Skip to Main content Skip to Navigation

Exploration des techniques de fouille de données pour un monitoring efficace des systèmes intégrés sur puce

Abstract : Over the last decades, the miniaturization of semiconductor technologies has allowed to design complex systems, including today's several billions of transistors on a single die. As a consequence, the integration density has increased and the power consumption has become significant. This is compounded by the reliability issues represented by the presence of thermal hotspots that can accelerate the degradation of the transistors, and consequently reduce the chip lifetime. In order to face these challenges, new solutions are required, based in particular on the self-adaptive systems. These systems are mainly composed of a control loop with three processes: (i) the monitoring which is responsible for observing the state of the system, (ii) the diagnosis, which analyzes the information collected and make decisions to optimize the behavior of the system, and (iii) the action that adjusts the system parameters accordingly. However, effective adaptations depend critically on the monitoring process that should provide an accurate estimation about the system state in a cost-effective way. The monitoring is typically done by using integrated sensors (analog or digital). The industrial methods consist of placing one sensor per resource (static monitoring). However, these methods are usually too expensive, and require a large number of units to produce a precise information at a fine-grained resolution. This thesis proposes an innovative and ‘upstream' approach; a set of data mining techniques is used to analyze data extracted from various levels of abstractions from the design flow, in order to define the optimum monitoring in terms of cost and accuracy. Our method systematically identifies relevant information required for the implementation of effective monitoring. This thesis mainly focuses on the monitoring of the power and the temperature of the chip.
Complete list of metadatas

Cited literature [122 references]  Display  Hide  Download
Contributor : Abes Star :  Contact
Submitted on : Tuesday, January 29, 2019 - 11:17:27 AM
Last modification on : Wednesday, September 9, 2020 - 3:11:28 AM
Long-term archiving on: : Tuesday, April 30, 2019 - 3:17:32 PM


Version validated by the jury (STAR)


  • HAL Id : tel-01997733, version 1



Mohamad Najem. Exploration des techniques de fouille de données pour un monitoring efficace des systèmes intégrés sur puce. Micro et nanotechnologies/Microélectronique. Université Montpellier, 2015. Français. ⟨NNT : 2015MONTS154⟩. ⟨tel-01997733⟩



Record views


Files downloads