Sistema Regional de Información
en línea para Revistas Científicas de América Latina,
el Caribe, España y Portugal

In network topology, the sinkhole attacks ensue while a malicious node deceives it’s adjust mode in the routing traffic and it results a predominant interruptions like amplified latency, depletion of energy with conceded data reliability. Moreover a decentralized feature of MANETs are specifically exposed to treats and emphasize the necessity of distinct solutions. In this study, Federated Learning is utilized to improve the security and privacy through empowering nodes to train the model deprived of distribution complex data. Then, every node gathers information about the local routing and contributes towards an inclusive model, which captures behaviour of the entire network when conserving its specific privacy. Further, the Hierarchical Deep Belief Network Convolutional Neural Network (HDBNCNN) algorithm has analysed the accumulated data in detecting the anomalies revealing the sinkhole activity centred on learning routing patterns. Besides, detection, the model can implement mitigation approaches, which includes redirecting traffic from conceded nodes and informing the routing tables to reinforce flexibility from imminent attacks. The proposed study aims to enhance the reliability and efficiency of MANETs in attack situations when minimizing overhead from malicious traffic management. By combining progressive ML methods through decentralized network structure, hence the study expressively provides a secure communication in MANETs and contributes a scalable and effective results through evaluating its performance metrics. 

The rapid advancement in the deployment of Cyber-Physical Systems (CPS) within autonomous vehicles has taken numerous of security challenges. These challenges primarily arise from the intricate and interconnected nature of CPS components, which can lead to vulnerabilities during operation. The complexity of these systems often results in connected devices undergoing errors, and there is an occurrence of risk at various cyber-attacks. In addition to security threats reliable and concerns the impact of overall CPS functionality and safety. Besides, the security and reliability problems have stimulates in protection and detection methodologies. A significant instance of improvement in this field is the incorporation of Machine Learning (ML) techniques into CPS, which has effectively streamlined operations and reduced the complexity associated with autonomous vehicles. This paper aims to provide a comprehensive review of the security measures pertinent to autonomous vehicles, specifically focusing on the integration of ML techniques within CPS. It delves into several protection approaches considered to address the multiple reliability along with security challenges handled by vehicles. Furthermore, the paper surveys the crucial feature of risk assessment associated to autonomous vehicles, emphasising the performance can considerably improve when risk assessments are confirmed by large datasets in the training phase, predominantly while engaging learning models. In general, the findings emphasise the significance of leveraging progressive data analytics and ML to improve the security and reliability of autonomous vehicles, confirming the safe functions in progressively multifaceted atmospheres.

Urban image classification is a challenging task due to factors such as occlusions, diverse appearances, and ambiguous visual contexts. This paper addresses these challenges by consolidating and discussing methodologies focused on tree-powerline interactions, a critical problem for urban safety and infrastructure planning. Using state-of-the-art Deep Learning Networks (DLNs), prior studies evaluated performance on a curated dataset of 11,000 labeled street-level images collected and annotated with the INvestigate and Analyze a CITY (INACITY) platform and the Street-Level Image Labeler (SLIL) tool. These evaluations revealed significant limitations in baseline models, with a maximum accuracy of 74.6\% on this dataset. To address these limitations, earlier work introduced a ``Challenging'' class and applied semi-supervised learning techniques, including the Noisy Student protocol and Focal Loss, achieving recall rates of 83.7\% for positive cases and 78.8\% for negative cases. Dimensionality reduction using the Self-Supervised Neural Projection (SSNP) method was explored for clustering and visualization tasks, while the Supervised Decision Boundary Maps (SDBM) technique improved classifier interpretability by addressing scalability and clarity in decision boundary visualizations. This paper consolidates these contributions, presenting an integrated narrative that highlights tools, datasets, and methodologies for advancing urban image classification. These efforts provide scalable, interpretable, and accurate solutions for urban safety and infrastructure management challenges.

Diabetic retinopathy is a vision impairment associated with diabetes mellitus, a common medical condition. Retinographic imaging analysis of retinal fundus images is the predominant method of diagnosing this eye complication. Recent advances in deep learning show effectiveness in detecting diabetic retinopathy, rivaling the diagnostic accuracy of human inspection. However, the success of these deep learning methods largely depends on the architecture model. In our research, we introduce a method for training a deep learning model using a Low-Rank Adaptation (LoRA) technique to differentiate between three stages of diabetic retinopathy: no diabetic retinopathy, non-proliferative diabetic retinopathy, and proliferative diabetic retinopathy. LoRA significantly reduces the training parameters needed by employing a low-ank representation. Our experimental results on four data sets of different sizes show that the LoRA model delivers top-tier results in classifying stages of diabetic retinopathy.

Software-defined networking (SDN) and network functions virtualization (NFV) represent promising technologies for on-demand services that require the development of flexible multicast transmission mechanisms incorporating data processing functions at the network nodes. The multicast routing problem in NFV-SDN networks aims to compute multicast routing trees and deploy virtual network functions (VNFs) to satisfy traffic demands while optimizing resource utilization and ensuring equitable data transmission. Due to the inherent computational complexity and the presence of conflicting objective functions, this paper formulates multicast routing and VNF placement as a multiobjective optimization problem (MOP), specifically minimizing the total resource cost and the maximum transmission delay variance.Within this context, this study develops solutions based on Multi-objective Evolutionary Algorithms (MOEAs).Simulations performed on test instances demonstrate the efficacy of the proposed methodologies, yielding efficient and non-dominated solutions when compared to a state-of-the-art single-objective approach.

Network reliability computation is an NP-hard problem which has attracted much attention in literature. This problem consists in, given a network where the links may fail or operate with known probabilities, to compute the probability that a given subset of nodes (known as terminals) are connected by the operational links. Given the difficulty to compute the exact value of the network reliability, an alternative which has been much explored in the literature is the use of Monte Carlo estimation methods. In this work we discuss Permutation Monte Carlo, a highly efficient network reliability estimation method. The method is prone to numerical limitations for networks with a high number of links. We discuss this situation and present a simple way to rewrite the algorithm's calculations to make it numerically more stable. We also present a variant of Permutation Monte Carlo implementation for the particular case where all the links in the network under study share the same failure probability distribution (homogeneous network). For this type of network the Permutation Monte Carlo implementation can be redesigned to be much more efficient. We present some experimental test results proving that both proposed implementation variants are extremely efficient.

In this paper, we delve into the necessity of applying coupled application-container elasticity in the context of resource allocation within multi-tenant scenarios, specifically targeting CNN training procedures on multi-core architectures. Our focus is primarily on elucidating the process of endowing a parallel application, exemplified by Tensorflow, with dynamic thread elasticity support. Additionally, we emphasize the criticality of harnessing this functionality in tandem with container resource management by container orchestrators to avert undesirable resource oversubscription and under-utilization effects, all the while optimizing core occupancy. The empirical findings comprise a wide range of resource allocation and dynamic re-allocation strategies, and reinforce the compelling case for elasticity support not only at the container level but also at the application level. These findings underscore the substantial improvements achieved in terms of time-to-solution, resource utilization, and productivity across diverse workloads.

Semantic Technologies for the Internet of Things provide opportunities to improve interoperability, considering the multidimensional challenges. This study proposes the development of an enhanced ontology that integrates different ontologies in the IoT domain, including key aspects such as interoperability, security, privacy, energy efficiency, and ethics. A formal methodology for ontology design and construction was used in four phases: ontology requirements, design, construction, and evaluation. As a result, an enhanced ontology called IoTO++ was built. This five-module ontology improves operational efficiency in IoT environments. Furthermore, it facilitates the automation of data-driven decisions and actions, thereby enabling faster adaptation to changing environments. The validation emphasizes competence and quality criteria. OQuaRE framework was also used considering the quality model and metrics. IoTO++ achieves 100% in modularity and testability. In the rest of the subcategories it exceeds 89%, demonstrating its maintainability.

This paper summarizes the thesis ”MEC Placement Problem in Protected 5G Networks”, which tackles the question of where to position Multi-access edge computing nodes (MECs) in 5G networks to achieve stringent 5G requirements at a minimum cost.The thesis formulated the MEC placement problem with traditional protection schemes to provide redundancy in the event of MEC failures, and proposed a new scheme called ’enhanced shared protection 1 : N : K’. The work investigated the impact on the costs of new variations of the MEC placement approaches considering different distributions of VNFs forming SFC requests. The results indicate that the enhanced protection scheme is cost-effective, and the VNF distribution impacts significantly on the cost.

Cyber ranges are specialized computer systems designed to simulate realistic cybersecurity scenarios for training purposes. An in-depth evaluation process is essential for determining whether users have met their training objectives. This article examines the implementation of Tectonic, an academic cyber range, along with a thorough assessment methodology for both offensive and defensive scenarios. This approach employs process mining to analyze training activities from various perspectives. The methodology has been applied to training sessions conducted within Tectonic.