PhD Thesis: « Interferences sensibility of multi-core applications for autonomous vehicle »

The recent generation of embedded computers are relying on multicore
microprocessors for safety critical functions. This solution is
seducing as it offers a high computing power alongside with reasonable
power consumption and cost. Several software functionalities can be
ran in parallel. The problem is they slow down each other when using
shared resources, buses and memories, in a non-deterministic way. This
phaenomenon is called interferences.  Without restrictive assumptions,
gathering an upper bound on interferences is known to be an open
problem, especially considering industrial multicore microprocessors
that have been conceived to offer high average performance. Worse, it
has been showed that in practice, denial-of-service approaches could
demonstrate high level of interferences.

This thesis aims at studying the sensibility of software to
interferences in the context of autonomous vehicles for
safety-critical and hard real-time functions. The precise topic
consists in studying software that may generate or may suffer from
interferences, and estimate the execution time overhead caused by
these interferences. The PhD candidate will perform her/his research
using a combination of static analysis and empirical methods. The work
location will be part time at the EasyMile office in Toulouse
(https://easymile.com/), and at the Inria laboratory in Rennes. The PhD is
intended to start in late 2020 or early 2021. The PhD will be
supported by ANRT under the CIFRE program.

Contact information :
– Gilles Muller <gilles.muller@inria.fr>
– Isabelle Puaut <isabelle.puaut@irisa.fr>
– Xavier Jean <xavier.jean@easymile.com>

Fichier PDF
Télécharger le PDF