Description
Date depot: 30 septembre 2019
Titre: Fault diagnosis of analog and mixed-signal circuits
Directrice de thèse:
Marie-Minerve LOUËRAT (LIP6)
Directeur de thèse:
Haralampos STRATIGOPOULOS (LIP6)
Domaine scientifique: Sciences et technologies de l'information et de la communication
Thématique CNRS : Non defini
Resumé:
The IoT concept will have an unprecedented impact on the automotive industry in the next decade. It will connect drivers and vehicles into a flow of information enabling real-time decisions and enhancing automotive experiences. IoT relies heavily on electronic systems that are capable of sensing, processing, and exchanging information. Apart from satisfying the application objectives, such electronic systems will have strong requirements in terms of on-line self-testing, reliability, and maintainability. Failures in such electronic systems may result in catastrophic consequences, placing in danger human lives, causing environmental accidents, jeopardizing the trustworthiness and integrity of data communication, causing non-repairable damages, etc.
The scientific and technological objectives of this thesis are to develop meaningful and comprehensive diagnosis methodologies and tools for heterogeneous, mixed analog-digital SoCs employed in automotive IoT applications. Diagnosis refers to the in-depth and meticulous analysis performed to identify the root cause of a failure that occurred either during manufacturing or in the field during normal operation. In particular, diagnosis aims at identifying the type of the defect that led to failure, including its localization and quantification. Automotive IoT applications demand zero defective parts per million, thus putting in place an effective diagnosis flow is essential for achieving this high reliability. Given a SoC that has failed, the objective is to develop a unified diagnosis flow that first pinpoints the IP block or interconnection that has failed (i.e. system-level diagnosis) and secondly, if the failure is attributed to an IP block, then it outputs a list of probable defects within the IP block and ranks these defects according to their probability of occurrence (i.e. IP block-level diagnosis).
This thesis will focus on IP block-level diagnosis specifically for analog and mixed-signal IPs. As of today there are no diagnosis solutions that have been proven to be efficient and practical. In fact, unlike digital IP blocks for which there exist diagnosis tools provided by EDA vendors, such as Mentor Graphics, Synopsys, etc., there are no diagnosis tools for analog and mixed-signal IP blocks.
To demonstrate the proposed concepts and quantify their impact, we will use as case study a mixed-signal IP block within a large SoC, which will be provided by STMicroelectronics.
Doctorant.e: Pavlidis Antonios