This work proposes the Horus-Cyber Detection for Smart Grids (Horus-CDS), a tool based on temporal neural networks for detecting anomalies in the network traffic of a Smart Grid (SG). The model employs Temporal Convolutional Networks (TCNs), which leverage dilated causal convolutions to capture sequential patterns and enhance real-time inference. Unlike conventional approaches, the tool enables greater parallelization, lower latency, and an interface for monitoring and decision-making. The tool was validated in two scenarios: a simulated SG environment and a real-world deployment at an brazilian Smart grid managing company, where its resilience against Distributed Denial-of-Service(DDoS) and Distributed Denial-of-Control (DDoC) attacks was tested. The results demonstrated that Horus-CDS outperformed other approaches, ensuring the operational continuity of power distribution systems.

Abstract Detecting deepfakes has become one of the most pressing challenges in the field of cybersecurity and digital forensics. With the increasing use of deepfake technology for malicious purposes, such as misinformation and identity theft, there is a critical need for effective detection methods. This work introduces ARNet ( A lexNet & R esNet Net work), a hybrid deep learning model specifically designed for deepfake detection. ARNet combines the computational efficiency of AlexNet with the robust feature extraction capabilities of ResNet, addressing the limitations of traditional models. The architecture of ARNet integrates residual connections and a global pooling mechanism to improve detection accuracy while reducing computational cost and processing time. The proposed model demonstrates superior performance in both accuracy and inference speed, outperforming popular models such as AlexNet, ResNet, and MobileNet in deepfake datasets. Our results highlight the effectiveness of ARNet in both low-resource environments and large-scale deepfake detection tasks.

Abstract This study empirically assesses how effectively two Large Language Models (LLMs), and , can transform terse feature-request titles from open-source software (OSS) issue trackers into well-formed software requirements. It further examines how prompt-engineering strategies shape requirement quality and evaluates a scalable “LLM-as-a-Judge” approach for automated quality assessment based on three ISO/IEC/IEEE 29148:2018 quality attributes: Unambiguity , Verifiability , and Singularity . We extract 150 feature-request titles from five OSS repositories and pair each title with every combination of two LLMs and three prompt styles, producing 900 candidate requirements. An independent evaluator LLM rates each requirement on these three quality attributes using a Likert scale and provides textual rationales. Ordinal data are analyzed with descriptive statistics and non-parametric tests, complemented by thematic analysis of the evaluator’s explanations. A targeted human validation study with five evaluators on a stratified sample of 50 requirements assesses the reliability of the automated judge. LLMs often produce high-quality requirements; however, scores vary with input clarity and prompt design. Few-shot prompting consistently boosts Singularity , while Expert Identity prompting sometimes raises Verifiability but frequently harms Singularity . shows a modest but significant edge in Singularity . Qualitative review echoes these patterns, revealing trade-offs between added detail and focus. The LLM-as-a-Judge protocol delivers consistent, scalable evaluations whose scores correlate significantly with aggregate human judgment ( $$p < 0.05$$ for all three attributes). Modern LLMs can expedite the drafting of initial software requirements from informal OSS inputs, but their output quality hinges on careful prompt selection and the inherent clarity of the source title. Prompt effects are model-dependent and may introduce trade-offs among quality attributes, so human oversight remains indispensable for refinement. The LLM-as-a-Judge framework proves a practical, human-validated technique for large-scale evaluation, enabling rapid, reproducible insights into LLM-driven requirements engineering workflows.

This article tackles the critical challenge of optimizing the buffer pool, a core component of Database Management Systems (DBMS) that caches frequently accessed data pages, where manual configuration often proves inadequate in dynamic, high-demand environments. To address this gap, we present an automated, data-driven methodology that combines advanced Machine Learning techniques with Bayesian optimization. Our approach follows a systematic three-phase process: (1) Exploratory Factor Analysis (EFA) coupled with K-means clustering to uncover latent factors and reduce the dimensionality of performance metrics; (2) LASSO regression to identify and rank the most influential configuration parameters; and (3) Bayesian optimization using Gaussian Process modeling with acquisition functions (Expected Improvement, Probability of Improvement, and Upper Confidence Bound) to fine-tune buffer pool settings. The main contributions of this work include a novel automated framework for DBMS tuning that simplifies configuration, enhances memory management, and boosts performance efficiency. We validated the proposed solution using real workloads collected from a large-scale financial system in Latin America, achieving up to a 45% reduction in maximum data access wait times, confirming improvements in performance and scalability.

Wireless Underground Sensor Networks (WUSNs) are complex systems comprised of subterranean sensors interconnected through wireless communication technologies. These networks fulfill a crucial role in monitoring subsurface environments. However, they grapple with a formidable challenge concerning their Network Lifetime (NL), which can be defined as the maximum duration over which the network remains operational and thus connected to a designated observation area. Given the paramount significance of prolonging NL to ensure comprehensive coverage of the observed region, the deployment of wireless power transfer stands out as a preeminent solution for augmenting NL. Nonetheless, the existing sleep-wakeup protocols have not been originally engineered to support this paradigm, which has subsequently resulted in suboptimal network performance. Therefore, we present a study to introduce a novel sleep-wakeup protocol explicitly tailored for wireless power transfer in WUSNs with the overarching aim of optimizing the network’s operational lifetime called Sleep-Wakeup Power-Transfer Media Access Control (SWPTMAC). The evaluation of SWPTMAC has been conducted through comprehensive simulations leveraging the Castália simulator. The empirical findings disclosed an average improvement of approximately 24% when contrasted against incumbent protocols in the domain.

Research in Natural Language Processing, bio-medicine, and computer vision achieved excellent results in machine learning due to the success of the Transformer-based models. However, these excellent results depend on the labeled high-quality and large-scale datasets. If one of these requirements is not met, the model may lack generalization ability, and its performance will be unsatisfactory. To address these issues, this research proposes a Generic Multimodal Gradient-Based Meta Framework (GeMGF) trained from scratch to avoid language bias, learns from a few data, and reduces the model degradation trained on a finite dataset. GeMGF was evaluated using the benchmark dataset CUB-200-2011 for the text and image classification tasks. The results show that GeMGF outperforms the state-of-the-art models with 93.2% accuracy. GeMGF is simple, efficient, and adaptable to other data modalities and fields.

The continuous emergence of new applications for Internet-connected road vehicles is imposing unprecedented re-source demand. Motivated by the incorporation of ever more resources into vehicles, this is a trend that, on the downside, is causing vehicular networks to become increasingly more challenging to manage. Departing from the proposition that computing capabilities can help overcome resource allocation problems in vehicular clouds (VCs), in this paper, we formulate ALTAIC, a coalition game to maximize resource utilization while dynamically load-balancing the usage among the VCs. First, we define a Shapley value-based strategy to determine the order in which the tasks are allocated. Then, with the marginal contribution of each task calculated, we employ a simple queue to allocate the tasks in VCs using these values. Finally, we conduct a comparative performance analysis of ALTAIC and relevant approaches. Simulation results show that the proposed solution allocates more tasks than the others and reduces 27.12% the load average of the VCs.

Even with advances in Smart Grids and their cybersecurity recommendations, recent attacks on automation and protection systems of these structures show that it is still necessary to investigate this research problem. With that in mind, this work proposes STRAYER: a SmarT aRchitecture Against cYbERattacks to reduce the vulnerability of automation equipment in Smart Grids. STRAYER integrates cybersecurity for monitoring and shielding access, interoperability for maintaining communication between equipment/devices, and risk management for maintaining reliability and preventing real-time cyberattacks on Smart Grids. To validate the STRAYER, we built a prototype commonly used in smart grids. The results showed that STRAYER increases the security efficiency compared to the traditional architecture, reducing the amount of infected equipment and the undue access time to Smart Grids. In addition to the reductions in the amount of IED’s affected by invasions, it was also possible to notice that STRAYER avoided the collapse of a Smart Grid, having only minimal and reversible losses, unlike the traditional architecture.

本评论回顾1998年提出的“一次性学习”(once learning, OLM)机制, 和随后出现的用于图像分类的“一瞥学习”(one-shot learning)以及用于目标检测的“你仅看一次”(you only look once, YOLO)。基于目前人工智能(AI)研究现状, 提出将其划分为以下子学科: 人工类人智能、人工机器智能、人工仿生智能和人工量子智能。这些被认为是AI研发的主要方向, 并按以下分类标准区分: (1)以类人、机器、仿生或量子计算为本的AI研发;(2)升维或降维的信息输入;(3)小样本或大数据知识学习。

This work proposes to solve the maximal covering location problem of the Mobile Operations Coordination Center (CCOp Mv), which aims to support the operational command of the Brazilian Army. This problem consists of selecting, in a limited region and with poor communication infrastructure to the ground troops’s operating area, the positions of vehicles equipped with Base Transceiver Station (BTS), the amount needed, and theirs transmission power to be set that maximizes the coverage area and reduce the interference due to the overlap of signals. For this reason, analytical modeling based on the mixed-integer linear problem was proposed that guided two optimization solutions: (i) E-ALLOCATOR – Exact ALLOCATiOn seRvice; and (ii) M-ALLOCATOR – Metaheuristic ALLOCATiOn seRvice. The solutions were evaluated in a scenario that employs CCOp Mv to support a rescue operation based on the tragedy in January 2019 in Brumadinho-MG and compared with a heuristic. The performance evaluation results show evidence of efficiencies in terms of quality and resource savings of the proposed solutions. Furthermore, E-ALLOCATOR has been proven to be suitable for a low workload on the network. At the same time, M-ALLOCATOR is suitable for scenarios with a high workload providing almost optimal solutions within the adequate computational time for all problem instances.

Efficient content distribution over vehicular ad hoc networks (VANETs) is a challenging task due to highly topology changes caused by vehicle mobility. In this context, Vehicle Named-Data Networks (VNDN) architecture improves the performance and reliability in delivering content by providing content-centric network communication and caching capabilities. However, the success of VNDN architecture depends on mitigating the broadcast storm problem during the cache discovery process, where the network performance impairment occurs due to the waste of resources generated. In this paper, we propose a receiver-based cache discovery protocol based on degree-centrality for VNDN, called CLYMENE. The protocol paves the way for efficient content distribution by minimizing the broadcast storm problem. Simulation results show that CLYMENE enhances the cache discovery by 80.59% while allowing a content delivery rate of 39.49% and reducing the number of transmissions in the cache discovery process at 70.65% compared to existing protocols.

One of the main challenges in urban development faced by large cities is related to traffic jam. Despite increasing efforts to maximize the vehicle flow in large cities, to provide greater accuracy to estimate the traffic jam and to maximize the flow of vehicles in the transport infrastructure, without increasing the overhead of information on the control-related network, still consist in issues to be investigated. Therefore, using artificial intelligence method, we propose a solution of inter-vehicle communication for estimating the congestion level to maximize the vehicle traffic flow in the transport system, called TRAFFIC. For this, we modeled an ensemble of classifiers to estimate the congestion level using TRAFFIC. Hence, the ensemble classification is used as an input to the proposed dissemination mechanism, through which information is propagated between the vehicles. By comparing TRAFFIC with other studies in the literature, our solution has advanced the state of the art with new contributions as follows: (i) increase in the success rate for estimating the traffic congestion level; (ii) reduction in travel time, fuel consumption and CO2 emission of the vehicle; and (iii) high coverage rate with higher propagation of the message, maintaining a low packet transmission rate.

The development of new smart objects for the sensing and actuation of a given place or environment led both the academia and industry to research and propose new protocols and intelligent systems to support such objects. One of the systems that has been gaining prominence is the smart residential environments. In this context, homes are equipped with smart objects to manage the living resources. However, managing such objects in residential environments requires data contextualization, i.e. collecting data from heterogeneous devices and actuate on the environment through context information generated from such data. To solve this problem, we propose an intelligent decision system based on the fog computing paradigm, which provides an efficient management of residential applications. The proposed solution is evaluated both in simulated and real environments. When compared with other studies from the literature in a simulated environment, the proposed solution shows a higher success rate with a lower delay in the decision-making process, higher efficiency in information dissemination with a lower overhead in the communication infrastructure, and increased robustness in processing with a lower power consumption. These results are also observed when considering a real environment evaluation.

This paper proposes a fault-tolerant indoor localization system that employs Recurrent Neural Networks (RNNs) for the localization task. A decision module is designed to detect failures and this is responsible for the allocation of RNNs that are suitable for each situation. As well as the fault-tolerant system, several architectures and models for RNNs are exploited in the system: Gated Recurrent Unit (GRU), Long Short-Term Memory (LSTM) and Simple RNN. The system uses as inputs a collection of Wi-Fi Received Signal Strength Indication (RSSI) signals, and the RNN classifies the position of an agent on the basis of this collection. A fault-tolerant mechanism has been designed to handle two types of failures: (i) momentary failure, and (ii) permanent failure. The results show that the RNNs are suitable for tackling the problem and that the whole system is reliable when employed for a series of failures.

The growth in many countries of the population in need of healthcare and with reduced mobility in many countries shows the demand for the development of assistive technologies to cater for this public, especially when they require home treatment after being discharged from the hospital. To this end, interactive applications on mobile devices are often integrated into intelligent environments. Such environments usually have limited resources, which are not capable of processing great volumes of data and can expend much energy due to devices being in communication to a cloud. Some approaches have tried to minimize these problems by using fog microdatacenter networks to provide high computational capabilities. However, full outsourcing of the data analysis to a microfog can generate a reduced level of accuracy and adaptability. In this work, we propose a healthcare system that uses data offloading to increase performance in an IoT-based microfog, providing resources and improving health monitoring. The main challenge of the proposed system is to provide high data processing with low latency in an environment with limited resources. Therefore, the main contribution of this work is to design an offloading algorithm to ensure resource provision in a microfog and synchronize the complexity of data processing through a healthcare environment architecture. We validated and evaluated the system using two interactive applications of individualized monitoring: (1) recognition of people using images and (2) fall detection using the combination of sensors (accelerometer and gyroscope) on a smartwatch and smartphone. Our system improves by 54% and 15% on the processing time of the user recognition and Fall Decision applications, respectively. In addition, it showed promising results, notably (a) high accuracy in identifying individuals, as well as detecting their mobility; and (b) efficiency when implemented in devices with scarce resources.

Multilevel approaches aim at reducing the cost of a target algorithm over a given network by applying it to a coarsened (or reduced) version of the original network. They have been successfully employed in a variety of problems, most notably community detection. However, current solutions are not directly applicable to bipartite networks and the literature lacks studies that illustrate their application for solving multilevel optimization problems in such networks. This article addresses this gap and introduces a multilevel optimization approach for bipartite networks and the implementation of a general multilevel framework including novel algorithms for coarsening and uncorsening, applicable to a variety of problems. We analyze how the proposed multilevel strategy affects the topological features of bipartite networks and show that a controlled coarsening strategy can preserve properties such as degree and clustering coefficient centralities. The applicability of the general framework is illustrated in two optimization problems, one for solving the Barber's modularity for community detection and the second for dimensionality reduction in text classification. We show that the solutions thus obtained are statistically equivalent, regarding accuracy, to those of conventional approaches, whilst requiring considerably lower execution times.

Currently, there is an increasing number of patients that are treated in-home, mainly in countries such as Japan, USA and Europe. As well as this, the number of elderly people has increased significantly in the last 15 years and these people are often treated in-home and at times enter into a critical situation that may require help (e.g. when facing an accident, or becoming depressed). Advances in ubiquitous computing and the Internet of Things (IoT) have provided efficient and cheap equipments that include wireless communication and cameras, such as smartphones or embedded devices like Raspberry Pi. Embedded computing enables the deployment of Health Smart Homes (HSH) that can enhance in-home medical treatment. The use of camera and image processing on IoT is still an application that has not been fully explored in the literature, especially in the context of HSH. Although use of images has been widely exploited to address issues such as safety and surveillance in the house, they have been little employed to assist patients and/or elderly people as part of the home-care systems. In our view, these images can help nurses or caregivers to assist patients in need of timely help, and the implementation of this application can be extremely easy and cheap when aided by IoT technologies. This article discusses the use of patient images and emotional detection to assist patients and elderly people within an in-home healthcare context. We also discuss the existing literature and show that most of the studies in this area do not make use of images for the purpose of monitoring patients. In addition, there are few studies that take into account the patient's emotional state, which is crucial for them to be able to recover from a disease. Finally, we outline our prototype which runs on multiple computing platforms and show results that demonstrate the feasibility of our approach.

Cities with a large number of people are currently facing urban mobility problems, especially the problem of traffic congestions. This not only has an adverse effect on the economy of the city, but also impairs the quality of life of its citizens. One measure that can be adopted to mitigate these problems is the use of systems that help identify, reduce, and/or avoid these traffic jams, such as intelligent transport systems. In this context, we propose an intelligent traffic information system called UCONDES, which is based on inter-vehicle communications and can be applied to detect and reduce congestion in urban centers. Simulation results shows that, when compared to original vehicular mobility trace, our solution reduces the average trip time, and the overall CO2 emission and fuel consumption. More specifically, the average travel time for drivers was reduced by approximately 26%, resulting in a reduction of fuel consumption by 23% and the CO2 emission by 25%.

This work proposes an intelligent decision system for a residential infrastructure based on wireless sensors and actuator networks, called ResiDI. ResiDI is equipped with battery-powered nodes to ensure that they are deployable anywhere in the house without the need for wiring, drilling or any pre-existing infrastructure. The key intelligence of ResiDI is distributed in the decider nodes, which are able to make decisions locally without the need to send traffic from the sensor nodes to the sink. The network intelligence core is based on a neural network that seeks to improve the accuracy of the decision-making, together with a temporal correlation mechanism that is targeted at reducing the energy consumption. When compared with an approach adopted in the literature, the results show that ResiDI is efficient in different scenarios in all evaluations performed.

The use of pesticides in agriculture is essential to maintain the quality of large-scale production. The spraying of these products by using aircraft speeds up the process and prevents compacting of the soil. However, adverse weather conditions (e.g. The speed and direction of the wind) can impair the effectiveness of the spraying of pesticides in a target crop field. Thus, there is a risk that the pesticide can drift to neighboring crop fields. It is believed that a large amount of all the pesticide used in the world drifts outside of the target crop field and only a small amount is effective in controlling pests. However, with increased precision in the spraying, it is possible to reduce the amount of pesticide used and improve the quality of agricultural products as well as mitigate the risk of environmental damage. With this objective, this paper proposes a methodology based on Particle Swarm Optimization (PSO) for the fine-tuning of control rules during the spraying of pesticides in crop fields. This methodology can be employed with speed and efficiency and achieve good results by taking account of the weather conditions reported by a Wireless Sensor Network (WSN). In this scenario, the UAV becomes a mobile node of the WSN that is able to make personalized decisions for each crop field. The experiments that were carried out show that the optimization methodology proposed is able to reduce the drift of pesticides by fine-tuning of control rules.