Switch Directed Dynamic Causal Networks -- A Paradigm for Electronic System Diagnosis

Electronic Systems Diagnosis, be it at the Device, Board, or System level, is a complex and time consuming task. Various techniques have been developed to provide design aids to the maintenance technician, each with its own successes and limitations, typically in terms of performance versus complexity issues. This paper demonstrates a novel integration of such techniques to provide for an effective and efficient approach to expert diagnosis of complex systems. The integration of Behavior Graph concepts and Causal Network analysis allows for the diagnosis of systems at a fairly high level of abstraction, allows for on-line diagnosis with or without explicit control of input stimuli, and provides for such diagnosis with minimal design detail or apriori fault assumptions.