ABSTRACT This paper focuses on the impact of Deep Neural Networks (DNN), especially those utilizing back propagation models, in the design aspects of a fractal geometry‐based microstrip patch antenna under high—frequency applications chosen at 25–35 GHz millimeter (mm) wave frequency band. Designing a fifth generation (5G) mm wave frequency microstrip patch is a challenging task due to size constraints and performance needs, making traditional methods complex. Antenna design for precise applications under high frequencies with better performance, DNNs are chosen. The ability to learn from the complex datasets of antenna designs enables DNNs to effectively predict and optimize the dimensions of the fractal geometry‐based microstrip patch antenna to meet the performance characteristics like bandwidth, gain, and voltage standing wave ratio. DNNs speed up the design process, minimize computing burden, and allow engineers to experiment with unique, compact designs that would be impossible to produce using traditional approaches. The proposed ANN integrated fractal geometry‐based 2 × 2 microstrip patch antenna array achieved a bandwidth of 5200 MHz and a gain of 9 dB, which are very much suitable for applications like vehicular communications under 5G models and biomedical communications.
ABSTRACT This paper presents a low‐profile quad‐patch bowtie dipole ultrawideband (UWB) multiple‐input–multiple‐output (MIMO) specifically designed for internet of things (IoT)‐based smart home applications. Operating over a wide frequency range of 2–10 GHz, This antenna configuration achieves a high gain of 8.5 dBi and exhibits excellent inter‐element isolation exceeding dB—without relying on additional decoupling structures. Unlike conventional 2D measurement approaches, this work employs comprehensive 3D radiation pattern analysis to evaluate real‐world signal behavior in three‐dimensional indoor environments. The antenna's low profile, cost‐effective fabrication, and inherent isolation realize highly ideal for deployment in portable and embedded IoT devices. Furthermore, key evaluation parameters, such as the envelope correlation coefficient (ECC ), diversity gain (10 dB), and channel capacity Loss ( bps/Hz)—validate its suitability for high‐speed, reliable smart home communications. The proposed design allows for scalable, efficient solution for next‐generation UWB‐MIMO systems targeting intelligent indoor wireless networks.
ABSTRACT Autism Spectrum Disorder (ASD) is characterized by significant behavioral and social impairments, including irregular physical activities that may pose risks of self‐harm. Effective real‐time monitoring of these behaviors is crucial for enabling timely interventions and enhancing patient care. This study presents an innovative IoT‐inspired edge computing framework that employs a deep learning‐based Bidirectional Long Short‐Term Memory (Bi‐LSTM) network to detect abnormal physical activities in individuals with ASD. The framework facilitates the processing of wearable sensor data on GPU‐enabled edge devices, ensuring low‐latency recognition of movement patterns. Evaluation of the proposed system was conducted using the iFogSim simulator, yielding a remarkable accuracy of 95.83% in identifying irregular activities while significantly reducing computational overhead compared to conventional cloud‐based methods. The findings underscore the potential of edge‐based analytics to deliver efficient real‐time monitoring, thereby assisting caregivers in managing ASD‐related movement irregularities. Future research will aim to incorporate multimodal sensing techniques to further enhance behavioral analysis.
ABSTRACT The industrial Internet of things (IIoT) integrates IoT technologies into industrial environments, enhancing automation, monitoring, and manufacturing efficiency. However, interconnected devices and sensors introduce significant security and privacy challenges, making intrusion detection a critical requirement. This study proposes a comprehensive progressive graph convolutional networks approach for enhancing the security of IIoT through advanced intrusion detection systems for threat identification and mitigation (PGCN‐IoT‐AIDS‐TIM). Data were collected from the ToN_IoT dataset and preprocessed using a regularized bias‐aware ensemble Kalman filter (RBAEKF) to clean, normalize, and balance the data. Relevant features were selected using the multiplayer battle game‐inspired optimizer (MBGIO). The selected features were then classified using progressive graph convolutional networks (PGCN) to identify normal and attack traffic. To improve detection accuracy, the Bitterling Fish Optimization Algorithm (BFOA) was employed to optimize PGCN parameters. The proposed framework was implemented in Python and evaluated using accuracy, precision, recall, F ‐measure, specificity, ROC, error rate, and computational time. Results demonstrated superior performance compared with CNN‐ID‐IoT, DNN‐AI‐ID‐IoT, and LSTM‐IDA‐IoT methods, achieving substantial improvements in accuracy, precision, and recall.
ABSTRACT Rapid growth in the IoV domain has led to significant security risks, making it imperative to use state‐of‐the‐art intrusion detection systems. Conventional methods usually lack the ability to effectively learn spatial, temporal, and global properties from the vehicular traffic data streams. Therefore, this study introduces a novel IDS named FusionIDS. It is different from previous convolutional neural network long short‐term memory (CNN‐LSTM) approaches in that this study considers the integration of Transformer‐based global attention along with SHAP explanations that are specific to IoV IDS. FusionIDS consists of three components, namely, a 1‐dimensional convolutional neural network (1D‐CNN) for extracting local features, an LSTM network for learning sequential information, and a Transformer encoder for global information extraction. In addition, it uses SMOTE to manage the problem of severe class imbalance and SHapley Additive exPlanations (SHAP) for explainability. Results show that FusionIDS obtains around 98% accuracy when tested on various attack cases in IoV environments.
