Multilevel Modeling, Formal Analysis, and Characterization of Single Event Transients Propagation in Digital Systems

• Main Author: Bany Hamad, Ghaith
• Format: Dissertation
• Language: English
• Published: ProQuest Dissertations & Theses 01-01-2017
• Subjects: Electrical engineering
• ISBN: 9798209589914

The exponential growth in the number of transistors per chip brought tremendous progress in the performance and the functionality of semiconductor devices associated with reduced physical dimensions and higher speed. Electronic devices used in a wide range of applications such as personal entertainment systems, automotive industry, medical electronic systems, and financial sector changed the way we live nowadays. However, recent studies reveal that further downscaling of the transistor size at nano-scale technology leads to major challenges. Reliability (i.e., ability to provide intended functionality) is one of them, where a system designed in nano-scale nodes is expected to experience more failures in its lifetime than if it was designed using larger technology node size. Such failures can lead to serious conséquences ranging from financial losses to even loss of human life. Soft errors induced by radiation, which were initially considered as a rather exotic failure mechanism causing anomalies in satellites, have become one of the most challenging issues that impact the reliability of modern microelectronic systems, including devices at terrestrial altitudes. For instance, in the medical industry, soft errors have been responsible of the failure and recall of many implantable cardiac pacemakers. Depending on the affected transistor in the design, a particle strike can manifest as a bit flip in a state element (i.e., Single Event Upset (SEU)) or temporally change the output of a combinational gate (i.e., Single Event Transients (SETs)). Initially, SEUs have been widely studied over the last three decades as they were considered to be the main source of soft errors. However, recent experiments show that with further technology downscaling, the contribution of SETs to the overall soft error rate is remarkable and in high frequency systems, it might exceed that of SEUs [1], [2]. In order to minimize the impact of soft errors, the impact of SETs needs to be modeled, predicted, and mitigated. However, despite considerable progress towards developing efficient methodologies for the functional verification of digital designs, advances in non-functional verification (e.g., soft error analysis) have been lagging. This is due to the fact that the modeling and analysis of non-functional properties related to SETs is very challenging. This can be related to the random nature of these faults and the difficulty of modeling the variation in its characteristics while propagating. Moreover, many details about the design structure and the SETs characteristics may not be available at high abstraction levels. Thus, in high level analysis, many assumptions about the SETs behavior are usually made, which impacts the accuracy of the generated results. Consequently, the lowcost detection of soft errors due to SETs is very challenging and requires more sophisticated techniques.