Physical Layer Security for Channels with State and Active Eavesdroppers (PLAY SCATE)

• Applicants: Professor Dr.-Ing. Holger Boche; Professor Dr.-Ing. Eduard Axel Jorswieck
• Subject Area: Electronic Semiconductors, Components and Circuits, Integrated Systems, Sensor Technology, Theoretical Electrical Engineering
• Term: from 2016 to 2023
• Project identifier: Deutsche Forschungsgemeinschaft (DFG) - Project number 326920355

Nowadays, information theoretic approaches to security are intensively discussed as a complement to cryptographic techniques. Such approaches jointly establish reliable communication and data confidentiality at the physical layer by taking the properties of the noisy channel into account. Recently, this area of research has drawn considerable attention since it provides a promising approach to achieve unconditional security and to embed secure communication into wireless networks. In practical systems, CSI will be limited due to the nature of the wireless channel and estimation/feedback inaccuracy.In designing wireless systems resilient against failures caused by nature and robust against malicious attacks, the correct system-theoretic model is the compound and the arbitrarily varying channel, respectively. The CC model applies if the state of the channel is constant for the duration of one codeword. The AVC model is suitable if each channel use is affected by a different channel state. This correctly models malicious attacks.The goal of this project is to understand the fundamental properties of compound wiretap channels and arbitrarily varying wiretap channels with active eavesdroppers who voluntarily influence the channel states. For the system design it is important to consider constraints on the transmit as well as attacker nodes to correctly model their possibilities.In the literature, the characterization of the secrecy capacity of AVWCs relies on methods by Ahlswede. They cannot be applied directly under input and attacker constraints.In particular, it is interesting to determine whether the dichotomy holds under constraints, and for which constraints the resulting secrecy capacity is zero. It is expected that semi-deterministic coding will suffer from symmetrizability due to active attacks. A detailed description of countermeasures is required to achieve the goal to design resilient wireless systems. Furthermore, physical layer key agreement, using the channel model, where the channel is either a CC or an AVC, shall be investigated and optimal transmit strategies will be derived. Additionally, the secrecy capacities of the CWC and the AVWC depend on the underlying uncertainty set. The performance of a communication system should depend in a continuous way on the system parameters. This is because if small changes in the parameters were to lead to dramatic losses in performance, the approach at hand would most likely not be used. Indeed, one is interested in approaches that are robust against such variations, in the sense that small variations in the uncertainty set result in small variations in the secrecy capacity. Such a continuous dependency is desirable in the context of active adversaries who can influence the system parameters in a malicious way.The project will develop information theoretic approaches based on general channel models to model attacks on communication systems with embedded security.

• DFG Programme: Research Grants