This paper outlines a cognitive reasoning platform which integrates semantic reasoning with fuzzy logic, and the Analytic Hierarchy Process, (AHP) for producing intelligence comparable to human intelligence for making the intricate decisions. By simulating human cognitive processes, it interprets relational hierarchies and synthesizes imprecise assessments into structured metrics, generating autonomous decisions by processing a variety of inputs. Context-aware adaptation that imitates human cognition and the close linkage of computational rigor and semantic reasoning are two important developments. 89% of experimental outcomes match expert evaluations, outperforming conventional techniques. It uses Human Resources (HR) evaluation as the main application context, even though the cognitive reasoning platform is architecturally broad and appropriate for a variety of domains needing structured judgments, such as healthcare triage, project risk analysis, and academic performance review. This explains why HR examples are used frequently in the manuscript as both an instructive use case and a validation area.
Twitter is a popular platform for the JavaScript community to share their opinions and thoughts. These tweets contain valuable knowledge about the Node Package Manager (NPM) ecosystem. Links are an example of such knowledge. Therefore, a thorough investigation into the role of links in tweets by NPM maintainers can reveal noteworthy trends and patterns of information dissemination. This study investigates the prevalence, targets, purposes, categories, and decay of links in tweets shared by NPM maintainers to understand how these links connect to the larger NPM ecosystem. To accomplish our goal, we conducted a mixed method analysis of 18,408 links. Our study found that links are prevalent in tweets and majority of the tweets are related to package management. The links are mostly blog posts, tutorials, and articles, and their primary function is to provide tweet context. Surprisingly, 80% of the links are unique, while repeatedly mentioned links make up one-fifth (20%). In addition, github.com was the most frequent domain other than twitter.com, and approximately 770 (5%) of the links shared in tweets are dead. Among the dead links, the domain github.com has the highest number of these links. The results of this study indicate that referencing external resources using links is prevalent practice for the NPM maintainers community on Twitter space. In addition, we identified some research gaps and open challenges that can guide future research efforts.
Primarily two areas of research progress are made in this paper: Discrete Wavelet Transform based Multicarrier Modulation (DWT-MCM), and the Discrete Wavelet Multitone based Multicarrier Modulation (DWMT-MCM). In this work, an efficient design of one and two-dimensional DWT-MCM is proposed for DWMT-MCM. Different channel conditions for different environments are simulated. Various wavelet families are employed with DWMT-MCM system. Additionally, bit error probability is compared to the familiar Orthogonal Frequency Division Multiplexing (OFDM) system for AWGN, Rayleigh flat fading, and frequency selective fading channels. The obtained outcomes demonstrate that the probability of error in the DWMT-MCM system is improved in manyscenarios, even though the guard interval (Cyclic Prefix (CP)) is not used, which leads to increase the spectral efficiency. The results also demonstrate that there is an error floor for some families of wavelet trans form. The obtained simulation results for DWMT-MCM system have significantly improved compared with conventional OFDM system in many scenarios and the proposed approach able to mitigate error floors introduced by channel variation effectively as well as the insufficient CP with high bandwidth efficiency.
paper proposed HyFlaNK, a hybrid federated learning threat detection framework combining Convolutional Neural Network (CNN) and Long Short Term Memory (LSTM) in a Flower-based federated learning model integrated with Apache Kafka to simulate live data ingestion, model updates, and feedback loops. The system is scalable, supports self-learning, real-time evaluation using TensorFlow/Keras for model creation and Flower for federated orchestration. Performance analysis was conducted to evaluate the model using accuracy, loss, precision, recall, F1-score, and ROC-AUC. Confusion matrices generated for the clients and global model shows good classification performance. Experimental results show consistently high performance across the local models and the aggregated global model, achieving accuracies above 99.7% and ROC-AUC of 1.0, highlighting the effectiveness and reliability of HyFlaNK. Aline plot of accuracy and loss over federated rounds revealed a consistent upward trend inaccuracy and a corresponding decline in loss, validating the capability of HyFlaNK to maintain high detection performance while preserving data privacy in a distributed environment. Additionally, a comprehensive performance evaluation comparing HyFlaNK with a traditional Random Forest-based approach further underscores its superior accuracy, precision, and scalability, making it amore robust solution for real-time threat detection in decentralized environments.
Social media platforms became the hub for conveying messages and responses to current events, but contain more harmful aspects by influencing negative stereotypes, spreading false information, and enabling misogyny in some scenarios. Detecting misogynistic language in social media is challenging for code-mixed languages due to demographic variations, transliteration, and noisy texts. The model has been evaluated on a Hindi-English code-mixed dataset of misogynistic comments. We proposed a hybrid misogyny classification model that combines byte-level ByT5 encoder embeddings with a Bidirectional Gated Recurrent Unit (BiGRU) augmented by the Bahdanau attention mechanism. ByT5 produces robust, subword-agnostic representations that reduce sensitivity to spelling variations and code switching. The BiGRU captures contextual sequential patterns and bidirectional dependencies, while attention emphasizes the most indicative tokens of abusive intent. It is demonstrated that the proposed hybrid model outperforms recurrent neural networks with static and dynamic embeddings, producing more stable misogyny predictions in low-resource and noisy texts.
memes, which commonly spread humor, ideas, or even harmful materials such as hate and propaganda, are a significant part of t he Internet culture. The m eme consists of an image and supporting text. Memotion Analysis, or meme Emotion Analysis, is automatic processing of memes using artificial intelligence. U nimodal solutions are now being taken over by multimodal solutions such as feature concatenation, weighted fusion, and Gated Multimodal Unit (GMU) for better Memotion Analysis. In this work, we proposed two deep learning based multimodal models for meme emotion classification. In the first model, we used ResNet and DeBERTa separately for single image-text fusion. In the second 'MemeCLIP' model an integrated CLIP-based representation with GMU employing a gated mechanism for adaptive visual and text feature fusion is used. In contrast to simple concatenation techniques, GMU demonstrates superior capability in extracting fine-grained emo-tional cues embedded in Memes. For the Memotion Analysis task 8 of SemEval-2020 competition, the CLIP-based model 'MemeCLIP' achieved a F1-score of 0.65, closely followed by the ResNet+DeBERTa model with a score of 0.64, compared to the SemEval baseline of 0.5118. These findings demonstrate the strength of selectively regulating modality contributions.
Protecting sensitive data has become an urgent necessity in wireless communication, which predates the emergence of Internet development, where signals transmitted over a channel are distorted by fading, noise, attackers, and interference, which leads to the loss of sensitive information. This paper introduces an efficient approach for image steganography over a wireless channel based on the New Radio (NR) polar code with Differential Chaos Shift Keying (DCSK) modulation to protect secret information over a Rayleigh fading model in the presence of an Additive White Gaussian Noise (AWGN) channel. The DCSK is a non-coherent detection method widely used in broad-spectrum communications due to its high security and effectiveness against fading. On the other hand, steganography is one of the most essential techniques for protecting secret information by embedding it in multimedia. In this work, the secret images are hidden in cover stego using the Least Significant Bits (LSB) technique. The simulation outcomes indicate that the Bit Error Rate (BER) is improved by 8 dB gain in this design over the AWGN channel and approximately 4 dB over the combination channel of AWGN and Rayleigh fading. Besides that, the Peak Signal to Noise Ratio (PSNR) reached 80.2750 dB at 26 dB SNR with 32 Spread Factor (SF). This work introduces a hybrid data protection system that consists of steganography and DCSK techniques, where the data has effectively been retrieved by the receiver.
paper presents a new solt-enhanced dual-band microstrip ring resonator for non-invasive glucose estimation fabricated on FR-4. The slot placed beneath the sample to increase electric-field participation in the sensing volume. The structure supports two well-separated resonances near 2.4 GHz and 5 GHz whose minima shift monotonically with glucose concentration across 0-400 mg/dL, enabling linear inverse modeling and ratio metric features that mitigate common-mode drift. Five regression models were evaluated-single-band linear and quadratic (for each band) and a dual-band linear mapping using (f1,f2)-on a dataset generated with a Debye-based permittivity parameterization; the dual-band linear model delivered the best accuracy-robustness trade-off with training MAE = 8.53 mg/dL, RMSE = 10.13 mg/dL, R2=0.9938, and leave-one-out crossvalidation R2=0.9881, MAE = 12.23 mg/dL, RMSE = 14.11 mg/dL, whereas a quadratic dual-band alternative over fit. Frequency-detection resolution and normalized sensitivity are reported for both bands to support fair comparison among configurations, and the slot-enhanced layout achieves higher sensitivity without sacrificing compactness, indicating a practical path toward calibration-efficient bio sensing. Index Terms-Dual-band resonator; FR-4; Microwave bio
summarization systems are becoming more popular due to the growing volume of text information in many real-life applications. This paper presents a novel approach to summarising a single document by modeling it as an optimization problem and using the Reptile Search Algorithm (RSA) to solve it. This algorithm is inspired by crocodile hunting behaviour, which includes two main steps encircling and hunting. The encircling step requires high walking or belly walking phases while the hunting step requires coordination or cooperation. In this study, we propose a binary version of this algorithm called BRSA-ESDS to implement an automatic text summarization system by choosing a subset of the sentences in the original text. This algorithm optimizes an objective function to preserve linguistic quality based on many factors, including readability and consistency in the compressed summary while improving its coverage. This model ensures the diversity and coverage of selected sentences in the summary by optimizing a harmonic average of the objective function factors. Additionally, this model controls the summary's length to ensure its readability. The results are compared with state-of-the-art approaches using ROUGE measures on the Document Understanding Conference (DUC) benchmark datasets. According to ROUGE scores, our approach consistently performs better than other methods.
The scarcity of annotated images significantly hin ders the development of robust deep learning models for early wildfire smoke detection. Traditional augmentation methods, such as rotation or mirroring, are often insufficient. This is particularly true for detecting subtle smoke formations at long distances, where smoke is often barely perceptible even to human observers. Existing smoke datasets predominantly feature devel oped smoke plumes or closer views, making them unsuitable for training models for this critical early-phase detection. To address this, we propose generating synthetic images using Generative Adversarial Networks. Unlike typical GAN applications that aim for high-fidelity object representation, our objective is different. We synthesize realistic, fuzzy images of subtle, distant smoke — blurred and blended with the background — yet retaining characteristic features essential for classifier training. We propose a GAN architecture based on a modified Super-Resolution GAN, specifically adapted without B residual blocks, in order to produce realistic images of smoke at long distances. Experimental evaluation demonstrates that augmenting datasets with GAN generated smoke images significantly improves the performance of classifiers in detecting early-stage wildfire smoke, affirming the utility of GANs for data enhancement even when generating low quality, realistic imagery. This method mitigates data scarcity, offering a viable solution for training effective early wildfire detection systems.
The growing availability of low-cost positioning technologies enables broader access to advanced Positioning, Navigation, and Timing (PNT) services for regional and global applications. Building on our previous work, which introduced a Low Earth Orbit (LEO)-PNT optimization framework for Africa, we extend this framework to the European and Arctic region and demonstrate that well-designed regional constellations with only 65-85 satellites can match the performance of much larger LEO-PNT systems, such as those comprising 240282 satellites. We evaluate both regional and global LEO-PNT performance in outdoor and indoor environments for several constellations: three newly designed configurations optimized for European coverage, a previously optimized African constellation, and seven benchmark constellations.
antenna parameters is essential to improve coverage and quality of service in wireless cellular networks. Electrical tilt and azimuth strongly influence the Reference Signal Received Power (RSRP) and the Signal-toInterference-plus-Noise Ratio (SINR), which are key Quality of Service (QoS) indicators. Conventional approaches, based on field measurements and manual tuning, are costly and inefficient in dynamic urban environments. To overcome these limitations, this paper proposes a hybrid framework integrating a high-fidelity electromagnetic simulator and a Geographic Information System to realistically model radio propagation and accurately evaluate performance. The optimization targets electrical tilt-remotely adjustable via the Remote Electrical Tilt (RET) mechanism- as well as azimuth, which requires on-site reconfiguration. The search relies on advanced metaheuristics, namely Genetic Algorithms and Artificial Immune Systems, ensuring efficient exploration and robust convergence. Experiments conducted on the LTE-Advanced network of Algeria Telecom Mobile - Mobilis in Oran demonstrate performance gains of up to 23% in the fitness function, which combines an average RSRP greater than -85 dBm and an average SINR greater than 10 dB, compared to the operator's configurations obtained through manual optimization based on drive tests. These results confirm the effectiveness of the proposed approach for optimizing antenna parameters in complex urban environments. Beyond performance gains, the proposed framework reduces operational costs and is compatible with Self-Organizing Networks (SON) and RET systems, providing a scalable solution for current and future cellular networks in large-scale Internet of Things (IoT) scenarios.
This paper investigates the performance of reflec tive intelligent surface (RIS)-assisted bi-directional full-duplex (FD) systems considering hardware impairments (HIs), and self interference (SI) under Nakagami-m fading. Unlike previous research that assumes ideal hardware or Rayleigh fading, this work derives closed-form expressions for outage probability, asymptotic diversity, throughput, and energy efficiency. The analysis employs a novel Gamma approximation, based on moment-matching and a first-order Laguerre series expansion, to accurately model Nakagami-m product channels. The derived formulations allow efficient performance comparisons, including RIS-assisted half-duplex (HD) systems with and without hard ware impairments. Numerical results reveal that using HIs/SI re sults in error floors. Increasing the number of reflective elements N, reduces the outage probability and enhances throughput and energy efficiency. The Monte Carlo simulations validate the analytical findings. Therefore, the proposed framework provides valuable design insights into the trade-offs among transceiver non-idealities, RIS size, and overall system robustness, serving as a critical reference for designing energy-efficient, hardware aware RIS-assisted FD communication architectures in future wireless networks.
the fifth generation (5G) and beyond networks, packet scheduling plays a critical role to fulfil the requirement of quality-of-service (QoS) while minimizing energy consumption. However, existing schedulers are struggling to balance between delay-sensitive traffic, energy efficiency, and fairness, especially in heterogeneous environments that have mixed traffic classes. This paper proposes EnergySlot, a lightweight energy-aware scheduler based on fixed slot partitioning between high-priority traffic and standard-priority. The method assigns a specified percentage of the frame to high-priority packets, to ensure it will be serviced, while reserving the remainder for standard-priority traffic to prevent starvation and improve fairness. Unlike traditional approaches that focus exclusively on either strict prioritization or energy minimization, EnergySlot introduces a balanced framework. Simulation results show that EnergySlot reduces deadline violations by 57% compared to first-come-first-served (FCFS) and EnergyOnly, achieves QoS levels within 5% of strict-priority scheduling, and consumes approximately 41% less energy than PriorityOnly. Additionally, it ensures 50% coverage for standard-priority traffic, which in turn significantly outperforms methods that neglect lower-priority users. These results confirm that EnergySlot provides a robust trade-off between delay, fairness, and energy consumption that makes it well-suited for real-time scheduling in heterogeneous 5G environments.
performance of the Affine Projection Algorithm (APA) and its Set-Membership (SM-APA) variants in adaptive channel equalization is limited by step-size selection. While fixed step-sizes offer a poor trade-off, many variable step-size (VSS) schemes fail to achieve optimal performance, particularly in non-stationary environments. This paper introduces a novel step-size control mechanism by integrating the Distributed Step-Size LMS (DSSLMS) strategy into APA and SM-APA frameworks. The method employs a two-stage process where a base step-size is dynamically adapted based on signal and error characteristics and then distributed to form an adaptive step-size vector, creating the DSSLMS-APA and two DSSLMS-SM-APA algorithms. The proposed algorithms are evaluated against conventional and state-of-the-art VSS benchmarks. Simulation results confirm superior performance. In a stationary environment, the proposed DSSLMS-SM-APA achieves a steady-state MSE of (2 & times; 10-3), outperforming both the conventional APA (3 & times; 10-3) and the benchmark VSS methods (VSS-APA-MSE and VSS-APA-EC). In three distinct non-stationary scenarios, the proposed DSSLMS-based algorithms exhibit the fastest reconvergence and maintain the lowest post-change MSE, successfully tracking system dynamics where other methods struggle. The algorithms provide a robust, efficient solution for step-size adaptation that balances convergence speed, steady-state accuracy, and tracking agility, making them suitable for real-time channel equalization applications.
inpainting is rapidly developing currently due to the progress of deep learning techniques and generative models. This has led to a loss of integrity in digital content, where inpainting techniques enable the production of highly realistic altered images that cannot be easily detected. To address these risks, this paper proposes a new deep learning-based image inpainting forensic network called W2SC-Net. The proposed architecture adopts a W-structured encoder-decoder design that integrates the Swin transformers with convolutional neural networks (CNNs). Specifically, the encoder block consists of two parallel streams to effectively extract both local textures and global contextual information. The decoder block is connected with the downsampling stages to enable accurate reconstruction. In addition, a high-pass filtered enhancement block is employed to highlight inpainting artifacts. Extensive experiments demonstrate not only the high detection performance of the proposed model but also its strong generalization capability. Although it was trained only on one inpainting method, it can accurately detect image manipulations across ten inpainting methods and diverse image datasets. Moreover, the W2SC-Net's robustness against anti-forensics attacks is further improved by introducing an state-of-the-art forensic approaches in terms of F1-score and AUC evaluation metrics.
the atomic broadcast problem, which consists in delivering messages atomically to multiple processes, is critical for maintaining consistency in distributed systems. This paper introduces a new atomic broadcast protocol for mobile distributed systems. Existing consensus-based protocols have drawbacks such as performance dependencies and specific consensus requirements. The proposed protocol addresses these limitations by allowing processes to cooperate through consecutive rounds to agree on the message delivery sequence without additional building blocks. It tolerates crash failures and uses an unreliable failure detector for fault-tolerance. The protocol simplicity is enhanced by the use of the lozenge S failure detector to select a decentralized round leader. Performance evaluation carried out by simulation considers message overhead, latency, energy consumption, and additionally examines the impact of the consensus block on the atomic broadcast protocol.
