Cities aiming for Industry 5.0 must contend with a persistent bottleneck: urban mobility. We propose a data-driven traffic management approach that couples the Internet of Things (IoT) with large-scale analytics to improve sustainability, operational efficiency, and environmental outcomes. Our framework instruments are critical assets, signalized intersections, sensing and surveillance networks, and public transport nodes, with IoT devices that stream continuous vehicle and pedestrian activity. The resulting multimodal signals are standardized and modeled with predictive algorithms that flag emerging congestion, adapt signal timing, and support dynamic rerouting to reduce delay. Using heterogeneous, real-world datasets, we observe consistent gains in mobility performance and measurable reductions in emissions, while also enabling evidence-based planning for infrastructure investments. Beyond these technical results, the work advances a shift toward mobility systems that are adaptive, human-centric, and ecologically responsible. In sum, the study shows how IoT-enabled data pipelines and predictive models can operationalize the goals of Industry 5.0 by delivering safer, faster, and more sustainable urban travel.
The online monitoring of news related to pests and diseases is a key component of early warning systems, as it enables the localization, extraction, and processing of relevant documents to generate timely information about potential agricultural threats. The National Service for Agro-Alimentary Health, Safety and Quality (SENASICA), the agency responsible for protecting Mexico’s agricultural, aquacultural, and livestock resources, requires efficient tools to identify and analyze phytosanitary events that may pose a risk of spread within the national territory. This work presents FREM, an automated news processing system that integrates web scraping, natural language processing (NLP), and Transformer-based models to detect and analyze phytosanitary risk events. The system extracts key information from online news – including the title, date, country, pest, and affected crop – and presents it in a structured and interactive way. To enhance accuracy, FREM combines named entity recognition (NER) with Transformer-based models (RoBERTa and BERT), achieving 95
Photovoltaic modules installed in dense urban environments often operate under high irradiance and limited convective cooling, leading to elevated cell temperatures and reduced power output, requiring research on thermal dissipation to keep the temperatures under near-optimal conditions. Most data acquisition on photovoltaic modules focus on electrical variables and global temperature measurements, limiting research on new cooling systems. This work presents an integrated software and thermal imaging framework to support the experimental evaluation of passive PV cooling systems in intelligent urban environments. The proposed solution combines an Internet of Things‑based data acquisition system combined with a thermal image processing pipeline. The data acquisition system supports dual dense temperature sensor grids on the back of cooled and reference solar modules, acquisition of electrical and environmental variables, configurable polling policies, and multiple data sinks including a time‑series database and cloud‑based visualization dashboards. The thermal imaging framework provides multi‑modal calibration between radiometric images and onsite temperature measurements, online generation of temperature maps, and computation of similarity metrics and spatial statistics that characterize thermal gradients, hot spots, and differences between cooled and noncooled modules. The resulting platform is reusable and extensible, enabling rapid integration of new sensors, cooling concepts and analysis routines, and lays the groundwork for creating rich multi‑source datasets to support future model‑based and data‑driven studies of photovoltaic thermal behavior in urban setting.
The deployment of an extremely high number of base stations (BSs) is a widely recognized technology for meeting the expected performance of next-generation wireless networks (5G/B5G/6G), leading to what is known as ultradense networks (UDNs). However, UDNs have also become a major challenge regarding sustainability due to the energy consumption of such a high number of BSs. Among the different research lines proposed in the literature to address this issue, the selective deactivation of the cells installed in the BSs in periods of low traffic demand is considered in this work. It is an NP-complete nonlinear optimization problem for which we are proposing two novel linear and quadratic approximations aiming at making them affordable for mathematical programming solvers. Computational experiments are conducted by comparing the approximated models and the original nonlinear problem and using different solvers to evaluate their performance. The results suggest that the approximated models can obtain feasible solutions in reasonable computing times and are a promising research line. Finally, a prototype of a multiobjective metaheuristic that uses these approximated models as a local search operator is developed, and the preliminary results show a consistent enhancement over the canonical algorithm with an average improvement in the hypervolume quality indicator of 27.77
Solid waste management is a global challenge affecting governments and the population. It is estimated that more than two billion metric tons of municipal solid waste are generated each year, and this number is expected to increase to 3.4 billion metric tons by 2050. In this context, this article presents a Simulated Annealing metaheuristic approach for solving the municipal solid waste management problem by optimizing the collection routes of waste trucks to minimize total travel distance while remaining within vehicle capacity limits based on the capacitated vehicle routing problem. The exploration operator of the proposed approach applies two neighborhood procedures to analyze different collection routes with varying vehicle capacities. Four realistic problem instances are analyzed, considering the distances for waste truck scheduling and the volumes of waste disposed in the historic center of Cuernavaca, Mexico. The proposed approach was efficient in computing municipal solid waste collection routes and improved from 2.2 to 15.1
This article elaborates a practical mixed-integer programming (MIP) formulation for addressing the short-term hydrothermal coordination (STHTC) in Uruguay, a country with highly diversified energy mix, where renewable sources typically supply over 95
An Erratum to this paper has been published: https://doi.org/10.1134/S0361768826040018
Urbanization has intensified the demand for efficient passenger transportation systems, reinforcing the strategic role of public bus services in smart-city development. This study evaluates the operational efficiency of bus transport in 37 Brazilian municipalities (2022–2024) by applying a Simar–Wilson bootstrap data envelopment analysis with bias correction, integrated with beta and tobit regression models to points determinants of technical efficiency. The results indicated that no municipality achieved full efficiency, revealing substantial performance disparities: mean CRS efficiency was 0.597, mean VRS efficiency was 0.824, and mean scale efficiency was 0.712. The 22.7
This article addresses the problem of synthetic tabular data generation, specifically energy consumption of high performance computing resources. The model proposed by Wasserstein generative adversarial Networks is applied to create realistic synthetic data with statistical properties similar to real data, which are useful for energy efficiency studies when data are scarce and obtaining new data implies costly experiments. Different energy models were developed and applied to data obtained from experiments on real computing infrastructures, involving four high-end computing nodes. The similarity analysis of the obtained distribution showed that the generated synthetic data have similar statistical properties to real data. The quality of estimation models was satisfactory when using generated data instead of real data with generated data and improved when using both. Also, the obtained synthetic data outperforms SMONG, a generation method for regression, in terms of data independence and diversity, resulting in a more robust and less biased dataset.
Hydrothermal power-system management involves high-dimensional uncertainty (curse of dimensionality), challenging traditional stochastic programming. Stochastic dynamic programming (SDP) and scalable variants remain robust industrial baselines but face rising computational strain as grids add flexibility and variability. This paper presents an enhanced reinforcement learning (RL) framework for long-horizon hydrothermal dispatch, extending tabular approaches via a scalable Gymnasium-based representation of the Uruguayan power system. Alongside a tabular Q-learning baseline, it introduces a proximal policy optimization (PPO) agent for continuous state-action control, representing reservoir volumes and turbine releases without discretization. Optuna-based hyperparameter optimization is applied to both methods, and stochastic results use multiseed ensemble statistics. Performance is evaluated under deterministic inflows and under stochastic settings using a nonhomogeneous Markov inflow model (in-sample) and century-scale historical chronicles (out-of-sample). Within the adopted environment, PPO achieves a 3.96
Self-organizing swarm systems of autonomous agents (peers) achieve their missions collectively relying on solely internal interactions without centralized coordination. The inherent nature of swarms lies in their decentralized, collective behavior supported by ad-hoc architecture, enabling agents to dynamically join or leave the swarm without any impact on its overall mission performance. This collectively demonstrating intelligence and problem-solving abilities surpass those of individual agents. However, self-organizing nature of swarms introduces challenges in ensuring consistency and explainability of run-time decision-making processes due to probable “black box” nature of the embedded AI/ML models. Peers in the swarm are intended to make adaptive decisions guided by distributed intelligence to handle events while maintaining interpretable reasoning pathways for reinforcement learning upon completion of the mission. Explainable event handling assumes providing transparent insights into the swarm’s emergent behavior and decision-making strategies, which can be complex and unpredictable, and cannot be inferred from the properties of individual peers. The paper introduces a framework that ensures explainability of the swarm behavior by introducing a formal language for specifying event signatures, adaptive probabilistic methods for run-time event detection, and an event log manager that incorporates new event categories while continuously updating the event probability space.
Most critical software developed today relies on multiple threads, a technique that is notoriously difficult to implement correctly. This complexity often leads to concurrency-related bugs, such as data races. Furthermore, emerging trends in robotics and unmanned vehicle software add further complications. These systems are not only multi threaded but also interact with hardware, require specific OS environments, utilize IPC, and demand rigorous testing. In this paper, we build upon a multi core version of S2E to present an algorithm devised to better detect data races in these complex environments. To do so, we leverage S2E’s full-system emulation, its plugin infrastructure, and our custom CPU core scheduler.
The rapid evolution of network threats has surpassed the capabilities of traditional intrusion detection systems and static network configurations. In this work, we present an integrated real-time anomaly detection and remediation framework that combines machine learning (ML) and software-defined networking (SDN) technologies in a graphical network simulator (GNS3) emulated environment. We create our dataset containing both benign and malicious traffic, including web browsing, DNS queries, port scans, and DoS attacks, by orchestrating flows through OpenvSwitch and emulated routers and switches. The monitored traffic undergoes flow-level feature extraction to generate packet counts, inter-arrival times, and byte distributions, which are used to train and evaluate both supervised and unsupervised ML models, including random forests, support vector machines (SVMs), and isolated forests. This research proposes a new replicable framework based on GNS3 for adaptive network security using ML and SDN, as well as publicly available datasets and administrator scripts for further experimentation.
Hierarchical federated fog computing architectures often have serious difficulties when a cluster is in a heavy load for a long time and it does not have enough local capacity for the requests. In this paper a new model of intercluster resource sharing is presented which makes it possible to redistribute the load dynamically. It includes two main strategies: (1) request redirection through direct gateway-to-gateway connections, and (2) temporary node migration between clusters. To be in control of the system, the architecture uses intelligent master node coordination, applying a limit of 30
This article researches the problem of calibrating probabilistic predictions in machine learning models for detecting obfuscated malware. Isotonic regression was used as the research method. The research was conducted in a virtual environment using obfuscated malware samples. The analysis showed that isotonic regression, when used in neural network calibration, improves the detection accuracy of obfuscated malware using machine learning methods. The datasets used were CIC-IDS 2017, EMBER 2018 and datasets generated using malware obtained from public sources. To evaluate the calibration accuracy of the machine learning model under noisy and class-imbalanced conditions, the following evaluation metrics were used: area under the receiver operating characteristic curve, area under the precision-recall curve, brier score, log loss, and expected calibration error. The results of the research confirm the applicability of isotonic regression to imbalanced and noisy datasets typical for obfuscated malware detection tasks.
This paper presents a mathematical analysis of a multiserver queueing model in which each arriving task requires the simultaneous use of multiple servers. The model reflects realistic scenarios in modern computing environments where parallel task execution is essential. Tasks arrive according to a Poisson process and require a randomly selected number of servers and an exponentially distributed service time. A key contribution of this study is the formulation of a recursive structure for analyzing the system’s busy period—the duration until the system becomes idle. Using combinatorial event conditions, the recursive equations are construct that describe the dynamics of task completion and waiting behavior. The mathematical expectation of the busy period is derived by incorporating transition probabilities and system parameters. This model extends existing results in multiserver queueing systems by explicitly accounting for simultaneous service requirements. The proposed analysis offers insights into system performance metrics and lays the groundwork for future computational and simulation-based studies.
Resource-constrained IoT devices require secure communication protocols that minimize both latency and energy consumption while maintaining cryptographic security guarantees. This paper presents a hardware-accelerated implementation of TLS 1.3 with 0-RTT (zero round-trip time) session resumption for IoT devices using ARM CryptoCell-310 cryptographic acceleration. Our implementation addresses the fundamental challenge of enabling immediate encrypted communication upon reconnection without sacrificing forward secrecy or replay protection. The system leverages hardware-isolated key derivation, temporal nonce validation, and efficient PSK caching to achieve sub-100 ms connection establishment while maintaining security properties equivalent to full TLS 1.3 handshakes after ephemeral key exchange completion. Experimental evaluation with 30 nRF9161 devices equipped with integrated CryptoCell-310 demonstrates average connection establishment latency of 73ms (standard deviation 8.2 ms), representing an 87
Efficient matrix multiplication remains one of the most important computational tasks in scientific computing, engineering simulations and data analysis. This work presents a high-performance implementation of double-precision general matrix multiplication designed for modern ×86 processors. The main objective is to approach the practical performance limits achievable through software-level optimization by exploiting register-level parallelism, cache hierarchy characteristics and thread-level parallel execution. The proposed method is built around an efficiently designed 6 × 8 micro kernel that utilizes vector registers and fused multiply-add operations. A two-level cache blocking strategy is used to increase data reuse and reduce memory traffic. The implementation also employs parallel processing with manual control of work distribution among processor cores to improve scalability on systems with different numbers of cores and cache sizes. The study evaluates performance on two contemporary x86 processors with different core counts and cache configurations. The implementation is compared to widely used numerical libraries. Results show that the presented approach achieves solid sustained performance and consistently outperforms the NumPy/OpenBLAS backend, while reaching a substantial fraction of the throughput provided by highly optimized libraries. Additional experiments include benchmarking and a performance model that explains the observed behavior in terms of arithmetic throughput and memory bandwidth. The work demonstrates that a combination of vector-optimized micro kernels, cache-aware blocking and multi threading can provide a portable and efficient solution for double-precision matrix multiplication on current ×86 architectures.
Zero-knowledge range proofs (ZKRPs) allow a prover to convince a verifier that a committed value lies in a given interval without revealing the value itself. Such proofs are widely used in financial applications and cryptocurrencies. This paper presents a new noninteractive ZKRP protocol derived from an order-revealing encryption (ORE) construction, enabling comparisons over encrypted data. The proposed protocol adapts a large-domain ORE structure to obtain an efficient range-proof mechanism and introduces a corresponding key-management/setup procedure. We discuss correctness, security considerations under standard ORE leakage, and provide performance and memory estimates.
In this paper, we propose a method for constructing classification models for three or more ordered classes when the initial sample is sufficient only to construct a binary classifier capable of recognizing the top and bottom classes relative to the class order. The classification model is constructed based on the available information about the distribution of all classes in the general population and the information about the frequency of positive result occurrence in the existing binary classifier for each ordered class. This method is based on the missing sample simulation modeling technique, which uses the Monte Carlo approach. To illustrate the proposed method, we consider a model that solves the problem of detecting the degree of carbohydrate metabolism disorder (CMD) across three classes with sufficiently high quality, based on a series of electrocardiograms (ECGs) with 11 or more ECG measurements per patient.
