|Title:||Optimising Inter-core Data-Propagation Delays in Multi-core Vehicular Embedded Systems|
A large share of innovation in vehicular systems can be attributed to computer-controlled functionality. Traditionally, this functionality is realised by software that runs on single-core embedded processors. However, the advanced features and new customer requirements in modern vehicles require new levels of computation power. Multi-core embedded systems are being considered as a promising solution to support such advanced and data-intensive functionality. Apart from the computational requirements, vehicular systems are often required to be predictable, i.e., it should be possible to prove or demonstrate that all timing requirements specified on these systems are satisfied. The challenge of supporting predictable execution of the software on multi-core embedded systems is more difficult to deal with compared to single-core embedded systems, due to an additional data-propagation delay caused by shared resources, e.g., the system bus. One way to support predictable execution in multi-core embedded systems is to use a time-triggered schedule for the system bus. However, such a solution can result in large inter-core data propagation delays in the worst-case scenario.
This thesis focuses on developing new approaches to create optimised schedules for the system bus and cores to minimise the inter-core data propagation delays. The goal is to develop the scheduling tool prototypes that implement these approaches and conduct a performance evaluation using the prototype.
Expected main results:
1- Literature review and exploration of related schedules in the state of the art and practice
2- Developing an offline scheduling approach to optimise the bus schedule with respect to the static schedules of the application on individual cores.
3- Developing an offline scheduling approach to optimise the application schedules on individual cores with respect to the static schedule of the bus.
4- Develop and implement tool prototypes of the two approaches.
5- Performance analysis of a use case using the prototypes and provide the analysis results
6- Write the thesis report
• Programming knowledge in C
• Knowledge about real-time systems
The thesis can be adapted for 1-2 students for 30 Credit points
|IDT supervisors:||Saad Mubeen|