Cyber-physical systems play a central role in critical infrastructures, particularly in power plants, factories, and water treatment facilities. These systems face increasingly sophisticated threats that can have severe operational and societal consequences. A Cyber-Physical Honeynet (CPHoneynet) is a network of honeypots designed to replicate the operation of a real cyber-physical environment. When attackers interact with it, a CPHoneynet enables the collection of valuable threat intelligence regarding attacker methods, tools, and attack vectors.The domain lacks a recent and comprehensive structuring. Therefore, this survey identifies, reviews, and analyzes the central publications published between 2010 and June 2026 through a Systematic Literature Review methodology. The survey analyzes 451 publications and focuses on three major themes: realism, attractiveness, and emerging applications of CPHoneynets.The analysis reveals that IoT and ICS are the dominant application domains, realism remains the primary research challenge, and attractiveness remains comparatively underexplored despite its direct impact on threat-intelligence collection. The survey synthesizes a reference architecture, proposes a comprehensive taxonomy, and identifies evaluation metrics for both realism and attractiveness. It further highlights emerging directions such as Artificial Intelligence, Digital Twins, and Large Language Models, while showing that standardized evaluation frameworks remain largely absent from the literature.Finally, the survey identifies open challenges and research gaps related to scalability, deployment complexity, operational safety, and adaptive cyber-deception, providing a structured foundation for future research and practical deployment of CPHoneynets.
Abstract Floating treatment wetlands (FTWs) have been increasingly studied and applied over the last 15 years, yet the conditions for their effective implementation are still not fully understood. To refine current understanding and identify conditions that maximize particulate pollutant removal, a flume experiment with a real FTW was conducted to investigate how the mean flow velocity ( U 1 = 0.018 m · s − 1 ; U 2 = 0.083 m · s − 1 ; and U 3 = 0.171 m · s − 1 ) and root-network characteristics (i.e., R = ratio of the water column occupied by the FTW over the total pond water depth) influence the flow structure and associated suspended solids (SS) retention within FTWs. Flow acceleration (up to twice the mean flow velocity) and deceleration occurred underneath and downstream, respectively, of the FTW having a rooting depth ratio R of 0.56 ( FTW 0.56 ). A substantial fraction of the flow was diverted toward the flume bottom—particularly under U 1 —reducing root–flow contact and particle entrapment. Under higher velocities, the flow could penetrate deeper into the root network, increasing opportunities for particles’ entrapment and increased treatment benefits. The results indicated that FTW 0.56 could enhance pond treatment efficiency by up to ∼ 35 % , with greater benefits at higher velocities. This is partly due to reduced performances of conventional ponds to settle particles under these higher flow velocities ( U 2 and U 3 ). Increasing the rooting depth ratio (e.g., 0.87) would further improve efficiency, especially under low flow (up to three times higher removal efficiency), where FTW 0.56 offered limited additional benefit because of greater short-circuiting around the root network. Overall, the results of the present study suggest that careful consideration should be given to the flow the FTWs are subject to, the original pond design and its performance toward natural settling, to evaluate whether inclusion of FTWs would provide sufficient additional treatment benefits.
Constant efforts to make head-mounted displays (HMDs) more compact and low-power consumption have naturally led to the idea of integrating them into contact lenses. Mojo Vision was the first company to take on this ambitious challenge, though with limited success. While enabling technologies, particularly in the micro-display sector, are advancing rapidly, it remains essential to understand the fundamental optical constraints, including those related to retinal physiology, in order to identify the most viable path forward. In this work, we review and evaluate the main eligible projection approaches used in HMDs, especially in Augmented Reality (AR) and Head-Up Display (HUD) systems, and consider how these concepts could be adapted to contact lenses with a greatly reduced form factor while maintaining comparable functionality. (c) 2026 Optica Publishing Group under the terms of the Optica Open Access Publishing Agreement
Generating accurate extremes from an observational data set is crucial when seeking to estimate risks associated with the occurrence of future extremes which could be larger than those already observed. Applications range from the occurrence of natural disasters to financial crashes. Generative approaches from the machine learning community do not apply to extreme samples without careful adaptation. Besides, asymptotic results from extreme value theory (EVT) give a theoretical framework to model multivariate extreme events, especially through the notion of multivariate regular variation. Bridging these two fields, this paper details a variational autoencoder (VAE) approach for sampling multivariate heavy-tailed distributions, i.e., distributions likely to have extremes of particularly large intensities. We illustrate the relevance of our approach on a synthetic data set and on a real data set of discharge measurements along the Danube river network. The latter shows the potential of our approach for flood risks' assessment. In addition to outperforming the standard VAE for the tested data sets, we also provide a comparison with a competing EVT-based generative approach. On the tested cases, our approach improves the learning of the dependency structure between extremes.
Generative AI and misinformation research has evolved since our 2024 survey. This paper presents an updated perspective, transitioning from literature review to practical countermeasures. We report on changes in the threat landscape, including improved AI-generated content through Large Language Models (LLMs) and multimodal systems. Central to this work are our practical contributions: JudgeGPT, a platform for evaluating human perception of AI-generated news, and RogueGPT, a controlled stimulus generation engine for research. Together, these tools form an experimental pipeline for studying how humans perceive and detect AI-generated misinformation. Our findings show that detection capabilities have improved, but the competition between generation and detection continues. We discuss mitigation strategies including LLM-based detection, inoculation approaches, and the dual-use nature of generative AI. This work contributes to research addressing the adverse impacts of AI on information quality.
