Colloquium lecture: December 15 2026, 10:15 Uhr.

Betreuer/in: Dibaei Asl

Bild Besprechungsraum 04.137

Normally, safety analyses of railway systems start with the analysis of the hardware system, e.g. cables, functional units like switches and power supplies. The functional units mentioned above (cables, switches and power supplies) can fail, and this can lead to the failure of the whole system under analysis. Undergoing the possible failures of the cables, switches and power supplies and their propagation in the system.

Then the possible failures of the functional units are examined and the consequences of these failures are traced upwards through the system hierarchy. This is a reasonable starting point, but it neglects a whole set of components of failures that live entirely in the functional units of the software system, and in the assumptions made by safety-relevant applications about the network below. This communication infrastructure is a main part of the Train Control System (TCS).
As the TCS greatly depends on the communication infrastructure, the test cases will concentrate on failures of this communication. The communication of process data within the TCS is realized by the real- time Data Protocol (TRDP), IEC 61375-2-3. This communication protocol publishes process data in a cyclic publisher-subscriber model. Each datagram has a fixed size of 128 bytes as defined in IEC 61375-2-3. in IEC 61375-2-3. TRDP delivers process data in a cyclic publisher-subscriber model, transmitting a fixed-size datagram every 100 ms. To design a test for such a subtle failure mode, we have to describe how the data points that the publisher is sending to the sender through the PDcom are not arriving at the sender. For this, we have to generate some false data points that start their origin from the publisher to the sender, but meanwhile, during this, we have to give the sender some false data points so that, if the sender does not know, we have solid proof that our hypothesis is right. These late datagrams (of the publisher) are then, as with the other late datagrams of the publisher, accepted by the subscriber without any indication of loss of previous datagrams of the publisher having occurred in the meantime. Enough to periodically send new datagrams which also reset the subscriber’s timeout counter. Enough that some datagrams still arrive and reset the subscriber’s timeout counter. Enough that some still arrive and reset the subscriber’s timeout counter. The timeout never fires. The subscriber will continue to read from the same buffer with older and older data, and there will be no alarm generated elsewhere in the system, silently lost. From the same buffer with older and older data, and there will be no alarm generated elsewhere in the system, silently lost. A one-line error in a widely used open-source implementation can cause errors within the safety classification boundaries; such errors need to be detected. For such errors, there should be some hardware system or some experiments that can prevent these errors from happening again and again. In this thesis, we used FTA to track down the error and to propagate the fault inside them and how we can find a better solution for each component that causes such problems in the system by
proposing some solutions that use redundant gates, which gives the chance to lower the rate of error inside any functional component.


Raum 04.137, Martensstr. 3, Erlangen

oder

Zoom:
https://fau.zoom-x.de/j/68350702053?pwd=UkF3aXY0QUdjeSsyR0tyRWtLQ0hYUT09

Meeting-ID: 683 5070 2053
Kenncode: 647333