High expansion of urban population and infrastructure complemented by recent geopolitical world events, triggered an increasingly alarming number of threats. Law enforcement authorities are now challenged to redesign societal security concepts. Although current technological advances eased the information access, e.g., via video surveillance cameras, satellite data, drones, wearables, manual analysis of such big and diverse data to extract strategic knowledge is not a solution anymore. There is the critical need for automatic solutions. Artificial Intelligence (AI) and the breakthrough of deep neural networks opened a new perspective for providing such solutions with a human-grade accuracy. Here, we provide a snapshot of our AI research for counter terrorism on automatic person and object identification, retrieval of speech intelligence, and dissimulated behavior analysis. The research was carried out during the UEFISCDI SPIA-VA research project at CAMPUS Research Center, University “Politehnica” of Bucharest, with the participation of UTI Grup and of the Military Equipment and Technologies Research Agency (ACTTM), having as public beneficiary the Protection Guard Service, Romania (SPP).

The Internet of Things (IoT) became widely utilized during last years, due to the large number of object that are connected to it - not only computers, but also humans, sensors and actuators. In the near future, a large number of objects will communicate to each other in a way never experienced before by the humankind. Connecting all these objects together will come with a wide area of challenges. One of the most important issues concerns trust. In majority of cases, centralized architectures are used to solve the problem, but the enormous number of object that can be connected in IoT will make almost impossible to manage and implement these solutions. Blockchain technology offers powerful solutions for decentralized architectures, which can be the future of IoT systems. A strong solution needs to be flexible to changes and must offer powerful resources to be successful. A hardware platform that uses FPGAs is suitable for those systems. In this paper we propose a solution for securing IoT systems using blockchain technology, implemented on a FPGA based architecture.

Mission-critical services (MCX) comprise a standardized suite of capabilities including Mission-Critical Push-to-Talk (MCPTT), MCVideo, and MCData, designed to meet stringent requirements for availability, reliability, latency, security, and Quality of Service (QoS). These services are essential for public safety, emergency response, and other critical infrastructure domains, where communication performance directly affects operational effectiveness. Integration into 4G and 5G mobile networks, supported by targeted standardization efforts, has extended broadband capabilities to mission-critical environments. 5G networks provide the technical foundations for MCX through ultra-low latency (below 1 ms), high availability (99.999%), broadband throughput over 100 Mbps per user, deterministic QoS via network slicing, massive device connectivity (over one million devices per square kilometer), and seamless Non-Terrestrial Network (NTN) integration. Technical enablers such as Proximity Services (ProSe), network slicing, and Ultra-Reliable Low-Latency Communications (URLLC) are fundamental to delivering these capabilities. This paper reviews MCX architectures, service frameworks, and protocols, relating MCPTT, MCData, and MCVideo to the key performance requirements defined in ITU-T M.2377-2. It also examines the frozen features of 3GPP Release 19, including enhancements to MC services, NTN integration, Reduced Capability device support, sub-meter positioning, extended network slicing for Public Protection and Disaster Relief (PPDR), and strengthened security mechanisms. Finally, the study addresses challenges such as standard maturity, interoperability, and deterministic QoS, identifying research priorities toward 6G readiness. By consolidating advances from standards bodies, research initiatives, and deployments, this work serves as a technical reference for scalable, secure, and standards-compliant MCX solutions in current and future networks.

The IoT market has grown significantly in recent years, and it is estimated that it will continue to do so. For this reason, the need to identify new solutions to ensure security is vital for the future development in this field. Inadequate sources of entropy are one of the factors that negatively influence security. In this study, inspired by NIST’s latest entropy estimation recommendations, we proposed a methodology for analyzing and validating a sensor-based entropy source, highlighted by an innovative experiment design. Moreover, the proposed solution is analyzed in terms of resistance to multiple types of attacks. Following an analysis of the influence of sensor characteristics and settings on the entropy rate, we obtain a maximum entropy value of 0.63 per bit, and a throughput of 3.12 Kb/s, even when no motion is applied on the sensors. Our results show that a stable and resistant entropy source can be built based on the data obtained from the sensors. Our assessment of the proposed entropy source also achieves a higher complexity than previous studies, in terms of the variety of approached situations and the types of the performed experiments.

Speaker verification methods are various and use different types of features, but each system alone do not perform satisfactory results. This paper makes a comparison of different features and methods for score fusion for an independent speaker verification application. Several types of spectral features are used as speaker data. The scores obtained with these types of features were fusioned with combination methods (as: mean, sum, max, min, weighted sum) and classification methods (as: SVM, linear discriminant). The experiments were performed on a laboratory registered database for Romanian language and demonstrate that fusion methods outperformed the baseline GMM-UBM method.

In modern data processing systems where large amount of data needs to be processed in order to achieve its designated performance, the use of hardware accelerators becomes mandatory. Every FPGA manufacturer proposes its unique method of transferring data between embedded processors (e.g. Picoblaze, Microblaze, Nios, PowerPC, ARM Cortex) and hardware peripheral made from FPGA's reconfigurable resources or high speed dedicated communication modules. These hardware peripherals can perform a lot of different functionalities, like cryptographic and multimedia accelerators or digital signal coprocessors. This paper presents the design and implementation of several data busses along with its associated communication protocols that interconnects embedded processors and cryptographic hardware accelerator. At the end of this paper the implementation results are compared to provide developers the information needed to choose the right method to transfer data in new designs.