|Sept 03, 2019, 02:00 p.m.
|Room HFT-TA 617, Einsteinufer 25, 10587 Berlin
|Wiretap Code Design by Neural Network Autoencoders
In industrial machine type communications, an increasing number of wireless devices communicate under reliability, latency, and confidentiality constraints, simultaneously. From information theory, it is known that wiretap codes can asymptotically achieve reliability (vanishing block error rate (BLER) at the legitimate receiver Bob) while also achieving secrecy (vanishing information leakage (IL) to an eavesdropper Eve). However, under finite block length, there exists a tradeoff between the BLER at Bob and the IL at Eve. In this presentation, it is shown how neural network autoencoders can be used to flexibly design finite blocklength wiretap codes. To attain this goal, a multi-objective programming problem is formulated, which takes the BLER at Bob and the IL at Eve into account. Simulation results show that the proposed scheme can find codes outperforming polar wiretap codes with respect to both BLER and IL simultaneously.
Karl-Ludwig Besser received his Dipl.-Ing. degree in electrical engineering from Technische Universität Dresden in 2018. In August 2018, he joined the Communications Theory group at TU Dresden. His research interests are in the area of physical layer security and the application of machine learning in communications.