Bulletin of Electrical Engineering and Informatics
Not a member yet
2885 research outputs found
Sort by
Handling partial occlusions in facial expression recognition with variational autoencoder
Full text linkFacial expression recognition (FER) is essential in various domains such as healthcare, road safety, and marketing, where real-time emotional feedback is crucial. Despite advancements in controlled settings such as well-lit, frontal, and unobstructed conditions, FER still faces significant challenges in natural, unconstrained environments. One of the most difficult issues is the presence of occlusions, which obscure key facial features. To overcome this, multiple strategies have been proposed, generally falling into two categories: those focused on analyzing visible facial regions and those aimed at reconstructing hidden facial features. In this study, we present a variational autoencoder (VAE)-based solution designed to reconstruct facial features obscured by occlusions. Experimental results show our VAE model optimized with the structural similarity index measure (SSIM) cost function achieves superior performance, with recognition rates of 91.2% for eye occlusions and 89.7% for mouth occlusions. The SSIM-optimized VAE effectively reconstructs occlude facial features while preserving structural details, demonstrating significant improvements over conventional approaches. This VAE-based solution proves particularly robust for real-world scenarios involving common facial obstructions like masks or sunglasses, making it valuable for applications in healthcare monitoring, driver safety systems, and human-computer interaction
Replay attacks and sniffing in Bluetooth low energy communications with mobile phone
Full text linkThis article analyzes vulnerabilities in Bluetooth low energy (BLE) connections in smartphones against replay and tracking attacks using software defined radio (SDR), evaluating four scenarios with BLE headsets and smartphones from different manufacturers through HackRF one, GNU radio, and Wireshark. In scenario 1, the advertising message ADV_NONCONN_IND was captured and retransmitted, generating persistent and deceptive pairing pop ups on smartphones. In scenario 2, fake pairing request signals were replicated to simulate a connection attempt, causing interface errors and deceptive notifications for the user. In scenario 3, complete pairing sequences were captured and replayed, producing false connection alerts and fabricated information such as battery level indicators from non existent devices. In scenario 4, passive tracking enabled the extraction of sensitive data during the pairing process, including ADV_IND packets, media access control (MAC) addresses, frequencies, manufacturer identifiers, and transmission power levels. A total of 93 successful and 123 failed attacks were recorded, with abnormal behaviors observed such as false pairing requests and manipulated device data, exposing users to risks of identity spoofing, denial of service (DoS) attacks, or targeted interference. The results highlight BLE protocol weaknesses against radio frequency (RF) based attacks and demonstrate the potential of SDR tools as powerful instruments for wireless protocol validation and cybersecurity research
Enhancing cloud resource management: leveraging adversarial reinforcement learning for resilient optimization
Full text linkThis paper introduces the first adversarial reinforcement learning (ARL) framework for resilient cloud resource optimization under dynamic and adversarial conditions. While traditional reinforcement learning (RL) methods improve adaptability, they fail when faced with sudden workload surges, security threats, or system failures. To address this, we propose an ARL-based approach that trains RL agents using simulated adversarial perturbations, such as workload spikes and resource drops, enabling them to develop robust allocation policies. The framework is evaluated using synthetic and real-world Google Cluster traces within an OpenAI Gym-based simulator. Results show that the ARL model achieves 82% resource utilization and a 180 ms response time under adversarial scenarios, outperforming static policies and conventional RL by up to 12% in terms of cost-effectiveness. Statistical validation (p0.05) confirms significant improvements in resilience. This work demonstrates the potential of ARL for self-healing cloud schedulers in production environments
The A3C-CCTSO-R2N2 algorithmic framework for precise edge-cloud parameter estimation
Full text linkEfficient resource allocation is crucial in fog computing environments due to dynamic conditions and different user requirements; this work addresses the scheduling issues of internet of things (IoT) applications in such situations. Our proposed method, chaotic crossover tuna swarm optimizer (CCTSO), is based on metaheuristics and aims to reduce energy usage, reaction time, and SLA breaches; it should help with these problems. Improved system responsiveness and dependability are outcomes of the suggested approach's use of machine learning models for scheduling decision prediction and dynamic workload adaptation. The framework achieves a good balance between performance and energy efficiency by adjusting critical parameters and application settings. By reducing energy usage, reaction time, and operational cost while retaining reduced service level agreement (SLA) violation rates, our solution greatly outperforms previous techniques, according to experimental assessments. In real-world implementations, our results demonstrate that CCTSO is a strong solution for fog-based IoT scheduling, providing greater scalability and adaptability. Taken together, the results of this study provide a strong algorithmic foundation for better resource management in cloud, fog, and edge computing environments
Optimized weighted non-local mean filter for enhanced denoising and improved quality of medical images
Full text linkImage quality is significantly influenced by noise, light, and artifacts, particularly in medical images where precision is essential for accurate diagnosis. Denoising is a significant pre-processing for enhancing the overall quality of images to enable efficient classification, feature extraction, and segmentation. Conventional denoising filters smooth out boundaries and lose texture because they are ineffective to process color images. To address these limitations, a weighted factor-based non-local means (WF+NLM) filter is proposed as an improvement over the non-local means (NLM) filter, with an additional weight factor based on pixel similarity. This addition reduces blurring while maintaining fine details, resulting in improved quality. The proposed filter performs effectively in blood smear images, with a peak signal-to-noise ratio (PSNR) of 39.6904, SSIM of 0.9551, and gradient SSIM of 0.9889. Statistical tests indicates that the WF+NLM filter improves image quality in terms of structure, gradients, and feature similarity. Statistical inference for a one-tailed paired t-test validates statistical significance with the highest t value of 9.323829 with p-value 0.00037 by the wavelet-based non-local moment mean (W-NMM) filter asserts higher image restoration quality
Modeling broken rotor bar faults in induction motors: a combined SSFR and FEM approach
Full text linkThe fault of broken rotor bar (BRB) yields to high levels of stress in induction motor drive (IM) and being a common fault. This paper proposes a novel hybrid approach combining standstill frequency response (SSFR) testing and finite element method (FEM) modeling to improve fault diagnosis accuracy. The findings were verified experimentally using a 7.5 kW three-phase IM by SSFR approach under various failure scenarios. Reliability of SSFR method is confirmed by the use of FEM, flux 2D magnetic analysis software is employed on the same IM using in SSFR to determine the magnetic field under different fault and load conditions. The work is finished by current harmonics analyses and the outcomes of the BRB model demonstrate that the combined method enhances fault detectability, particularly for incipient and partial bar breakages, reducing false alarms compared to conventional techniques
A review on radio-frequency transceiver architectures for low-power wireless sensor networks
Full text linkWireless sensor networks (WSNs) have garnered significant scientific attention because of their many uses, but their power usage is a fundamental barrier to their deployment. Energy constraints have a direct effect on important design elements including battery capacity, energy harvester effectiveness, and network longevity. To enable sustainable WSN operation, radio-frequency (RF)-based transceiver (TR) design has become a key area of study. A thorough examination of current RF-TR architectures is given in this paper, with a focus on low-power (LP) implementations designed for WSN applications. Amplifier-sequenced hybrid (ASH), superheterodyne (SHD), zero-intermediate frequency (Zero-IF), low-intermediate frequency (Low-IF), sliding-intermediate frequency (Sliding-IF), and super-regenerative (SRG) architectures are among the TR system designs that are categorized, with an emphasis on the performance trade-offs associated with each. Comparative evaluation shows that Zero-IF and SRG architectures are more energy efficient than other designs that were studied, which makes them viable options for ultra-low-power (ULP) WSN installations. Along with outlining important research issues in RF-TR design, such as hardware minimization, security, synchronization, and energy optimization, this review also suggests possible future paths to improve the sustainability and performance of WSN-based RF-TRs
Memory management principle for dynamic isolation in agent-based epidemic modeling
Full text linkThis paper presents a new epidemiological modeling approach that adapts the working set (WS) concept from computer memory management to the dynamics of infectious diseases. Traditional compartmental models provide valuable insights but are limited in their ability to capture dynamic isolation and heterogeneous contact patterns. In contrast, the WS model conceptualizes a time-varying subset of agents actively participating in social interactions, allowing for dynamic adjustments to the rate of infection and the explicit identification of superspreaders. By incorporating isolation states for both susceptible and infected individuals, the model more realistically captures quarantine and targeted interventions. Including an incubation period reduces epidemic peaks by nearly 40% and delays them by more than three weeks, providing critical time for public health response. Within the WS model, moderate isolation reduces peak infection rates by more than three times compared to uncontrolled scenarios, while high isolation almost completely prevents large-scale spread. These results highlight the model's ability to estimate the intensity and timing of interventions with greater accuracy than traditional models. By integrating the time window parameter and computer resource management principles, the adapted WS model represents a robust and adaptable tool for analyzing epidemic dynamics. The results highlight its potential for advancing epidemic modeling and supporting real-time public health decision-making
Design of a secure-cloud remote medical monitoring system using the P-QRS-T electrocardiogram detection algorithm
Full text linkThe COVID-19 pandemic has highlighted the limitations of traditional healthcare, resulting in higher mortality rates among children, the elderly, and healthcare workers. This situation has created a pressing need for urgent medical care from healthcare professionals. This paper presents a secure cloud-based remote medical monitoring system that integrates the internet of things (RMMS-IoT) with advanced P-QRS-T electrocardiogram (ECG) detection algorithms to enable real-time, accurate vital sign analysis. The system combines microcontroller devices, wearable sensors, and medical-grade equipment, leveraging hypertext transfer protocol secure (HTTPS) and Blynk bridge cloud technologies to ensure data security and interoperability. The RMMS-IoT system demonstrated high accuracy in monitoring vital signs by comparing its results with data from actual measuring devices, showing errors in body temperature readings below 1% and heart rate (HR) measurements below 2.8%. The algorithm used to detect P-QRS-T features from the ECG exhibited robust performance in differentiating between normal and abnormal ECG patterns in patients, and it achieved an accuracy rate of 90% in ECG classification
Isolation enhancement of four port multiple-input multiple-output antenna for sub-6 GHz 5G communication
Full text linkFor 5G communication, this research suggests a small, broad band, 4-port multiple-input multiple-output (MIMO) antenna with an impedance bandwidth of 2.0 GHz (or 3.0-5.0 GHz). The n77, n78, and n79 bands are covered. The single antenna is realized by inserting the stubs and creating the ‘HI’ slot on the rectangle patch with the defect in the ground plane, using FR-4 substrate. Next, four MIMO antennas are built utilizing the reference antenna. Due to mutual interaction, implementing MIMO systems presents a substantial challenge: achieving good isolation between antenna parts in the confined space. To increase isolation with decoupling procedures, the four antennas are placed orthogonally to one another. Because the antennas are positioned orthogonally, the MIMO antenna has an isolation of 28.0 dB. The diversity gains (DG) and envelop correlation coefficient (ECC) are used to analyze the recommended antenna's diversity performance characteristics, and the results show that the values are 9.99 dB and 0.0003, respectively. The simulated S-parameters have been compared with orthogonal and adjacent positions of quad port MIMO antenna. Anritsu MS2037C VNA is used to measure the parameters, and HFSS software is used to simulate it