Dissertations/Theses - Department of Electrical and Electronic Engineering
Browse
800 results
Search Results
Item Polarization insensitive electrically reconfigurable metasurface for metalensing at near infrared wavelength(Department of Electrical and Electronic Engineering (EEE), BUET, 2025-05-19) Asif Hossain Bhuiyan, Md.; Choudhury, Dr. Sajid MuhaiminThe conventional fiber communication band at 1.55 µm is approaching its capacity limit due to the growing demand for high-speed and large-volume data transmission. In re- sponse, the 2 µm waveband is emerging as a promising candidate for next-generation communication systems, offering higher capacity and lower transmission loss. In align- ment with this technological shift, we present an optically engineered metasurface tai- lored for this waveband, suitable for applications such as fiber coupling and lensing. The proposed device exhibits polarization insensitivity and dynamic tunability between its transmissive (ON) and reflective (OFF) states. To enable this tunability, we incorpo- rate a novel phase change material, In3SbTe2 (IST), which features fast, reversible, and non-volatile transitions between its metallic and insulating phases. Electrical control is achieved by integrating indium tin oxide (ITO) as a micro-heating element, which modulates the optical properties of IST, thereby enhancing the device’s applicability in point-of-care systems. Utilizing the Finite-Difference-Time-Domain (FDTD) method, we demonstrate a modulation depth of 90%, a high focusing efficiency of 76%, and an ON-OFF switching ratio of 26 dB. The use of multiple thin IST layers ensures uni- form and energy-efficient switching, with a low energy requirement of 232.98 nJ/µm2. With its outstanding performance at the 2 µm communication band and dynamic mul- tifunctional capabilities, the proposed metasurface stands poised to transform future telecommunication technologies and broader optical systems.Item Design and implementation of microwave plasma reactor for the production of silicon-nano-particle from locally available quartzite(Department of Electrical and Electronic Engineering (EEE), BUET, 2025-05-28) Das, Sagar Kumar; Chowdhury, Dr. NadimPlasma nanotechnology plays a crucial role in the large-scale synthesis of nanoparticles, which are widely used in various modern technological applications. Conventional microwave-based plasma reactors typically incorporate components such as circula- tor and directional couplers to protect the magnetron from reflected microwave power. However, these components significantly increase the overall system cost and complex- ity. In this study, a cost-effective alternative design for a microwave plasma reactor was proposed and analyzed using finite element simulations. The design replaces the conventional circulator with a three-decibel (3dB) waveguide bridge, which passively redirects reflected microwave energy away from the magnetron, thereby mitigating the risk of damage. During the design process, particular emphasis was given on mini- mizing microwave reflections on the magnetron side. This was achieved by optimiz- ing for a low reflection coefficient and a low Voltage Standing Wave Ratio (VSWR), ensuring minimal power loss and enhanced operational stability. An operational mi- crowave plasm reactor was successfully fabricated using locally sourced and affordable components. The functionality of the system was demonstrated by synthesizing fumed silica nanoparticles, which were subsequently characterized using Scanning Electron Microscopy (SEM) and Energy Dispersive X-ray Spectroscopy (EDS). However, sili- con nanoparticles production from quartz is also possible using the plasma reactor. But due to some complexity this work has been suggested as future prospect for further study and analysis. This alternative reactor design provides a practical and economical solution for microwave plasma generation, eliminating the need for expensive compo- nents while maintaining effective performance. It holds promising potential for a wide range of applications, including nanoparticle synthesis, chemical processing, biomass conversion, and waste treatment.Item Beryllium sulfide monolayer as a biosensor for early lung disease detection(Department of Electrical and Electronic Engineering (EEE), BUET, 2025-04-15) Saha, Sudipta; Kawsar Alam, Dr. Md.Considerable attention has been directed towards the prognosis of lung diseases primarily due to their high prevalence. Despite advancements in detection technologies, current methods such as computed tomography, chest radiographs, bold proteomic patterns, nuclear magnetic resonance, and positron emission tomography still face limitations in detecting diseases related to the lungs. Consequently, there is a need for swift, non-invasive and economically feasible detection methods. Our study explores the interaction between BeS monolayer and breathe biomarkers related to lung disease utilizing the density functional theory (DFT) method. Through comprehensive DFT analysis, including electronic properties analysis, charge transfer evaluations, work function, optical properties assessment and recovery times, the feasibility and efficiency of BeS as a VOC (volatile organic compound) detection are investigated. Findings reveal significant changes in bandgap upon VOC adsorption, with notable alteration in work function for selective compounds. Optical property analyses demonstrate the potential for selective detection of biomarkers within specific wavelength ranges. Moreover, the study evaluates the impact of electric fields and strain on VOC-2D BeS interaction. Furthermore, the desorption of these VOCs from the BeS surface can be achieved through a heating process or under the illumination of UV light. This feature enables the reusability of the 2D material for biosensing applications. These findings highlight the potential of the BeS monolayer as a promising material for the sensitive and selective detection of breath biomarkers related to lung disease.Item Multi-speaker end-to-end text to speech synthesis for low resource languages(Department of Electrical and Electronic Engineering (EEE), BUET, 2025-04-15) Shahruk Hossain; Ariful Haque, Dr. MohammadThis thesis presents a high-quality, end-to-end multi-speaker text-to-speech (TTS) system for Bangla - a language spoken by millions yet lacking in open-source, high- quality speech resources. TTS systems have broad applications, including virtual assistants, audiobooks, dubbing, and accessibility tools. Despite Bangla’s large speaker base, its representation in modern open-source speech synthesis remains limited. Motivated by this gap, and the lack of accessible tooling for building contemporary TTS systems in Bangla, this work presents a curated speech dataset, tentatively named Bani, compiled from publicly available corpora and community- driven projects. To address this, a curated dataset - tentatively named Bani - was compiled from public and community-driven sources. A key contribution is a remastering pipeline using deep learning-based denoising and enhancement, substantially improving audio quality for TTS training. Bani served as the foundation for training single-speaker and multi-speaker TTS models. The architecture is based on the Variational Inference with Adversarial Learning for end-to-end TTS (VITS) model, with two core modifications intro- duced in this work: explicit duration modeling and integration of a pretrained speaker embedding model jointly trained with the system. These changes aimed to improve convergence, speaker similarity, and synthesized speech naturalness. Evaluation combined objective metrics - Mel-Cepstral Distortion (MCD), tran- scription error rates, and speaker similarity - with subjective Mean Opinion Score (MOS) tests from native Bangla speakers (1=poor, 5=excellent). The modi- fied multi-speaker model achieved a MOS of 3.64 ± 0.48, surpassing the baseline (3.46 ± 0.50) and single-speaker (3.10 ± 0.61) models. Objective scores showed a 10% drop in MCD and 9.5% boost in speaker similarity. A commercial Google Bangla TTS system scored 4.12 ± 0.33. These results show that both audio re- mastering and architectural changes significantly enhance perceived and measured synthesis quality, while explicit duration modeling improves training efficiency without sacrificing fidelity.Item Development of a smart power efficient patch antenna for wireless communication(Department of Electrical and Electronic Engineering (EEE), BUET, 2025-01-25) Ziaul Islam, Md.; Saha, Dr. Pran KanaiSmart antenna technologies have revolutionized communication systems, particularly in the context of the IoT (Internet of Things) applications. This research introduces an innovative approach to smart antenna technology by proposing a modified polygonal layout which utilizes dynamic beam direction adjustment based on user availability and the Direction of Arrival (DOA). CST Microwave Studio is used for the initial antenna design and the most efficient version is fabricated for experimental validation. Parasitic elements, slots and the defected ground structure technique are used to create a fully directional antenna capable of forming a beam that covers an acute angle. By appropriately arranging multiple identical patch antenna elements, a configuration is established that allows signals to be transmitted in all directions, ensuring full 360 degree coverage for desired users. Several polygonal antenna models are developed and tested experimentally to identify the most optimized configuration in the ISM band (Industrial, Scientific and Medical band) operating in 2.4 to 2.5 GHz range. An algorithm is also developed to steer beam through a control circuit consisting of radio transceiver modules and microcontrollers. This algorithm combines adaptive beamforming, feedback based beam steering and automatic retransmission control (ARQ) to optimize signal transmission to targeted users. User location is tracked by measuring parameter like the received signal strength indicator (RSSI) which allows the scheme to estimate the user's position relative to the antenna array. The proposed smart antenna array design is then validated experimentally to demonstrate significant improvements in both performance and reliability. The antenna performance is evaluated based on radiation efficiency, power efficiency, signal quality, constant signal strength, Quality of Service (QoS) and its ability to increase communication capacity while eliminating dead zones. The optimized result of the proposed antenna design demonstrates a gain of 6.38 dBi, a return loss of -26.2 dB, VSWR of 1.01 with beamwidth of 49.3°. These results indicate a reshaped and strongly directional radiation pattern. In the proposed polygonal antenna array arrangement, the dead zone is minimized to only 3.33%, ensuring near-complete coverage and minimizing interference with radiation efficiency of 89.3% and electrical power efficiency of 74.6%. The proposed scheme offers significant improvements over traditional methods by enabling adaptive beamforming within a single sector and facilitating interference free sector-to-sector transitions, while efficiently radiating power toward the intended user without signal interruption. This thesis highlights the potentiality of the proposed smart antenna system to meet the growing demand for efficient, reliable and uninterruptible communication solutions particularly in the context next generation smart antenna technology.Item Design of silicon carbide based single quantum well white LED(Department of Electrical and Electronic Engineering (EEE), BUET, 2025-01-12) Mahfuzul Haque, Md.; Choudhury, Dr. Sajid MuhaiminThe unique optical and electrical properties of two-dimensional silicon carbide (2D-SiC) have lately attracted considerable attention as a potential element for optoelectronic systems. The initial goal of this work was to investigate the impact of different materials doped into SiC nanosheets on their electrical and optical properties, along with the impacts of strain, vacancy defects, and antisite defects in SiC. The direct energy bandgap of 2D SiC is calculated to be 2.57 eV. After a satisfactory examination of the electrical and optical properties, I suggested a 2D-SiC-based LED as the active layer. This LED depends on a single quantum well and is subsequently microscale-simulated utilizing the SILVACO TCAD program. The LED structure I created can generate visible light, namely in the blue, green, and red wavelengths, which together provide white light. The suggested LED device has been analyzed for power spectrum density, current-voltage characteristics, luminous power, CIE color coordinates, and light extraction efficiency. The light emission characteristics of this product offer several attractive attributes, including higher efficiency, heightened electrical conductivity, and higher electrical power.Item Design and analysis of a low-noise on-chip three-dimensional plasmonic biological cell imaging system(Department of Electrical and Electronic Engineering (EEE), BUET, 2025-01-07) Maliha Momtaj; Anisuzzaman Talukder, Dr. MuhammadEnhancing accessibility to affordable and efficient cell imaging tools has been a long- standing worldwide objective for rapid disease detection and diagnosis. We present an innovative approach to this goal, introducing a compact, portable, and user-friendly three-dimensional (3D) cell imaging platform leveraging silicon photonics and the sur- face plasmon coupled emission (SPCE) phenomenon. Central to our method is a spe- cially designed slide that incorporates a fundamental SPCE structure seamlessly inte- grated with a silicon nitride (SiN) waveguide. We introduce a grooved array on the slide to couple light into the waveguide, interfacing with a broadband light source. This source is employed to excite fluorescently labeled cells. The excitation is achieved using edge coupling through the SiN waveguide, directing the excitation light to the specimen placed on the SPCE platform. This integrated architecture eliminates the ne- cessity for an additional filter to extract the required light for fluorophore excitation while enabling precise excitation of labeled cells. Following the fluorophore excitation, the emitted SPCE signals are captured and subjected to analysis. We have developed an imaging algorithm based on the emitted light patterns, which we comprehensively detail through theoretical demonstration. This algorithm has the remarkable capability of achieving 3D cell imaging. This fusion of optics and computational techniques can significantly impact the domain of cell imaging by enabling the development of easily accessible and portable point-of-care (POC) imaging tool.Item Privacy-preserving human activity recognition using principal component-based wavelet CNN(Department of Electrical and Electronic Engineering (EEE), BUET, 2024-08-27) Pervin, Nadira; Imtiaz, Dr. HafizHuman activity recognition (HAR) is crucial in applications such as smart homes, in- teractive games, surveillance, security, and healthcare. In recent years, Channel State Information (CSI) data extracted from Wi-Fi signals has garnered significant interest for applications in HAR. This interest stems from CSI’s several advantages, including its immunity to illumination variations and environmental disturbances, and the elim- ination of the need for wearable devices. Despite being widely used, existing HAR system’s performance suffers when used in new environments without system improve- ment or retraining. This constraint can be overcome by gathering and annotating data from various locations, and then retraining the system. However, it is far from ideal from the privacy perspective, as the training algorithms access the data from differ- ent privacy-sensitive environments. This motivates us to design a reliable and robust privacy-preserving HAR system. In this work, we introduce a Differentially Private Principal Component-based Wavelet Convolutional Neural Network (DP-PCWCNN) that offers accurate and robust HAR performance across different environments, while preserving strict privacy constraints. We evaluate the performance of our proposed algorithm on two publicly available real dataset and demonstrate that our proposed sys- tem closely approximates the non-private system’s performance for some parameter choices.Item Human activity recognition on edge devices using salient region-guided knowledge-distilled lightweight model(Department of Electrical and Electronic Engineering (EEE), BUET, 2024-11-18) Abrar Zahin, Md.; Haque, Dr. Mohammad ArifulUnderstanding human behavior is crucial in disaster management, surveillance and el-der care. In disaster management, drone surveillance can quickly identify activities like walking, running, lying down, standing, and sitting. This helps prioritize rescue opera-tions and save lives. Accurate activity recognition also boosts security and effectively monitors public spaces. Additionally, in elder care, it assists in keeping an eye on the well-being of elderly individuals by promptly detecting falls or unusual behavior, which allows for timely intervention. Overall, these applications highlight the importance of real-time activity recognition for safety and support. Therefore, it’s important to de-velop a lightweight and accurate model that can run on edge devices like Raspberry Pi used in drones and security cameras. Faster models often sacrifice accuracy compared to larger, complex Deep Neural Networks (DNN) which are highly accurate, require a lot of computational power, making it difficult to implement Human Activity Recog-nition (HAR) systems on devices with limited processing capabilities. In this thesis, we have made two major contributions. First, we introduced a new lightweight DNN model called YOLOv5n-light, which is 4 times smaller and 1.5 times less computa-tionally intensive than YOLOv5n, the smallest model in the YOLOv5 family. Second, we proposed a novel training strategy called Pre-trained Weight Pruned Retrained by Salient Feature Guided Knowledge-Distillation (PWPR-SFGKD) to enhance the accu-racy of lightweight YOLO models. In our approach, we first train the model, followed by pruning its weights, and then perform a retraining process using a Salient Feature Guided Knowledge Distillation (SFGKD) technique. For loss calculation, we compute the difference between the feature maps of the full image and the background region, derived from both the teacher and student models. This allows us to capture more meaningful, context-specific information during the distillation process. When we ap-plied this method to the YOLOv5n model, we observed a significant improvement in performance, with the mAP-50 score increasing from 0.64 to 0.71. This result high-lights the effectiveness of our proposed PWPR-SFGKD approach in enhancing model accuracy, outperforming traditional YOLOv5 training strategies.