The internet of things (IoT) encompasses various devices for monitoring, data collection, tracking people and assets, and interacting with other gadgets without human intervention. Implementing a system for predicting the development and assessing the criticality of detected attacks is essential for ensuring security in IoT interactions. This work analyses existing methods for detecting attacks, including machine learning, deep learning, and ensemble methods, and explores the federated learning (FL) method. The aim is to study FL to enhance security, develop a methodology for predicting the development of attacks, and assess their criticality in real-time. FL enables devices and the aggregation server to jointly train a common global model while keeping the original data locally on each client. We demonstrate the performance of the proposed methodology against structured query language (SQL) injection and brute force attacks using the CICIOT2023 dataset. We used accuracy and F1 score metrics to evaluate the effectiveness of our proposed methodology. As a result, the accuracy in predicting SQL injection reached 100%, and for brute force attacks, it reached 98.25%. The high rates of experimental results clearly show that the proposed FL-based attack prediction methodology can be used to ensure security in IoT interactions.

The constant emission of polluting gases is causing an urgent need for timely detection of harmful gas mixtures in the atmosphere. A method and algorithm of the determining spectral composition of gas with a gas analyzer using an artificial neural network (ANN) were suggested in the article. A small closed gas dynamic system was designed and used as an experimental bench for collecting and quantifying gas concentrations for testing the proposed method. This device was based on AS7265x and BMP180 sensors connected in parallel to a 3.3 V compatible Arduino Uno board via QWIIC. Experimental tests were conducted with air from the laboratory room, carbon dioxide (CO2), and a mixture of pure oxygen (O2) with nitrogen (N2) in a 9:1 ratio. Three ANNs with one input, one hidden and one output layer were built. The ANN had 5, 10, and 20 hidden neurons, respectively. The dataset was divided into three parts: 70% for training, 15% for validation, and 15% for testing. The mean square error (MSE) error and regression were analyzed during training. Training, testing, and validation error analysis were performed to find the optimal iteration, and the MSE versus training iteration was plotted. The best indicators of training and construction were shown by the ANN with 5 (five) hidden layers, and 16 iterations are enough to train, test and verify this neural network. To test the obtained neural network, the program code was written in the MATLAB. The proposed scheme of the gas analyzer is operable and has a high accuracy of gas detection with a given error of 3%. The results of the study can be used in the development of an industrial gas analyzer for the detection of harmful gas mixtures.

This study proposes a four-layer AI agent architecture for automating routine web security operations, integrating Large Language Model (LLM) reasoning with a hybrid Convolutional Neural Network-Long Short-Term Memory (CNN-LSTM) detection engine and implementing a Reasoning-Acting (ReAct) loop for autonomous testing with human-in-the-loop validation. The proposed architecture was empirically evaluated across 50 web applications sourced from OWASP WebGoat, DVWA, and custom-developed test environments over a six-month period. The experimental results demonstrate that the AI agent achieved an overall detection accuracy of 89.2% (95% CI: 86.4-92.0%), significantly outperforming traditional automated methods (67.4% accuracy, p < 0.001). Mean Time to Remediation (MTTR) decreased from 74.3 days to 28.5 days (61.6% reduction), while false positive rates decreased from 24.3% to 4.8%. According to these findings, AI agent-driven automation can substantially enhance the efficiency and reliability of web security testing. However, human expertise remains important for assessing complex vulnerabilities and detecting zero-day threats.

This paper shows the implementation of a physically unclonable function (PUF) of the arbiter type and its modifications, the generation of random numbers and unique identifiers, and a statistical analysis of the generated sequences of random numbers. Within the framework of this work, experiments were carried out on different SoC, SoM crystals and within the same crystal with different placement on it, experiments with PUF modeling based on machine learning were simulated. The results and conclusions provide recommendations for the optimal and safe use of PUF.

Today, the Internet of Things is one of the most popular technologies having a significant impact on various aspects of our lives, including social, commercial and economic spheres. In the areas of automation, productivity and comfort, IoT offers tremendous promise for improving the quality of life, being applied in education, smart cities and other areas. However, cyber-attacks and threats pose a serious challenge to smart applications in IoT environments. The presented work investigates the privacy and security issues in the Internet of Things (IoT). The paper analyzes various security techniques, focusing on machine learning, including Naive Bayes, Random Forest and Logistic Regression, and their application to detect anomalies and attacks in IoT networks. The article emphasizes the importance of integrating advanced technologies and ethical principles to create reliable and secure IoT systems.

The rapid deployment of 5G networks and the proliferation of Internet of Things (IoT) devices have significantly increased the complexity of urban electromagnetic radiation (EMR) environments. Conventional ground-based monitoring systems are spatially limited and unable to provide three-dimensional field characterization. This paper proposes an integrated IoT–UAV framework for high-resolution EMR monitoring, spatial reconstruction, and intelligent source classification. A four-layer architecture combining distributed sensing, edge computing, cloud analytics, and visualization is developed. A formal electromagnetic propagation model is introduced to ensure consistency between broadband exposure measurements and frequency-selective spectral analysis. A CNN–LSTM architecture is implemented for spectral–temporal source classification, achieving 95% validation accuracy across five EMR categories. Simulation-based validation demonstrates up to an eightfold improvement in spatial coverage compared to fixed ground networks while maintaining a practical anomaly detection threshold of −55 dBm in the spectrum-analysis RF chain. The proposed framework establishes a mathematically consistent and practically deployable solution for next-generation EMR monitoring systems.