Dimitrios Schoinianakis

Privacy-preserving machine learning and encrypted statistics increasingly require evaluating long chains of matrix products directly on ciphertexts. In the CKKS homomorphic encryption scheme, however, every multiplication amplifies noise and enlarges ciphertexts, so the available precision budget is exhausted after only a few products. This work establishes c F H E (compressed FHE), a unified analytical and empirical framework that integrates low-rank matrix factorization techniques into the CKKS homomorphic encryption scheme. Its central idea is a co-design: instead of tuning the low-rank approximation and the cryptographic parameters in isolation, c F H E balances the two error sources against each other so that neither wastes precision. Theoretical bounds are derived for the accumulation of relative error across sequences of factorized matrices, leading to an explicit expression for the attainable computation depth as a function of target accuracy, norm amplification behavior, and per-layer approximation quality. Extensions to tree-based evaluation structures are also formulated, allowing depth to scale logarithmically with the number of factors. Concretely, this co-design is realized as a precision-balancing model that, for a target accuracy, automatically selects the CKKS parameters: the polynomial modulus degree, the modulus chain, and the scaling factor. Experimental evaluations demonstrate that encrypted low-rank matrix multiplications achieve both significant runtime improvements and reduction of ciphertext sizes over direct or tree-based encrypted multiplications while maintaining the prescribed accuracy. c F H E is agnostic to other CKKS optimizations and can be combined with them for further gains.

In this work, the conventional cryptographic algorithms used in the 5G Core are replaced with post-quantum alternatives and the practical impact of this transition is evaluated. Using a simulation environment, we model the registration and deregistration of varying numbers of user equipments (UEs) and measure the resulting effects on bandwidth consumption and latency. Our results show that the deployment of post-quantum cryptographic algorithms has a measurable effect on performance, but that this effect is small, and perhaps more crucially, that the extra overhead needed in terms of computation and bandwidth does not have any substantial impact on the usability of the network and the efficiency of its network functions. Overall the experimental results in this work corroborate earlier research: the 5G Core is technically able to support post-quantum cryptography without any inherent issues connected to the increased computational overhead or larger message size.

Quantum Computers (QCs) differ radically from traditional computers and can efficiently solve mathematical problems fundamental to our current cryptographic algorithms. Although existing QCs need to accommodate more qubits to break cryptographic algorithms, the concern of ''Store-Now-Decrypt-Later'' (i.e., adversaries store encrypted data today and decrypt them once powerful QCs become available) highlights the necessity to adopt quantum-safe approaches as soon as possible. In this work, we investigate the performance impact of Post-Quantum Cryptography (PQC) on TLS 1.3. Different signature algorithms and key agreements (as proposed by the National Institute of Standards and Technology (NIST)) are examined through black- and white-box measurements to get precise handshake latencies and computational costs per participating library. We emulated loss, bandwidth, and delay to analyze constrained environments. Our results reveal that HQC and Kyber are on par with our current state-of-the-art, while Dilithium and Falcon are even faster. We observed no performance drawback from using hybrid algorithms; moreover, on higher NIST security levels, PQC outperformed any algorithm in use today. Hence, we conclude that post-quantum TLS is suitable for adoption in today's systems.

The continuous growth of the number of Internet of Things (IoT) devices and their inclusion to public and private infrastructures has introduced new applciations to the market and our day-to-day life. At the same time, these devices create a potential threat to personal and public security. This may be easily understood either due to the sensitivity of the collected data, or by our dependability to the devices’ operation. Considering that most IoT devices are of low cost and are used for various tasks, such as monitoring people or controlling indoor environmental conditions, the security factor should be enhanced. This paper presents the exploitation of side-channel attack technique for protecting low-cost smart devices in an intuitive way. The work aims to extend the dataset provided to an Intrusion Detection Systems (IDS) in order to achieve a higher accuracy in anomaly detection. Thus, along with typical data provided to an IDS, such as network traffic, transmitted packets, CPU usage, etc., it is proposed to include information regarding the device’s physical state and behaviour such as its power consumption, the supply current, the emitted heat, etc. Awareness of the typical operation of a smart device in terms of operation and functionality may prove valuable, since any deviation may warn of an operational or functional anomaly. In this paper, the deviation (either increase or decrease) of the supply current is exploited for this reason. This work aimed to affect the intrusion detection process of IoT and proposes for consideration new inputs of interest with a collateral interest of study. In parallel, malfunction of the device is also detected, extending this work’s application to issues of reliability and maintainability. The results present 100% attack detection and this is the first time that a low-cost security solution suitable for every type of target devices is presented.

It is clearly redundant to extensively discuss the upcoming era of 5G and pervasive computing or, in other words, the Internet of Things (IoT). Scientific and industrial communities have been analyzing the design challenges, architectures, and security requirements of our brandnew, all-connected world for quite some time now, and standardization is still a work in progress.

With the increasing adoption of end-to-end encryption in industrial systems, the risk of distributing hidden malware by exploiting encrypted channels gradually turns to a major concern. Due to encryption, the stateof-the-art, signature-based mechanisms might fail to detect malware sufficiently, thus new approaches are required. In this work, a method for malware detection in encrypted traffic based on Machine Learning is presented. A supervised learning approach is adopted and the efficiency of the solution is demonstrated by a set of exhaustive simulations. Further considerations for incorporating the proposed method in a reference industrial network are also discussed.