Software instrumentation for failure analysis of USB host controllers

Failures caused by electrostatic discharge (ESD) compromise the reliability of embedded systems. Peripherals such as the universal serial bus (USB) are particularly vulnerable, as isolating them to avoid electromagnetic interference would defy their purpose - facilitating communication with and/or by the embedded system. Better understanding the propagation of failures that result from ESD would facilitate defensive development of hardware and software for embedded systems, but is hampered by the lack of non-invasive and lightweight instrumentation techniques. This paper proposes the use of software instrumentation for monitoring the reaction of the USB peripheral of an embedded system to ESD. It describes our efforts towards detection and root cause analysis of ESD-induced failures - correlating changes in the operation of the peripheral with the specific pin subjected to ESD. The work described is intended as proof-of-concept for the development and use of (in situ) software instrumentation for lightweight acquisition of data that can be used for runtime failure analysis and actuation of self-healing mechanisms, as well as postmortem statistical analysis of system reliability, availability, and survivability.