For a practical usage, we want the responses of Physical Unclonable Functions (PUFs) to be unpredictable for an attacker, but reproducible for a legitimate user—intuitive criteria which need to be specified in the form of statistical tests to be useful for a practical evaluation. Practical tests range from simple ones, e.g. calculating the bias of the responses (bit 1 should be as likely as a 0 overall), to more complex tests like estimating spatial correlations between response bits. Each checks for a different aspect, pointing towards particular classes of possible issues.

The aim of this work is to consolidate existing tooling for the assessment of PUFs from measurement data into a newly built framework. The targeted end result is a common test suite which is

• generic regarding the concrete dataset, its data format, its dimensions, and the applicable tests,
• extensible, i.e. includes the currently existing tests but can be easily adapted to cover additional ones, and
• maintainable and auditable to allow for confidence in the correctness of the results.

Voraussetzungen

For a practical usage, we want the responses of Physical Unclonable Functions (PUFs) to be unpredictable for an attacker, but reproducible for a legitimate user—intuitive criteria which need to be specified in the form of statistical tests to be useful for a practical evaluation. Practical tests range from simple ones, e.g. calculating the bias of the responses (bit 1 should be as likely as a 0 overall), to more complex tests like estimating spatial correlations between response bits. Each checks for a different aspect, pointing towards particular classes of possible issues.

The aim of this work is to consolidate existing tooling for the assessment of PUFs from measurement data into a newly built framework. The targeted end result is a common test suite which is

• generic regarding the concrete dataset, its data format, its dimensions, and the applicable tests,
• extensible, i.e. includes the currently existing tests but can be easily adapted to cover additional ones, and
• maintainable and auditable to allow for confidence in the correctness of the results.

Voraussetzungen

In principle, an artificial neural network (ANN) can be trained to closely approximate any function. Progress in the domain of machine learning (ML) has shown that this universal approximation of functions has not only theoretical, but also practical relevance.

Error correction codes (ECCs) map information to a larger space, adding redunandancy so that the original information can be recovered despite erroneous data transmission. To decode a received data word and correct transmission errors, typically bespoke classical algorithms are used.

In principle, an ANN could be used in place of a classical decoding algorithm and prior research has shown that this specific application is indeed possible. The goal of this work is to look into this idea in more detail by

• doing a literature search on ML-based ECC decoders,
• assessing the feasibility of adapting the structure of an ANN to different ECCs, and possibly
• conducting practical experiments by training and evaluating such a decoder.

Voraussetzungen

For a practical usage, we want the responses of Physical Unclonable Functions (PUFs) to be unpredictable for an attacker, but reproducible for a legitimate user—intuitive criteria which need to be specified in the form of statistical tests to be useful for a practical evaluation. Practical tests range from simple ones, e.g. calculating the bias of the responses (bit 1 should be as likely as a 0 overall), to more complex tests like estimating spatial correlations between response bits. Each checks for a different aspect, pointing towards particular classes of possible issues.

The aim of this work is to consolidate existing tooling for the assessment of PUFs from measurement data into a newly built framework. The targeted end result is a common test suite which is

• generic regarding the concrete dataset, its data format, its dimensions, and the applicable tests,
• extensible, i.e. includes the currently existing tests but can be easily adapted to cover additional ones, and
• maintainable and auditable to allow for confidence in the correctness of the results.

Voraussetzungen

Es gibt einen Praktikumsteil zur Vorlesung, in dem verschiedene
Aspekte der IT-Sicherheit mithilfe eines eigenen Linux-Systems
und verschiedener Server-VMs praktisch geübt werden.

Deine Hauptaufgabe als Tutor*in sollte es sein, die
Studierenden während der Tutorstunden (2× wöchtentlich
à 1½ h) vor Ort bei der Bearbeitung dieser Aufgaben zu
unterstützen.

Daneben kannst du an der Wartung und Weiterentwicklung
der Aufgaben mitwirken und diese kreativ mitgestalten. Es gibt
stets Verbesserungspotential, was Verlässlichkeit und Inhalte
angeht!

Du solltest solide Linux-Kenntnisse mitbringen, da du häufig
Studierende, die vor der Vorlesung noch keinen Kontakt mit
Linux hatten, bei der Fehlersuche unterstützen wirst. Ein Besuch
der Vorlesung ist von Vorteil, aber keine zwingende
Voraussetzung.

Die Anstellung beläuft sich auf 6 h/Woche während der Vorlesungszeit im Wintersemester.