Loading...
![Thumbnail Image](assets/images/Collection.png)
Theses
Browse
53 results
Search Results
ThesisItem Open Access SPEECH DEREVERBERATION BY FEATURE MAPPING USING SKIP CONVOLUTIONAL NEURAL NETWORK(G. B. Pant University of Agriculture & Technology, Pantnagar-263145, 2023-02) Ganhwar, Utkarsh; Mathur, SanjayReverberation is the phenomenon that degrades the quality of a speech signal when reflected off objects in space causing multiple reflections to build up. Reverberation disrupts information contained in an audio signal recorded by remote microphones and in hands free telephonic plot. This corruption in signal caused by reverberation makes speech recognition system ineffective while recognizing the voices in automatic voice recognition applications. Dereverberation is the method to reduce the reverberation in an audio signal and increase the speech quality. There is a growing demand of dereverberation techniques to remove corruption from the audio signals in real time scenarios and to make automatic speech recognition systems more effective. In this dissertation, an approach to the dereverberation of a speech signal is presented by feature mapping using Skip Convolutional Neural Network. The evaluation of the performance of the proposed network with a given data set is also calculated for a better understanding of the results with the proposed network. Pre-extracted clean speech and reverberant speech datasets are used for training/testing speech dereverberation algorithms to train the proposed network. The original reverberant data has only a little degree of reverberation. So, strong reverberation effects were added to the clean speech datasets using a reverberator object (with predefined functions of MATLAB) to generate synthetically reverberant datasets. This synthetically generated reverberant speech data make the system effective against a high degree of reverberation. Afterwards, the network performance is evaluated by calculating the cepstral distance (CD), log likelihood ratio (LLR) etc. In the end, predicted the log-magnitude spectra of the reverberated speech signal using the pretrained network.ThesisItem Open Access Implementation of 2-x function using polynomial approximation(G. B. Pant University of Agriculture and Technology, Pantnagar, 2022-10) Khan, Sameena; Tomar, AbhishekThe primary building block of arithmetic circuits is the implementation of elementary functions. Commonly utilised building blocks in sophisticated algorithms include elementary functions like logarithms, exponentials, trigonometric functions, square roots, and inverse square roots. With a limited amount of arithmetic operations, it is impossible to compute elementary functions precisely, only an approximation of these functions is obtainable. The aim of the proposed work is to implement one of the elementary functions i.e., ‘2-X’ which is a frequently used function in digital signal processing, machine learning, artificial intelligence and is frequently utilised in a variety of scientific application. So, we proposed to design an architecture of a polynomial approximation based 2-X function using polynomial approximation. This architecture is implemented with hybrid method and curve fitting techniques with linear and quadratic polynomial approximation method. Further, in the proposed architecture number of segments are increased to reduce error characteristics and to improve accuracy. The designed architecture is simulated using Verilog Xilinx ISE Design Suite software with Isim simulator and MATLAB simulation tools. The results show that both the architecture design methods provide the flexibility that one may choose either the linear approximation for shorter latency or the quadratic approximation for relative high accuracy result. The maximum error of proposed architecture is reduced in comparision to the previously described architecture. So, proposed architecture has better accuracy than previous architecture at the cost of small increase in memory.ThesisItem Open Access Design and analysis of metamaterial based patch antenna with low SAR for healthcare applications(G. B. Pant University of Agriculture and Technology, Pantnagar, 2022-10) Rawat, Himani; ParasA new metamaterial-based UWB antenna with a frequency range of 2–12 GHz is described. This antenna employs metamaterial with a low specific absorption rate. Wearable antennas are becoming more and more necessary, exclusively in the field of healthcare. The antenna is positioned in free space for the analysis and then close to the human body. The antenna's parameters are then measured, simulated, and compared. The designed antenna involves the use of an Fr4 substrate with a 1.3 mm thickness and dimension of 30 mm*30 mm. The curves are added along with slotting on the ground and patch. Metamaterial SRR is used to enhance the overall bandwidth of the antenna and reduces the SAR rate. The antenna shows better results for off and on the body when placed on the human body. It is observed that there is a reduction in gain and radiation efficiency when the antenna is placed near the human body. The proposed antenna is simulated using Ansoft HFSS software, fabricated in a PCB designing machine, and tested using VNA. The proposed antenna resonates in UWB frequency bands with a bandwidth of 9 GHz having a gain above 5 dB and efficiency above 90%. The measured return is in good agreement with the simulated one. The performance of the proposed antenna is far better than earlier reported work. It can be determined from the results that the antenna will work efficiently for healthcare applications with low SAR.ThesisItem Open Access Design of a microstrip antenna for early brain stroke detection(G.B. Pant University of Agriculture and Technology, Pantnagar, District Udham Singh Nagar, Uttarakhand. PIN - 263145, 2022-10) Singh, Sanjul; Verma, ReetaStroke is one of the world's most serious health issue. A decrease of blood flow to a portion of the brain that results in brain tissue damage is termed as stroke in medical science. Blood clots and ruptured blood arteries in the brain are the main causes of strokes Depending on which portion of the brain is injured and how long the brain is deprived of blood flow, a stroke may result in either temporary or permanent disability A brain stroke can be categorized under ischemic stroke and hemorrhagic stroke. Magnetic Resonance Imaging (MRI), Biopsy Test, Computerized Tomography (CT) scan, Positron Emission Tomography (PET) scan, Electro Encephalogram (EEG), and other tests are available for early diagnosis of brain stroke. Existing approaches are noninvasive, but they are constrained by cost and reliability concerns. These diagnostic approaches are insufficient to detect stroke in human head. In the dissertation, a microstrip slot antenna has been designed with stroke model and human head phantom model, and stroke is detected with the help of antenna. In the ISM band frequency range proposed antenna with circular edge slots and circular slot in slot in the front plane and ground slotted plane has been designed. A 3-D seven layer simulated head phantom model is designed and interfaced with the proposed antenna with and without stroke model. The seven layer head phantom model consists of skin, fat, muscle, skull, dura cerebro spinal fluid and brain layers with different thickness, permittivity and conuctivity. The stroke is detected by observing changes in antenna performance parameters viz. return loss, peak gain, radiation efficiency. Sphere shaped stroke is detected by observing variation in the antenna performance parameter. The stroke detection is applied on the cylindrical phantom on the HFSS software. The result shows that the change in performance is experienced as the distance between the antenna and the head phantom is changed. The value of peak gain and radiation efficiency decreases thus depicting the presence of stroke.ThesisItem Open Access Design and analysis of wearable AMC backed antenna on FR4 substrate for WBAN application(G.B. Pant University of Agriculture and Technology, Pantnagar - 263145 (Uttarakhand), 2022-02) Kavita; ParasWireless body area network contains sensors and actuators which work in the proximity of human body .For this purpose an antenna is needed which can work near the human body without any significant change in the bandwidth, gain and radiation efficiency of the antenna. This is a challenging task since it is observed that an antenna working near the human body faces shifts in the bandwidth, reduction in gain and radiation efficiency of the antenna. This is due to the fact that human body is lossy in nature and therefore most of the radiation is absorbed by the body resulting in very little gain and radiation efficiency. A rectangular patch antenna is designed on FR4 substrate and its characteristic is observed. To increase the bandwidth of the antenna the current path is changed by making proper slots on the antenna patch as well as ground plane. The characteristic of the antenna is observed for on body and free space conditions. When the antenna is loaded on human body a significant reduction in the radiation efficiency is observed. To mitigate this problem, an artificial magnetic conductor is used with the antenna. Antenna shows very good bandwidth for both on body and free space condition. With the AMC structure, the bandwidth of the antenna further increases. Also the radiation efficiency of the antenna with the AMC structure is good with very small back lobe radiation for both free space and on body condition. This shows the antenna is a good candidate for wearable application and WBAN.ThesisItem Open Access Analysis and design of substrate integrated waveguide based leaky wave antennas and compact self-quadruplexing antenna(G.B. Pant University of Agriculture and Technology, Pantnagar - 263145 (Uttarakhand), 2021-11) Singh, Abhay Kumar; ParasThe presented work includes different Substrate integrated waveguide (SIW) based leaky-wave antenna (LWA) for beam steering applications. Phased-array antennas, which offer beam steering capabilities by tuning the phases of individual radiating elements, are most commonly used in beam steering applications. However, the tunable phase shifters utilized are limited by the continuity of the phase shift. In contrast, frequency beam steering offers a more economical and convenient solution for a number of applications. This thesis focuses on the solutions of the problems like Open Stop Band (OSB) and beam scanning coverage area. The OSB problem is solved using the impedance matching and lumped modeling technique. In the first contribution, beam steering is defined, and enhanced technology based on a dispersive element is proposed. The dispersive element is used to engineer the slope of the dispersion curve of the unit cell in LWA. Though having a flat response in its dispersion curve, the beam steering coverage is enhanced. The proposed LWA has the capability to cover the entire range 360° of E-plane. In the second contribution, improved impedance balanced condition based on the distributed transmission line model is derived in periodic LWA based on EH2-mode are designed. The EH2-mode LWA has beam steering capabilities from backward to forward quadrants in a quasi-E plane. In the third contribution, two-dimensional beam steering is achieved by incorporating the LWA with the R-KR lens beamforming. By changing the ports of the R-KR lens, antenna beam scans in the E-plane, while by changing the operating frequencies beam scan in the Hplane has been achieved. In the fourth contribution, a SIW-based self-quadruplexing antenna is proposed. This design eliminates the need for employing external circuitry in quadruplexes design. The fabricated prototype responses closely matched theoretical predictions, verifying the proposed concept.ThesisItem Open Access A compact defected-ground UWB antenna for breast tumor detection using SAR analysis(G.B. Pant University of Agriculture and Technology, Pantnagar - 263145 (Uttarakhand), 2022-02) Kandpal, Rekha; ParasTumor is the world's most serious health issue. There are various methods for detecting a tumor in any region of the human body. Breast cancer is one of the most frequent cancers worldwide. Tumor generated by improper tissue growth. It is classified based on its size and location within the body. Magnetic Resonance Imaging (MRI), Biopsy Test, Computerized Tomography (CT) scan, Positron Emission Tomography (PET) scan, mammography, Electro Encephalogram (EEG), and other tests are available for early diagnosis of breast tumors. Existing approaches are noninvasive, but they are constrained by cost and reliability concerns. These diagnostic approaches are insufficient to detect tumors that are deeper than 20% and cause major problems. In the dissertation, an ultra wideband antenna has designed with variable size (i.e. radius) of the tumor and breast phantom models, and the tumor is detected with the help of antenna. An ultra wide band planer antenna with T-shape slot in the front plane and the modified defected ground plane has been designed. A 3-D three layer simulated breast phantom model with radius 30mm is designed and interfaced with the proposed antenna with and without tumor. The three layer breast phantom model consists of skin, fat and fiber layer. The tumor is detected by observing changes in antenna performance parameters viz. electric field distribution, return loss, current density, SAR (specific absorption rate). Sphere shaped tumor is detected by observing variation in the antenna performance parameter. The tumor detection is applied on the semi-hemispherical breast phantom on the HFSS software. The result shows that the change in the radius of tumor, location of tumor and the position of the UWB antenna changes the antenna performance parameters because of the electrical properties of tumor and breast phantom. The value of SAR decreases on increasing the distance between the antenna and phantom as well as increases on increasing the radius of the tumor. The current density, electric field and magnetic field distribution also decreases on moving the tumor tissue deep inside the cell. Thus, SAR is best choice for detection of the malignant tissue inside the phantom and the antenna therefore be used in bio-medical applications.ThesisItem Open Access Real-time detection of Salmonella Typhimurium using optical fiber plasmonic biosensor based on surface plasmon resonance(G.B. Pant University of Agriculture and Technology, Pantnagar - 263145 (Uttarakhand), 2021-08) Joshi, Abhishek; Mathur, SanjayFoodborne pathogens have remarkably caused food contamination at every stage of food production, processing, and distribution. Salmonella is one of the main organisms causing the outbreaks of foodborne illnesses, and pork is one of the major vehicles of foodborne salmonellosis throughout the world. It is therefore highly desirable that the presence of Salmonella Typhimurium be detected at the preprocessing stage of the product been made from any such source which may carry this bacteria and that may contaminate the product. In this thesis an optical fiber index sensor based on surface plasmon resonance is proposed for label-free and real-time detection of Salmonella Typhimurium. The sensing properties of the sensor are numerically simulated in COMSOL Multiphysics5.5. The first layer of silver nanoparticles is coated on the fiber probe sensor and the second layer as grapheme oxide. The reflection properties and the refractive index sensitivity of the sensor are systemically investigated. The results shows that by changing refractive index on the surface of multimode optical fiber gives a remarkable shift of resonance peak. In addition the proposed sensor is very easy to make and the layer of graphene oxide enhances the field to a great extent and also protect the silver film from oxidation. This refractive index sensor is highly sensitive and best choice for bio-medical applications.ThesisItem Open Access Design and implementation of high performance hybrid full adders using XOR-XNOR in 90nm CMOS technology(G.B. Pant University of Agriculture and Technology, Pantnagar - 263145 (Uttarakhand), 2020-12) Kandpal, Jyoti; Tomar, AbhishekThe demand for high computational power of processor and dependency on portable application is constantly growing day by day. In the processor thirty percent of power is consumed by the data path. Arithmetic unit is a critical component of this data path. In arithmetic unit, adder plays a vital role not only for addition but also in multiplier, comparator, error detection and correction etc. Many techniques and topologies have been proposed by various authors in the quest of improved performance of adder in terms of power consumption, lesser propagation delay, full voltage swing at the internal and external nodes, and a smaller number of internal nodes. Hybrid logic style is used for implementing the full adder circuit in this thesis. In the hybrid logic style, appropriate logic style is chosen to realize the different modules of the full adder. In this work, at first a XOR-XNOR module is implemented using the complementary pass transistor logic (CPL) and restorer circuit with 10 number of transistors. Furthermore, using this proposed XOR-XNOR design four new full adders are implemented. The proposed work is simulated using the cadence software tool. The proposed XOR-XNOR design showed 7.5% improvement in the power delay product (PDP). The proposed full adder design showed 2 to 28.13 % improvement in PDP. Inaddition proposed full adder circuits are compared with different reviewed designs in varying supply voltage, process corner and loading conditions. Among the proposed four full adder designs, Design-4 showed best performance. The results of the virtual fabrication of the proposed XOR-XNOR design and Design-4 of full adder circuit are presented and discussed in the present work. The proposed full adder is appropriate for usage in different everyday used electronic applications, for example, in memory and processor-based devices.