Inhalt des Dokuments
MA: Real-Time Operating Systems in the IoT
Real-Time Operating Systems (RTOSs) intend to serve applications with well-defined time constraints on embedded systems. Using priority-based scheduling algorithms, they are able to provide a high predictability concerning the time a task needs to perform its actions. Here, high predictability also has a higher priority than high throughput. Tasks are activated by I/O interrupts through real-time control systems like user interfaces or sensors or timer interrupts in the case of periodic tasks. With the advent of the internet of things (IoT), embedded devices have received more tasks that exhibit significant communication overhead. Interrupts generated by incoming network traffic lead to a preemption of tasks in the RTOS independent of the currently running task’s priority, which leads to increased jitter in systems under high load.
In our research we are looking for approaches to minimize the effect of incoming network traffic on the real time system. This might happen in between the operating system and hardware layers. To evaluate approaches, a suitable simulation tool has to be developed and a testing methodology for real systems must be found.
Interested students can develop a personal topic with regards to the following areas:
- Experimentation with modern RTOSs to assess their sensitivity to high I/O interrupt loads
- Evaluation of the applicability of state of the art RTOSs in (industrial) IoT applications
- Design of measures for real time guarantees in RTOSs in the IoT in NICs
- Development of a NIC-simulator to evaluate different interrupt moderation approaches
- Design and implementation of RTOS-based IoT applications for testing purposes
Some experience with operating systems and programming for microcontrollers (in C) is required. Additionally, a high degree of autonomy and some experience with scientific writing is expected.
If this sounds interesting to you, please send me an email with a little bit of background information on yourself, so we can quickly identify a fitting thesis topic together.