Buda Health Center

In today’s automotive market, the OEM’s bring out more and more new connected features in their cars to make the users driving experience more safe and amusing than ever, but this comes with a very big price: the attackers gain even more attack surface thanks to the broad connectivity. One of the biggest problem is the approach of the manufacturers to this subject, most of them are not really implementing the much needed security solutions and testing the system’s cyber resilience unless they are forced to do it by the law or regulations. But to be able to defend themselves against the attacks, first they have to detect it. To address this problem, we developed a hybrid solution that uses multiple machine learning models to leverage the classification capabilities of these algorithms and to combine this with a human-modifiable IDS ruleset to effectively detect various attacks, while providing explainability and auditablity.

Fifth-generation (5G) communication networks provide unprecedented opportunities for military operations, offering ultra-low latency, massive device connectivity, and enhanced bandwidth. At the same time, they introduce new vulnerabilities, including risks associated with network slicing, edge computing, and the proliferation of Internet of Things (IoT) devices. Traditional perimeter-based security models are insufficient to address these threats, particularly in contested or grey-zone environments where adversaries deliberately exploit ambiguities below the threshold of armed conflict. This paper proposes a Zero Trust Architecture (ZTA) framework for 5G-enabled military communication networks. The framework integrates continuous authentication, least-privilege access, microsegmentation of network slices, and AI-driven anomaly detection to ensure resilience against insider threats and distributed denial-of-service (DDoS) attacks. Scenario-based evaluation demonstrates that Zero Trust reduces both the likelihood and impact of such attacks, maintaining command-and-control (C2) continuity even under adversarial conditions. The findings underscore the importance of embedding Zero Trust into military communication infrastructures, both as a technical solution and as a strategic enabler of NATO and EU defence resilience.

The proliferation of standalone non-public 5G networks (SNPNs) in industrial and enterprise environments has introduced complex security challenges arising from software-defined infrastructures, virtualized network functions, and service-based architectures. This paper presents an attack graph–based approach for systematically modeling and analyzing vulnerabilities across the user equipment (UE), radio access network (RAN), and core domains of 5G systems. Building on experimental research at the NIK-SOC Laboratory of Óbuda University, the study demonstrates how observed attack vectors - such as IMSI catching, replay attacks, GTP-U manipulation, and slicing misconfigurations - can be represented within formal graph structures to visualize multi-stage adversarial progressions and support risk-driven security assessment.The results show that traditional attack graph methodologies must be extended to reflect 5G-specific characteristics including network slicing, edge computing, and dynamic service exposure through the Service-Based Architecture (SBA). The paper concludes by outlining future development directions that integrate artificial intelligence, multimodal data correlation, real-time dynamic graph analysis, and Zero Trust architectural principles. Together, these enhancements aim to enable predictive and adaptive defense capabilities for next-generation standalone 5G networks.

This paper investigates the influence of conductor materials - copper (Cu), aluminum (Al), brass (CuZn), bronze (CuSn), and copper-clad aluminum (CCA) - on starting torque, current, and thermal behavior of cage-start synchronous motors. A 6 MW, 6 kV, 50 Hz salient-pole motor was analyzed using coupled electromagnetic-thermal FEM with 0.5 ms time step and refined air-gap mesh to evaluate torque-slip, current-time characteristics, Joule losses, and rotor bar temperature rise.Based on the normalized per-unit results, the maximum starting torque was 2.87 p.u. for Cu, 2.71 p.u. for CCA, and 2.46 p.u. for Al. The acceleration time (t95) values were 13.2 p.u., 10.9 p.u., and 11.5 p.u., respectively, while normalized Joule losses (E_J) were 1.00, 0.83, and 0.93 normalized units. Cu achieves maximum starting torque but longest acceleration and highest losses. Al provides smoother acceleration with reduced heating. Brass and Bronze offer superior starting torque but extended acceleration and elevated losses. CCA material presents optimal compromise: achieves 99% of Cu pull-out torque (2.71 vs. 2.87 p.u.), 17% lower losses (0.83 norm.), and 21% faster acceleration than Al (10.9 p.u.). Material selection significantly impacts starting dynamics and thermal stress. The CCA hybrid conductor provides superior performance-efficiency balance for high-power synchronous motors in industrial and power generation applications.

LoRaWAN enables low-power, long-range IoT connectivity, but its security testing suffers from a lack of accessible, labeled datasets. Anomaly and intrusion detection methods require realistic communication logs, yet real-world traffic is limited by privacy regulations and proprietary constraints. This paper introduces a conceptual framework that integrates Large Language Models (LLMs) with deterministic protocol construction to generate synthetic, privacy-safe logs. This multi-stage process couples cryptographic precision with LLM adaptability to produce diverse, compliant data for Intrusion Detection System (IDS) evaluation. We outline the framework, validation methodology, and preliminary Proof of Concept, providing a foundation for reproducible and ethical IoT security testing.

The quest for more intuitive human-computer interaction continuously seeks pathways to bridge the cognitive gap between human intentions and digital execution. Among these, gaze tracking stands as a beacon, promising to unlock hands-free control and enhanced user experiences. However, hardware-based eye trackers often compete with the inflated cost and impractical setups of standard hardware-based eye trackers. The present work focuses on the development of an open source gaze estimation system, creating a user-calibrated, real-time focus prediction system based on a commonly available web camera and the robust technology of Google’s MediaPipe Face Mesh pipeline. The proposed solution is based on Support Vector Regression (SVR) models with custom feature engineering that incorporates both facial landmarks and head pose dynamics. The paper details a calibration methodology designed to forge a bespoke mapping for each user, adapting to individual biometric and environmental variations. Furthermore, this work addresses practical implementation considerations, including the crucial step of feature scaling and the process of hyperparameter optimization via exhaustive grid search. The presented results, as illustrated in the accompanying scatter plots, demonstrate the SVR models’ ability to capture complex nonlinear relationships, marking a promising step towards robust and accessible gaze tracking in diverse interactive paradigms.

Cybersecurity education increasingly benefits from realistic, gamified simulation environments such as Cyber Ranges. A design proposal is presented for integrating an automated NIS2 compliance assessment module into an existing Active Directory (AD)-based, gamified Cyber Range used in academic training. The platform’s capabilities are aimed to be extended beyond detection and incident response to include regulatory compliance evaluation. Technical and procedural controls derived from the NIS2 directive will be automatically checked by the proposed module, such as privileged account management, password policies, logging configurations, patch management, and backup procedures. The motivation, architecture, educational workflow, and expected benefits of the module, along with potential challenges and future development directions, are outlined in this paper.

With the increasing demand for heterogeneous Unmanned Aerial Vehicle (UAV) swarms to perform complex tasks in urban environments, system design now faces major challenges, including efficient semantic understanding, flexible task planning, and the ability to dynamically adjust coordination strategies in response to evolving environmental conditions and continuously changing task requirements. To address the limitations of existing methods, this paper proposes CoordField, a coordination field agent system for coordinating heterogeneous drone swarms in complex urban scenarios. In this system, large language models (LLMs) is responsible for interpreting high-level human instructions and converting them into executable commands for the UAV swarms, such as patrol and target tracking. Subsequently, a Coordination field mechanism is proposed to guide UAV motion and task selection, enabling decentralized and adaptive allocation of emergent tasks. A total of 50 rounds of comparative testing were conducted across different models in a 2D simulation space to evaluate their performance. Experimental results demonstrate that the proposed system achieves superior performance in terms of task coverage, response time, and adaptability to dynamic changes.

The increasing integration of industrial robots into digitized manufacturing systems requires unified, lightweight, and vendor-independent data models. Existing frameworks such as the OPC UA Companion Specification for Robotics offer rich semantics and interoperability but are computationally intensive, while MQTTbased approaches provide simplicity and speed at the cost of structure and meaning. This paper proposes the Sparkplug Robotics Profile v1, a lightweight extension to the Sparkplug B standard that bridges this gap by introducing robot-specific semantics and state management within an event-driven MQTT architecture. The proposed model defines structured telemetry (RobotData), command (RobotCmd), and lifecycle (RobotBirth) message types using Google Protocol Buffers for efficient transmission and consistent interpretation across platforms. This approach enables real-time data exchange, scalable multirobot coordination, and native compatibility with XRbased digital-twin environments. The results highlight that Sparkplug Robotics Profile v1 preserves the simplicity of MQTT while providing semantic depth comparable to OPC UA, facilitating interoperable and resource-efficient robot connectivity from edge to cloud.