The maximum likelihood attack strategy is known to be the optimal attack strategy for an eavesdropper in a wiretap channel scenario with additive white Gaussian noise channels under the distinguishing security criterion. The main drawback of this optimal attack is its high computational complexity. While this complexity doesn’t hinder the eavesdropper since he has unlimited computing power, it does present a significant challenge for legitimate parties. For them, it is extremely difficult, if not impossible, to estimate the outcome of the optimal attacker strategy to validate the secrecy of their communication system. In this paper, we introduce a low complexity method for generating upper and lower bounds on the attack performance of the eavesdropper to validate the security against the maximum likelihood attack strategy. We theoretically establish that the derived bounds represent valid constraints on the attack success probability under suitable constraints. The validation method is based on list generation and can be used for any linear block code. Furthermore, we propose a list generation algorithm for this validation method and show different ways to further reduce the complexity. We compare the proposed validation method with state-of-the-art attack strategies in numerical simulations for various error-correcting codes.

This paper compares different algorithms to check semantic security on AWGN wiretap channels. Each algorithm provides upper and lower bounds on the performance of an attack strategy that is close to the best attack strategy. The advantage of these algorithms is that they have lower computational complexity compared to the best attack strategy. We also show that the proposed algorithms can be further improved by including cyclic redundancy check bits and parity check bits generated in the algorithms, for example, when polar codes or LDPC codes according to the 5G standard are used in the coding layer. Finally, we show the compatibility of the algorithms for both polar codes and LDPC codes.

In this paper we show a way to check semantic security for AWGN wiretap channels. We introduce low complexity decoding methods that provide upper and lower bounds to the performance of an attack strategy that closely resembles the best attack strategy, which has the problem of large computational complexity. We show the assumptions under which these methods can be applied and compare simulation results of the bounds to the performance of the best attack strategy. We use a seeded modular coding scheme, which consists of a coding layer and a security layer. For the coding layer we use polar codes, but the method is neither restricted to the seeded modular coding scheme nor to the polar codes.

We experimentally verify the information-theoretic security of a seeded modular code for the AWGN wiretap channel consisting of a security layer, an error-correction layer and a modulation layer. The security layer is given by a universal family of hash functions. In the error-correction layer and the modulation layer we use polar codes and QAM, respectively. The eavesdropper uses the maximum likelihood (ML) test as an attack strategy. We analyze the security in different communication scenarios using simulations. We show that for small blocklengths the advantage (security measure) at the eavesdropper in the corresponding scenario can be close to 0 for suitable code parameters. Additional insights gathered from the simulation results include the impact of code parameters and seed choice on security.

In this paper we use a seeded modular coding scheme for implementing physical layer security in a wiretap scenario. This modular scheme consists of a traditional coding layer and a security layer. For the traditional coding layer, we use a polar code. We evaluate the performance of the seeded modular coding scheme in an experimental setup with software defined radios and compare these results to simulation results. In order to assess the secrecy level of the scheme, we employ the distinguishing security metric. In our experiments, we compare the distinguishing error rate for different seeds and block lengths.

For streaming applications, we consider parallel burst erasure channels in the presence of an eavesdropper. The legitimate receiver must perfectly recover each source symbol subject to a decoding delay constraint without the eavesdropper gaining any information from his observation. For a certain class of code parameters, we propose delay-optimal M-link codes that recover multiple bursts of erasures of a limited length, and where the codes provide perfect security even if the eavesdropper can observe a link of his choice. Our codes achieve the maximum secrecy rate for the channel model.

We consider transmissions of secure messages over a burst erasure wiretap channel under decoding delay constraint. For block codes we introduce and study delay optimal secure burst erasure correcting (DO-SBE) codes that provide perfect security and recover a burst of erasures of a limited length with minimum possible delay. Our explicit constructions of DO-SBE block codes achieve maximum secrecy rate. We also consider a model of a burst erasure wiretap channel for the streaming setup, where in any sliding window of a given size, in a stream of encoded source packets, the eavesdropper is able to observe packets in an interval of a given size. For that model we obtain an information theoretic upper bound on the secrecy rate for delay optimal streaming codes. We show that our block codes can be used for construction of delay optimal burst erasure correcting streaming codes which provide perfect security and meet the upper bound for a certain class of code parameters.

In this paper we consider a wiretap channel II with an active eavesdropper. Aggarwal et al (2009) were the first who studied the Ozarow-Wyner's binary wiretap channel II in the presence of an active eavesdropper. They derived achievable secrecy rates for two modification models where the eavesdropper can erase/replace the bits he observes. The existence of better achievable rates remains an open problem. Here we study a model with a less powerful eavesdropper. The eavesdropper is able now to observe any interval of μ symbol positions and erase the symbols in any interval of l positions of a transmitted codeword. We present an explicit construction of nested linear codes that achieve maximum secrecy rate for the finite length coding regime, with perfect secrecy and zero-error decoding, for any admissible code parameters.