI Risk Estimation for System-Level Functions Using Probabilistic Graphical Models

In general, the functions provided by complex systems often involve multiple sub-systems and components that are functionally dependent on each other. The dependency could be to receive power, control signals, input data, memory storage, feedback etc. With the increasing use of electronic systems to perform critical functions, the potential for malfunctions due to electromagnetic interference need to be identified and mitigated. Hence, a risk analysis, estimating the likelihood and severity of electromagnetic interference effects, is desirable from the very early stages of system development. In this paper, the use of probabilistic graphical models for estimating the likelihood of electromagnetic disturbances causing system malfunctions with various degrees of severity is demonstrated using a very simple case study. Statistical data are synthesised to illustrate the construction of conditional probability distribution tables for a Bayesian Network system model. Factorization and inference techniques are then applied to demonstrate the formulation and answer of queries that could be of value during system risk assessment.