School Of Electronics Engineering (Aug – Sep 2022)
- Nayak, A., Parida, M.K., Kumar, V. & Prasanna, G. (2022). Investigation of thermal neutron detection efficiency of Boron Carbide converter material using GEANT4 simulation for different types of detector configurations, Journal of Instrumentation, 17(7), P07012-P07040, https://doi.org/10.1088/1748-0221/17/07/P07012 (Impact Factor: 1.121)
A lot of advancement in the field of semiconductors has made it possible to design a solid-state neutron detector. They are small, have economical bulk fabrications, require low power for their operation. In the present research work, a systematic GEANT4 simulation have been performed on estimating the simulated thermal neutron detection efficiency (η) for different detector geometrical configurations design with Boron Carbide ( 10 B 4 C) as a converter material. These detectors geometry configurations designs are planar, rectangular parallel trenches, cylindrical perforation, stack and spherical. The objective of the simulations was to obtain critical geometrical features for which the efficiency reaches the maximum value.
- Behera, T. M., Samal, U. C., Mohapatra, S. K., Khan, M. S., Appasani, B., Bizon, N., Thounthong, P. (2022). Energy-Efficient Routing Protocols for Wireless Sensor Networks: Architectures, Strategies, and Performance. Electronics, 11(15), 1 – 26, https://doi.org/10.3390/electronics11152282) (IF: 2.690).
In a WSN environment, cluster formation and CH selection consume significant energy. Elementary protocols such as LEACH and C-LEACH are well proven, but gradually limitations evolved due to increasing desire and need for proper modification over time. This paper overviews the modifications in the threshold value of CH selection in the network. With the evolution of bio-inspired algorithms, CH selection has also been enhanced considering behaviour of the network. This paper includes brief description of LEACH-based and bio- inspired protocols, their pros and cons, assumptions, and criteria of CH selection. Finally, the performance factors of various protocols are compared and discussed.
- Nayak, D.; Ray, K.; Kar, T.; Kwan, C. (2022). Walsh–Hadamard Kernel Feature-Based Image Compression Using DCT with Bi-Level Quantization. Computers 2022, 11(7), 110. https://doi.org/10.3390/computers11070110
To meet the high bit rate requirements in many multimedia applications, a lossy image compression algorithm based on Walsh–Hadamard kernel-based feature extraction, discrete cosine transform (DCT), and bi-level quantization is proposed in this paper. The selection of the quantization matrix of the block is made based on a weighted combination of the block feature strength (BFS) of the block extracted by projecting the selected Walsh–Hadamard basis kernels on an image block.
- Bera, S., Shrivastava, V.K., and Satapathy, S.C. (2022), Advances in Hyperspectral Image Classification Based on Convolutional Neural Networks: A Review, Computer Modeling in Engineering & Sciences, 133(2), 219–250, https://doi.org/10.32604/cmes.2022.020601. (IF:2.027).
Hyperspectral image (HSI) classification has been one of the most important tasks in the remote sensing community over the last few decades. Due to the presence of highly correlated bands and limited training samples in HSI, discriminative feature extraction was challenging for traditional machine learning methods. Recently, deep learning based methods have been recognized as powerful feature extraction tool and have drawn a significant amount of attention in HSI classification. Among various deep learning models, convolutional neural networks (CNNs) have shown huge success and offered great potential to yield high performance in HIS classification. Motivated by this successful performance, this paper presents a systematic review of different CNN architectures for HSI classification and provides some future guidelines. To accomplish this, our study has taken a few important steps. First, we have focused on different CNN architectures, which are able to extract spectral, spatial, and joint spectral-spatial features. Then, many publications related to CNN based HSI classifications have been reviewed systematically. Further, a detailed comparative performance analysis has been presented between four CNN models namely 1D CNN, 2D CNN, 3D CNN, and feature fusion based CNN (FFCNN). Four benchmark HSI datasets have been used in our experiment for evaluating the performance. Finally, we concluded the paper with challenges on CNN based HSI classification and future guidelines that may help the researchers to work on HIS classification using CNN.
- Mukherjee, D, Raja, G.L, Kundu, P, Ghosh, A (2022), Modified augmented fractional order control schemes for cart Inverted Pendulum using constrained Luus-Jaakola Optimization, International Journal of Modeling, Identification and Control, Inderscience, 38(3-4), 367-379, https://doi.org/10.1504/IJMIC.2021.123361
The upright position of an inverted pendulum system is an unstable equilibrium and fractional order based control schemes are becoming increasingly popular in stabilising an unstable system. Therefore, a novel combination of FOLyapunov rule and FOPI/two-degree of freedom FOPI (2DOF-FOPI) controller is proposed to tackle this problem. Parameters of FOPI/2DOF-FOPI controllers are obtained using multi-objective constrained Luus-Jaakola multipass optimisation method. Comparative simulation studies are carried out with direct synthesis based PID control scheme, combination of FOMIT rule augmented with either FOPI/2DOF-FOPI controllers. It is evident that the proposed combination of FOLyapunov method and FOPI/2DOF-FOPI controllers outperforms the other schemes.
- Maity, S.K., Dutta, P. & Pandit, S. (2022) Compact Drain Current Modeling of Planar InGaAs Quantum well MOSFET. Micro and Nanostructures, 156(9), 207361.1-207361.14. https://doi.org/10.1016/j.micrna.2022.207361
In this article, we propose a physics-based compact drain current model of planar InGaAs channel-based quantum well MOS transistor. The effects of essential physical phenomenon such as quantum confinement, multiple sub-band energies, wavefunctions and perturbations in sub-band energies are considered in the model by deriving the time-independent Schroedinger wave equation. The potential and inversion carrier profiles are obtained through direct solution of Schroedinger and Poisson equations inside the device. The proposed model also considers other important physical aspects such as band non-parabolicity, velocity overshoot and threshold voltage roll-off. The model is thus physics-based and does not include any empirical fitting parameter. Professional numerical simulator data for a variety of bias voltages and channel thicknesses have been used to validate the expected outcomes of our model. A reasonable agreement between the transistor characteristics as predicted by our model and that available experimentally is obtained, thus justifying the accuracy of our model.
- Ahire, N., Awale, R.N., Patnaik, S. & Wagh, A. (2022). A comprehensive review of machine learning approaches for dyslexia diagnosis, Multimed Tools and Applications (2022). https://doi.org/10.1007/s11042-022-13939-0
EEG can add multiple dimensions towards the identification of learning disability being an abnormality of the brain. The machine learning techniques can examine, classify, and process EEG signals to accurately understand brain activities and disorders. This paper is a comprehensive review of the application of machine learning techniques in the classification of EEG signals of dyslexia and analysis of an improved framework to extemporize the classifier’s performance and accuracy in discriminating between dyslexics and controls. The presence of noises and artefacts often reduces the performance of classifiers and hampers results. This study reviews input pre-processing, feature selection, feature extraction techniques and machine learning algorithms for the early detection of disorder.
- Patnaik, S. (2022). Speech emotion recognition by using complex MFCC and deep sequential model. Multimedia Tools and Applications (2022). https://doi.org/10.1007/s11042-022-13725-y
This paper is about emotion classification by using Complex Mel Frequency Cepstral Coefficients (c-MFCC) as the representative trait and a deep sequential model as a classifier. The main contributions of this work are of two-folds. Firstly, introducing conception of c-MFCC and investigating it as a robust cue of emotion and there by leading to significant improvement in accuracy performance. Secondly, establishing correlation between MFCC based accuracy and Russell’s emotional circumplex pattern. Emotional signals are combinations of several psychological dimensions though perceived as discrete categories. Results of this work are outcome from a deep sequential LSTM model. Proposed c-MFCC are found to be more robust to handle signal framing, informative in terms of spectral roll off, and therefore put forward as an input to the classifier.
- Tappeta, V.S.R., Appasani, B., Patnaik, S. & Ustun, T.S.S. (2022) A Review on Emerging Communication and Computational Technologies for Increased Use of Plug-In Electric Vehicles, Energies, 15(8), 6580. https://doi.org/10.3390/en15186580
The electric vehicle (EV) shows unprecedented behaviour during vehicle battery charging, and sending the charge from the vehicle’s battery back to the grid via a charging station during peak hours has an impact on the grid operation. Balancing the load during peak hours, i.e., managing the energy between the grid and vehicle, requires efficient communication protocols, standards, and computational technologies that are essential for improving the performance, efficiency, and security of vehicle-to-vehicle, vehicle-to-grid (V2G), and grid-to-vehicle (G2V) communication. This paper presents existing literature on emerging protocols, standards, communication technologies, and computational technologies for EVs. Frameworks, standards, architectures, and protocols proposed by various authors are discussed in the paper to serve the need of various researchers for implementing the applications in the EV domain.
- Barreto, F., Sarvaiya, J. & Patnaik, S. (2022) Learning Representations for Face Recognition: A Review from Holistic to Deep Learning, Advances in Technology Innovations, 7(4), 279-294. https://doi.org/10.46604/aiti.2022.8308
This study reviews the development of different face recognition (FR) methods, namely, holistic learning, handcrafted local feature learning, shallow learning, and deep learning (DL). With the development of methods, the accuracy of recognizing faces in the labelled faces in the wild (LFW) database has been increased. The accuracy of holistic learning is 60%, that of handcrafted local feature learning increases to 70%, and that of shallow learning is 86%. Finally, DL achieves human-level performance (97% accuracy). This enhanced accuracy is caused by large datasets and graphics processing units (GPUs) with massively parallel processing capabilities. Furthermore, FR challenges and current research studies are discussed to understand future research directions. The results of this study show that presently the database of labelled faces in the wild has reached 99.85% accuracy.
- Barreto, F., Sarvaiya, J., Patnaik, S. & Yadav, S.K. (2022). Unsupervised Domain Adaptation using Maximum Mean Covariance Discrepancy and Variational Autoencoder, International Journal of Advanced Computer Science and Applications (IJACSA), 13(6), https://doi.org/10.14569/IJACSA.2022.01306104
Face Recognition models performed very well on the benchmark datasets, but their performance sometimes deteriorated for real-world applications. Few researchers looked at Unsupervised Domain Adaptation (UDA) to find the domain-invariant feature spaces. They tried to minimize the domain discrepancy using a static loss of maximum mean discrepancy (MMD). From MMD, the researchers delved into the higher-order statistics of maximum covariance discrepancy (MCD). MMD and MCD were combined to get maximum mean and covariance discrepancy (MMCD), which captured more information than MMD alone. We use a Variational Autoencoder (VAE) with joint mean and covariance discrepancy to offer a solution for domain adaptation. Analysis was done using the TinyFace benchmark dataset and the Bollywood Celebrities dataset. Three objective image quality parameters, namely SSIM, pieAPP, and SIFT feature matching, demonstrate the superiority of MMCD-VAE over the conventional KL-VAE model.
- Samant, T., Banerjee, S., Rath, M.K. & Swain, T. (2022) Human Activity Recognition using Signal Processing and Classical ML Algorithms, 2022 IEEE 7th International conference for Convergence in Technology (I2CT), 1-6. https://doi.org/10.1109/I2CT54291.2022.9825173
For many of the modern-day applications, human beings are being tracked based on their activity all throughout a given period. The applications may be in surveillance systems, health care, human interaction, marketing, etc. This has become a specific field of study altogether and people are extensively working on this field to find better solutions to the given real-world problems. The data is being collected through devices such as smartwatches (Fitbits, Apple Watch, etc.) or smartphones and is further used to analyze the activity of the human by whom the device is being used. Here we have used the data from UCI Machine Learning Repository, which gives us the information acquired from 30 peoples’ smartphones. We have gyroscope and accelerometer data upon which appropriate signal processing techniques have been applied to generate features. Then these features are fed to various classical ML and ensemble models, and we perform a comparative analysis by looking into their specifics and accuracies.
- Banerjee, S., Rath, M., Swain, T. & Samant, T. (2022) Music Generation using Time Distributed Dense Stateful Char-RNNs, 2022 IEEE 7th International conference for Convergence in Technology (I2CT), 1-5, https://doi.org/10.1109/I2CT54291.2022.9824167.
Sequence generation is one of the state-of-the-art topics in recent days, where given a sequence of inputs, the aim is to generate a similar sequence of outputs in a given context. The applications range from sentence autocompletion in mail bodies to text suggestions for automatic replies, etc. A similar idea can therefore be utilized to generate a sequence of musical notes using some LSTM/GRU based architecture, where we train our model based on given sequences of musical notes. If we look at music of a certain genre, it’s simply a time series data where the data is in frequency domain. Fortunately, for musicians these frequencies have a notation, which can serve as texts to train any given time series model. Hence, in this paper we propose a Char-RNN based model, which can understand the patterns in each composition or a raga and generate new piece of music based on that. The model must not simply copy paste the sequence or generate any random note at a given instant of time but be capable enough to grasp the patterns in which the given piece of music is based upon and create a similar, new piece of music out of that.
- Dash, B.B., Banerjee, S., Samant, T., Swain, T. & Rath, M.K. (2022). Large Scale Follower Recommendation in Instagram, 3rd International Conference for Emerging Technology (INCET), 1-6, https://doi.org/10.1109/INCET54531.2022.9824114
With advancements in predictive and prescriptive analytics over the past few years, various businesses have been extensively using recommender systems for marketing purposes. Say whether it’s some sort of association analysis for customer market or any social networking for connecting people together, this very concept of recommenders has been utilized up to a very large extent over the past decade. When it comes to implementing the underlying concepts behind these systems, most of the time we get to have a paradox of choice of various techniques, which need to be carefully analyzed and acted upon to get the desired accuracy of the model. In cases where the recommendations are made based on a connected graph, we need proper feature generation techniques, which play an important role in training the model as per our objective. In this paper, we propose a combination of feature extraction techniques employed over a Gradient Boost Decision Tree model for recommending people whom a person is likely to follow. We also shall investigate a few ways where the recommendation made based on the common followers of two or more profiles and how likely two or more people may or may not know each other. In any problem, hyperparameter tuning plays the most important part which is a key aspect to optimize the model (so that it doesn’t overfit or underfit the data available) and is totally problem specific.
- De, U.C., Banerjee, S., Rath, M.K., Swain, T. & Samant, T. (2022). Content Based Apparel Recommendation for E-Commerce Store, 3rd International Conference for Emerging Technology (INCET), 1-6, https://doi.org/10.1109/INCET54531.2022.9824870.
Recommendation systems are extensively in use these days. These recommendations may be in the form of friend suggestions on Facebook, suggesting similar questions in Quora, or product suggestions on e-commerce sites, etc. Whenever we use an e-commerce website or app, we get to see product recommendations based on previous search history. Starting from OTT platforms like Prime and Netflix to general e-commerce sites like Flipkart and Amazon all of them use this feature so that the product search for the end-user becomes easier. It’s estimated that e-commerce platforms generate around 35% of their revenue just by these recommendation systems which run for their users. In this paper, we present a content-based recommendation system for women’s apparel where, given an apparel, the system generates other apparel to the users, which are similar to the query apparel. We use various text-based techniques to retrieve the information from the product page based on the product image and its description.
- Mukherjee, D, Raja, G.L, Kundu, P, Ghosh, A. (2022). Design of optimal fractional order lyapunov based model reference adaptive control scheme for CSTR, Editors: Indrani Kar, Elsevier, IFAC, pp. 436-441.
When a first-order irreversible exothermic reaction takes place in a CSTR, the plant model relating the reactor temperatures along with delay is of second-order unstable type. Hence, this article focuses on developing an optimal FOL based MRAC scheme. The optimal values of adaptive gain, as well as extra degree of freedom, are obtained using a modified PSO algorithm. The closed-loop responses and control efforts are compared with that of FOL based MRAC scheme by using an ABC algorithm. Moreover, the aforementioned optimization methods are also employed to develop respective optimal FOMIT based MRAC schemes for comparison purposes. Simulation studies validate the effectiveness of the proposed particle swarm optimized FOL-based MRAC scheme.
- Mukherjee, D, Raja, G.L, Kundu, P, Ghosh, A (2022), Analysis of Fractional Calculus-Based MRAC and Modified Optimal FOPID on Unstable FOPTD Processes, Das, B., Patgiri, R., Bandyopadhyay, S., Balas, V.E. (eds), Springer, Singapore, pp. 202-209.
This article shows a novel approach of fractional calculus using both G-L and R-L methods which are used to develop fractional order tuning rule of MRAC for first-order CSTR with time delay. The efficacy of FOPID controller is also investigated, and its parameters are computed using a modified PSO algorithm. Comparative simulation studies are carried out between normal PSO and modified PSO on time domain metrics. Another comparative simulation studies are investigated between fractional and conventional Lyapunov rule. The effect of perturbation in plant model is also studied to show efficacy of the rule.