Diploma and Master Theses (authored and supervised):

D. Riezler:
"Emulating the Programming Interface of Commercial CAN Controllers in a Time-Triggered Environment";
Supervisor: H. Kopetz, R. Obermaisser; Institut für Technische Informatik, 2006; final examination: 2006.

English abstract:
Today, the Controller Area Network (CAN) is the most widely used communication protocol in the automotive industry thanks to its flexibility, extensibility and the high availability of hard- and software on the market. However, CAN is regarded as not suited for emerging safety-critical by-wire applications which require high bandwidth, deterministic communication behavior, ultra dependability, and composability. All these demands can be satisfied by time triggered architectures, which offer reliable message transport with time-triggered protocols such as TTP/C or FlexRay. During and after the change from the event-triggered CAN protocol to time-triggered architectures, it is important for the automotive manufacturers to reuse existing CAN-based legacy systems to avoid costly redevelopment. Therefore, it is necessary to offer services for transporting event-triggered CAN messages in a time-triggered architecture which can be accessed by the legacy application by means of common CAN Application Programming Interfaces (APIs). Several CAN APIs have been developed with different functionality for a broad range of application fields, but they all share a common goal: to make CAN-based applications independent from different communication controller implementations. Due to its wide-spread usage in the automotive sector, the Hersteller Initiative Software (HIS) CAN Driver API with its handle-based interface has been chosen to form the API for the legacy applications to exploit virtual CAN networks in this thesis. A virtual CAN network is an event-triggered overlay network on a time-triggered network which enables the transmission of CAN messages. An implementation of a virtual CAN network offering the HIS CAN Driver API is presented in this thesis, together with an example application to further explain the usage of the API. The functionality and performance of the implemented system is compared to a physical CAN network. The results of this comparison show that the functionality and performance of the physical CAN network can be established, and its dependability is even surpassed, when the CAN-based legacy systems are integrated into a time-triggered architecture.

Created from the Publication Database of the Vienna University of Technology.