2 positions on Physical Layer Security and Machine Learning (from April 2021)
Theoretical approaches to securing radio communication with mechanisms of physical layer security (PLS) are now well known in research. In particular, the method of generating cryptographic keys for two communicating parties using radio channel reciprocity (and variability) represents a promising approach in order to be able to efficiently secure devices in the Internet of Things (IoT) in the future. These methods are also of particular interest because, in contrast to classical cryptography, they are not based on assumptions of complexity of mathematical problems, but are considered safe from an information theory point of view. In this way, these techniques (potentially) solve the impending threat from quantum computers and reduce the scalability of attacks on complexity-based cryptographic techniques by implicitly excluding the physical proximity of the communicating parties as the basis of key establishment attacks from the remote. A feasibility in principle has already been shown in the successful project "PROPHYLAXE", and "SecureFog (see Past Project tab).
We propose the extension of reciprocity-based key generation to end-to-end (E2E) communication. One method that is supposed to make key generation usable in E2E communication is called so-called Extended Channel-Reciprocity Based Key Establishment (ECRKE). It uses untrusted relays (e.B. WiFi routers) and a so-called trust broker or authenticator (e.B.g. smartphone) to mask the channel.
In this work, the information-theoretical foundations and procedures for the transmission of secrets in distance-bound channels are to be developed, which forms the basis for E2E-capable authentication with the help of the authenticator in the ECRKE procedure. For this purpose, various designs for the transmission mask will be developed and tested, using both typical simulation tools and the Wi-Fi SDR testbed. An important element here is the design of eavesdropping-proof codes based on near-field measurements using learning algorithms. In conjunction with the key agreement procedures, the fundamental limits for the listening capacity (i.e. the rate at which a secret can be safely transmitted) are to be explored and corresponding metrics for the "correlation" to the authenticator and the untrusted relay are derived.
G. Wunder, R. Fritschek and R. Khan, „RECiP: Wireless channel reciprocity restoration method for varying transmission power“, in Proc. IEEE 27th Annual International Symposium on Personal, Indoor, and Mobile Radio Communications (PIMRC’16), 2016
R. Fritschek and G. Wunder, „Towards a Constant-Gap Sum-Capacity Result for the Gaussian Wiretap Channel with a Helper“, in Proc. IEEE International Symposium on Information Theory (ISIT’16), Barcelona, Spain, 2016
R. Fritschek and G. Wunder, „On-The-Fly Secure Key Generation with Deterministic Models”, IEEE International Conference on Communications (ICC’17), Paris, France, 2017
R. Fritschek, G. Wunder, On Full-Duplex Secure Key Generation with Deterministic Models, IEEE Conference on Communications and Network Security (CNS’17), Las Vegas, USA, October 2017
R. Fritschek and G. Wunder, „On the Gaussian Multiple Access Wiretap Channel and the Gaussian Wiretap Channel with a Helper: Achievable Schemes and Upper Bounds“, IEEE Transactions on Information Forensics & Security, 2018
R. Fritschek, R. Schäfer, G.Wunder, Deep Learning for the Gaussian Wiretap Channel, IEEE ICC’19, online: https://arxiv.org/abs/1810.12655, 2019
R. Fritschek, R. Schäfer, G.Wunder, Deep Learning for Channel Coding via Neural Mutual Information Estimation, IEEE SPAWC’19, online: https://arxiv.org/abs/1903.02865, 2019
Applicants must possess a master degree in computer science, mathematics, electrical engineering or similar, with coding skills in C, C++, Python/Matlab, TensorFlow, PyTorch
The candidate is expected to have a profound knowledge in the field of wireless networking (particularly Wi-Fi standards, Bluetooth standards) as well as good understanding of security. Additional knowledge on the fundamentals of Information Theory.
To be considered please send a short cover letter outlining your background, a detailed CV including possibly some references, and all relevant certificates to email@example.com.
By submitting an online application, you agree to process and electronically store your data. Please note that the Free University of Berlin cannot take responsibility for the security of unprotected transmitted personal data send electronically.
2 positions on Federated Learning and Blockchain