Thesis in ETE

Browse

Filters

Settings

Search Results

Now showing 1 - 6 of 6
  • Thumbnail Image
    Item
    SINGLE-FEED DUAL BEAM SWITCHABLE ARRAY ANTENNA FOR ISM BAND APPLICATION
    (CUET, 10-Oct-2023) Akther, Naznin
    This manuscript proposes one single-feed dual-beam switchable array antenna and one single-feed multi-beam switchable array antenna. The proposed antenna contains three and four microstrip square patch antenna elements and a switchable feed network respectively. In dual-beam design, the feed network creates a 900 phase difference between the center element and the right or left antenna elements. This antenna can tilt its main beam direction to θ = by exciting either right or left antenna elements along with the center patch through two PIN diodes, respectively. The dual-beam switching concept is experimentally verified with better than 8 dBi gain for every condition at ɸ = 00-plane. The proposed multi-beam array antenna has two 900 switchable phase shifters. The 900 switchable phase shifter consists of a single-pole double-throw (SPDT) switch and a hybrid coupler. The hybrid coupler generates a 900 phase difference between the antenna elements to tilt its radiation pattern. The two SPDT switches can generate three modes to control the input ports of the hybrid coupler. As a result, two tilted beams and one difference pattern are generated from the same structure. A prototype multi-beam antenna is fabricated to demonstrate the concept experimentally. The measured results confirm the concept with a good agreement between the simulation and measurement. Better than 20dB cross-polarization suppression is achieved in the measurement.
  • Thumbnail Image
    Item
    OPCNet: An optimized parallel convolutional neural network for classification of satellite imagery.
    (CUET, 21-May-2024) Tumpa, Priyanti Paul
    Satellite image classification is crucial for various applications, driving advancements in Convolution Neural Networks (CNNs). While CNNs have proven effective, deep models often encounter overfitting issues as the network's depth increases since the model has to learn many parameters. Besides this, traditional CNNs have the inherent difficulty in extracting fine-grained details and broader patterns simultaneously. To overcome these challenges, this research presents a novel approach using an optimized parallel CNN (OPCNet) architecture with an SVM classifier to classify satellite images. Each branch within the parallel network is designed for specific resolution characteristics, spanning from low (emphasizing broader patterns) to high (capturing fine-grained details), enabling the simultaneous extraction of a comprehensive set of features without increasing network depth. The OPCNet incorporates a dilation factor to expand the network's receptive field without increasing parameters, and a dropout layer is introduced to mitigate overfitting. Evaluation of two public datasets (EuroSAT dataset and RSI-CB256 dataset) demonstrates remarkable accuracy rates of 97.91% and 99.8%, surpassing previous state-of-the-art models. Finally, OPCNet, with less than 1 million parameters, outperforms high-parameter models by effectively addressing overfitting issues, showcasing exceptional performance in satellite image classification.
  • Thumbnail Image
    Item
    MRI Brain tumor detection and classification using parallel deep convolutional neural network.
    (CUET, 21-May-2024) Rahman, Takowa
    Brain tumors are frequently classified with high accuracy using convolutional neural networks (CNNs) and better comprehend the spatial connections among pixels in complex pictures. Due to their tiny receptive fields, the majority of deep convolutional neural network (DCNN)-based techniques overfit and are unable to extract global context information from more significant regions. While dilated convolution retains data resolution at the output layer and increases the receptive field without adding computation, stacking several dilated convolutions has the drawback of producing a grid effect. To handle gridding artifacts and extract both coarse and fine features from the images, this research suggests using a dilated parallel deep convolutional neural network (PDCNN) architecture that preserves a wide receptive field. To reduce complexity, initially, input images are resized and then grayscale transformed. Data augmentation has since been used to expand the number of datasets. Dilated PDCNN makes use of the lower computational overhead and contributes to the reduction of gridding artifacts. By contrasting various dilation rates, the global path uses a low dilation rate (2, 1, 1), while the local path uses a high dilation rate (4, 2, 1) for decrement even numbers to tackle gridding artifacts and extract both coarse and fine features from the two parallel paths. Using three different types of MRI datasets, the suggested dilated PDCNN with the average ensemble method performs better. The accuracy provided by the Multiclass Kaggle dataset-III, Figshare dataset-II, and Binary tumor identification dataset-I is 98.35%, 98.13%, and 98.67%, respectively. In comparison to state-of-the-art techniques, the suggested structure imp
  • Thumbnail Image
    Item
    An Integration of Planar Cavity-Backed Antenna with Artificial Magnetic Conductor in the Application of 2.4 GHz WBAN Application
    (CUET, 2-Nov-2023) Murshed, Abu Hena
    A high-gain array antenna is proposed using cavity-backed technology with the help of Substrate Integrated Waveguide Technology. The joint combination of both patch and the cavity-backed antenna is also known as a hybrid combination. Such types of antennas allow bandwidth enhancement and further division of patches to allow high gain and low cross-polarization behaviour. The whole patch antenna is further divided and based on the number of array elements in the patch, four array antenna cases have been proposed. A comprehensive investigation of them is presented. The proposed antenna operates at Ku-band, showing linearly polarized wave in TE101 mode and this hybrid combination provides several advantages. First, it has the advantages of a cavity-backed antenna and the conventional slot antenna. Second, an antenna is more compact. Third, it contains a much simpler matching network. In addition, being a member of the planner waveguide family, the proposed antenna also possesses a low profile and allows easy integration with the planner circuit. An exact representation of parameters of all antennas that have emerged during the designing course helps to grasp the sense of designing such kinds of hybrid array antenna of planner waveguide technology. Later, the proposed antenna is verified experimentally and exhibits a satisfying agreement with the data found during numerical simulation. The antennae’s simulated parameters were investigated with the help of CST (computer simulation technology) and HFSS (high frequency structure simulator) simulation software in microwave momentum mode.
  • Thumbnail Image
    Item
    APPLICATION OF DEEP LEARNING APPROACH IN TRANSCRANIAL MAGNETIC STIMULATION FOR PREDICTING ELECTRIC FIELD
    (CUET, 7-May-2023) Sathi, Khaleda Akhter
    As a non-invasive neuromodulation technique, transcranial magnetic stimulation (TMS) has already exhibited a great impact in clinical applications and scientific researches. For finding new clinical applications of TMS, the current study focused on a deep learning-based prediction model as an alterative of time-consuming electromagnetic (EM) simulation software. However, the main bottleneck of the existing prediction models is to consider fewer input parameters such as single coil type and coil position for predicting electric field value. To address these limitations, this research develops an improved approach based on a deep neural network (DNN) to predict electric field by considering several input parameters such as coil turns of single wing, coil thickness, coil diameter, distance between two wings, distance between head and coil position, and angle between two wings of coil. In addition, considering the fact of focality and depth tradeoff, the assembly coil is designed. The performance of the model is evaluated based on four verification statistic metrics including coefficient of determination (R2), mean squared error (MSE), mean absolute error (MAE), and root mean squared error (RMSE) between the simulated and predicted values. Compared to current state-of-the-art methods, the proposed DNN model outperformed with the value of R2=0.9992, MSE=0.0005, MAE=0.0188, and RMSE=0.0228 in the testing stage. Therefore, the proposed DNN model can accurately predict electric field from assembly coil in a lower period of time without using traditional simulation software.
  • Thumbnail Image
    Item
    Analysis on simple feed microstrip patch array antenna design for circular polarization diversity and beam switching with DOA estimation.
    (CUET, 21-May-2024) Das, Debprosad
    The terminal node antenna of a wireless communication device should possess