Skip to main content

2024 | Book

Recent Advances in Electrical and Electronic Engineering

Select Proceedings of ICSTE 2023

Editors: Bibhu Prasad Swain, Uday Shanker Dixit

Publisher: Springer Nature Singapore

Book Series : Lecture Notes in Electrical Engineering


About this book

This book presents select proceedings of the International Conference on Science, Technology and Engineering (ICSTE 2023) related to electrical and electronic engineering. Various topics covered include neural network classification, text detection from natural scene images, speech processing systems, Wi-Fi intrusion detection, machine learning, wireless sensor network, image retrieval, automatic speech recognition, device physics, power transfer, photovoltaics, antenna for ultra-wideband applications, electric vehicles, etc. The book is useful for researchers and professionals whose work involves electrical and electronics and computer science fields.

Table of Contents


Electrical Engineering and Electronics & Communication Engineering

Cost-Effective Solution of Grid Dispatch Problems Using a Novel Rao-4 Algorithm

One of the most pervasive challenges in the realm of energy systems is ensuring the economical operation of the systems that generate electric energy. To resolve the issue of grid dispatch, a novel approach known as the Rao-4 Optimization Algorithm is suggested in this piece of research. On the IEEE 30-bus test system, this approach is examined and evaluated for seven distinct cases. The valve-point impact and many additional difficulties have been taken into consideration, in addition to the conventional producing fuel cost. In addition, a variety of goals, including the strengthening of voltage stability, improvement of the voltage profile, and reduction of emissions, are taken into consideration. The results that are achieved are then contrasted with the results that are acquired utilizing several well-known optimization strategies. This comparison demonstrates how effective the Rao-4 approach is in solving a variety of grid dispatch problems, including those with complex and non-smooth objective functions.

Shuvam Sahay, Ramanaiah Upputuri, Pooja Kumari, Niranjan Kumar
Comparative Analysis of Gate Driver Circuits for GaN MOSFET-Based Class E Resonant Inverter

There are evolutionary changes that may be seen in power electronics applications and industries, which brought the necessities of higher power density and higher frequency semiconducting devices. These requirements can be achieved with the help of new generation wide band gap semiconducting devices such as silicon carbide (SiC) and gallium nitride (GaN). The power switching devices based on GaN material have greater performance in comparison with silicon (Si) and SiC material-based semiconducting devices. In this work, the fundamental driving requirements of the GaN MOSFET are discussed in detail. A totem pole resonant gate and transient resonant gate drivers are considered for driving the GaN MOSFET. The simulation studies for both types of driver circuits are analyzed by using LTspice simulation. The various losses such as gate driving loss, conduction loss, and switching losses are also compared for the performance of different gate driver circuits of GaN MOSFET-based class E resosnant inverter. Finally, the performance of the class E resonant inverter is validated using LTspice simulation software.

Vikram Kumar Saxena, Kundan Kumar
High-Power Wireless Power Transfer (WPT) System Using Series-Series (SS) Topology

Due to its various uses, wireless power transfer (WPT) technology has recently become one of the most significant study areas. This technology is being pushed into new fields of study and research as more and more innovative thoughts and ideas is emerging daily. WPT technology has undergone significant advancements due to research and development, both in low- and high-power applications. Electric vehicles are charged employing high-power sources. Therefore, it is always difficult to charge electric vehicles in a way that is secure, effective, and safe. Most of these difficulties can be overcome through wireless power transfer (WPT). WPT transformers are operated at high frequencies to provide significant amounts of power with excellent efficiency for electric vehicle charging stations. High-power WPT systems often need specialized high-frequency voltage source inverters. For high-power WPT applications, a high-frequency voltage source resonant inverter employing SS topology has been discussed in this study. The proposed WPT transformer design has been optimized using Ansys Maxwell and a 5.5 kW WPT system has been validated using Ansys Simplorer circuit simulation.

Prabhat Chandra Ghosh
A Multi-criteria Decision-Making (MCDM) Approaches for Systematic Analysis and Ranking of Solar Power Plant Site Using ANN

Currently, environmental policies dedicated to adopting the progress, application, and performance of renewable energies (RES) are in focus. With the development of renewable energy sources, utilizing solar power has also become greater. Government of India (GOI) has also introduced various policies to promote diffusing of solar energy and has also invested a huge sum of money in the development of renewable energy. Because of this reason, choosing a suitable location for a solar farm power installation and getting the best radiation has become a very important agenda. Solar power plant site selection itself requires identifying and analyzing different multi-criteria such as environmental criteria, geographical criteria, economic criteria, and climatic criteria. The goal of the current work is to create a model that will help decision-makers rank or categorize different solar power plants sites in India using multiple criteria attributes applying artificial neural network (ANN) such as multiple layer perceptron (MLP) back-propagation (BP) and genetic algorithm (GA). This work shows a decisive effect, careful use of resources, and systematic decision support framework that will help the future policy planner throughout the review process for choosing an appropriate solar farm site in different states of India.

Hormi Kashung, Benjamin A. Shimray
Real-Time Simulation of an Islanded Mode Solar PV System with New Elman NN-Based MPPT

The ever-increasing demand for clean electrical power and the ability to provide energy in the interior location where the main grid is not available has made solar PV systems popular and demanding. Therefore, in this article, a solar PV system is considered to analyze its performance with the real-time simulator. In the PV system to track the maximum power, a new Elman neural network-based MPPT is used and a battery energy storage system is installed to store surplus energy and supply power whenever required. In this study, one three-phase converter (load-side converters), unidirectional and bidirectional DC-DC converters, and their control schemes are used to manage the power flow. The system is implemented in MATLAB/Simulink 2021A environment and validated on real-time simulator OPAL-RT. The simulation results are analyzed. Results show the proposed control scheme works effectively and can be implemented in real-life.

Bappa Roy, Shuma Adhikari, Aribam Deleena Devi, Kharibam Jilenkumari Devi
Different Strategies for Estimation of State of Charge for Battery Packs of Electric Vehicle

The world is advancing toward an era with infrastructure that has been updated with more modern inventions. At the same time, pollution is increasing, causing constant harm to all living creatures. The problems associated with environmental contamination are currently the most concerning topic for the researchers. Internal combustion engines (ICEs) in conventional vehicles have a significant impact on this problem, causing the greenhouse effect and having a noteworthy emission rate. Because EVs use electrical energy to drive their motors efficiently, environmental researchers are paying close attention to the development of electric vehicles. As battery is one of the most important parts of electric vehicle, we need to pay extra attention on different types of battery management strategy, issues related to battery and its impact on environment.

Pooja Kumari, Durgesh Choudhary, Shuvam Sahay, Niranjan Kumar
Novel Current Mode First-Order Filter and Oscillator

A novel current mode first-order universal filter is developed using the dual X second generation multi-output current conveyor (DX-MOCCII), capacitors and resistors to produce current mode low pass, high pass and all pass responses. The filter’s sensitivity is quite low. In the paper being given, a new sinusoidal oscillator based on DX-MOCCII is introduced. The oscillator’s frequency can be adjusted without affecting the circuit structure. Using PSPICE simulator and parameters for a 0.25 µm CMOS (TSMC) model, the block was built. The effectiveness of the filter is assessed using the Cadence OrCAD simulator. Results from simulations back up the theory.

Ashok Kumar, Ajay Kumar Kushwaha
Eigenfrequencies of Functionally Graded Plates with Rectangular Cutout

In recent times Functionally Graded Materials, also they are commonly known as FGM widely used in place of composites for their immense advantages. Vibration analysis is crucial in the field of structural analysis due to its wide application in many practical engineering problems. Plates are fundamental elements of many structural engineering applications and have been studied extensively. Cutouts weaken a structure because they alter the load lines surrounding it. However, cutouts can only be avoided in some cases. Therefore, it is necessary to determine the structure's eigenfrequencies with cutouts. The present work employs finite element analysis to get the eigenfrequencies and mode shapes of the FGM plate with a rectangular cutout. “Nine-noded isoparametric elements” with forty-five degrees of freedom are used for the finite element formulation. The entire finite element program, based on “first-order shear deformation theory (FSDT)”, has been written in MATLAB. Power law governs the continuous change of the material properties through the thickness of the FGM plate. Initially, the current outcomes are compared to previously reported outcomes for FGM plates without a cutout. Then, new results are found for various boundary conditions, cracked corners, aspect ratio, power index and thickness ratio. The current formulation is robust and is capable of producing solutions with a high degree of accuracy.

Kushal Jana, Aditi Majumdar, Salil Haldar
Floating Inductance Simulator with EXCCTAs

This paper introduces a floating inductance simulator with two extra X current conveyor transconductance amplifiers (EXCCTA) and three passive elements. The presented floating simulator (FS) has no passive component matching condition and operates at high frequency. The simulator employs one floating resistor and two grounded capacitors; the circuit is suitable for an integrated circuit (IC) platform. The presented FS has electronically tunable property due to having an inbuilt transconductance term (gm). The proposed FS is utilized in the tunable band rejection filter (BRF) and higher-order low-pass filter (LPF) application circuit for the workability test. All the simulation results are performed with 0.18 µm TSMC CMOS technology parameter with ±1.25 V. The theoretical approach is verified by computer simulation and experimental responses. The introduced FS is simple and utilizes two current-feedback operational amplifiers (CFOA) ICs (AD844) and one operational transconductance amplifier (OTA) ICs (LM13700) in the experimental arrangement.

Y. Shantikumar Singh, Ashish Ranjan, Shuma Adhikari, Benjamin A. Shimray
Comparative Analysis of Microstrip and Co-Planar Waveguide-Fed Printed Monopole Antenna for Ultra-Wideband Application

In this design, a circular shape monopole antenna with dimension 30 × 35 × 1.6 mm3 is designed on the FR-4 substrate. It is excited either by microstrip or CPW-fed line. A comparative study is made for both the fed line for the ultra-wideband range. In the case of VSWR, both feeds show that it is below 2 which satisfies the UWB technology. The return loss of −41.86 dB, fractional bandwidth of 89% and voltage standing wave ratio (VSWR) are below 2 for CPW fed, whereas in the case of microstrip fed, return loss is −39.5 dB and fractional bandwidth of 80.1%. A gain of 8.77 dBi is seen in CPW and 6.45 dBi for microstrip fed. These parameters indicate that CPW fed gives better performance.

Samom Jayananda Singh, Rajesh Kumar, Kharibam Jilenkumari, M. M. Dixit
Estimation of the Post-burning Area of the Fire Hazard Severity Zone in California from Landsat 8 OLI Images Using Deep Learning Machine Intelligence Model

A forest fire is an unplanned and unexpected event that contaminates our environment by burning the grass, trees, and vegetables of any forest and have an impact on the country’s economic model and the functioning of our ecosystem. A forest fire begins as a result of a natural occurrence, human activities, global warming, or lightning strikes. California is one of them that is affected by fires everyday. However, the statistics and procedures needed to reliably characterize fire propagation, behavior, and consequences are still lacking. As a result, our understanding of wildfire behavior remains restricted. In this paper, we have analyzed the area affected by the fire of August 7, 2016, with the help of machine intelligence deep learning model from Landsat 8 OLI images. The model has a total of five hidden layers, and the model has been trained with seven input features. The accuracy of the model has been obtained as 99.4712% with a baseline error of 0.53%. The area affected by a large fire to the north of San Bernardino in the state of California (USA) is 171.730934 km2 (42,435.637956 Acres). We used this approach to map California’s fire history in August 2016.

Mohan Singh, Kapil Dev Tyagi, Arti Joshi
Algorithm Analysis in NOMA

Decrease in latency and massive connectivity are some of the focused application scenarios in the upcoming-generation wireless networks. Non-orthogonal multiple access (NOMA) is considered as one of the key technologies for 5th-generation wireless communication due to its high spectral efficiency. The two NOMA techniques are power-domain NOMA and code-domain NOMA. The fundamentals of power-domain NOMA are discussed in this paper. It also briefly discusses the difference in uplink and downlink communication in the system model. From that viewpoint, this paper surveys about the algorithms for detection and decoding in the order of power levels used in superposition coding at the transmitter side and successive interference cancellation (SIC) at the receiver side. The obtained result indicates how users are multiplexed in the power domain by following the algorithm of superposition coding at the transmitter. Furthermore, interference has been removed to give every individual user separate messages by following SIC at the receiver.

Moirangthem Rushdie Devi, Aheibam Dinamani Singh
A New High-Order Electronically Tunable Oscillator (HO-ETO)

In this work, a current conveyor (CCII) as an active element is used to design the novel high-order electronically tunable oscillator (HO-ETO) which offers electrical tenability with resistance in the X terminal. The suggested oscillator makes it easier to design integrated circuits with an extra-X terminal in CCII with three grounded capacitors and a resistor. The proposed HO-ETO circuit simultaneously generates both types of outputs (voltage and current). The implementation of HO-ETO design is well-validated mathematically, and Cadence software results are demonstrated to support the theoretical investigations.

Manoj Joshi, Rakesh Kumar, Mohan Singh, Shilpa Choudhary
Analysis and Simulation of Misalignment Issues in Dynamic Wireless Charging for Electric Vehicles

It is well known that internal combustion (IC) engines-based vehicles driven by fossil fuels are one of the major contributors to global warming. The new technology that is capable to replace IC engine-based vehicles is the battery-operated vehicle, i.e., electric vehicle (EV) as well as vehicles driven by means of electricity. The major benefits of developing EV technologies are economic, environmental-friendly, and highly efficient than fuel vehicles. The biggest drawback of EVs over IC engines is the longer time needed to completely charge their batteries. The various methods to charge the EVs are wired charging and wireless charging. Wireless charging can be further classified as static, quasi-static, or quasi-dynamic, and dynamic wireless charging. This work focuses on misalignment issues in dynamic wireless charging (DWC). A 3-D analytical model of a circular helix coil is presented and simulated through computer-assisted-based simulation software, i.e., ANSYS MAXWELL 3-D. The various simulation results such as coupling coefficient, flux linkage, magnetic flux density, and mutual inductance with the variation of axis parameters are presented to validate the various misalignment issues.

Kundan Kumar, Ngangoiba Maisnam
An Asymmetric CPW-Fed Compact Dual-Band Circularly Polarized Monopole Antenna

This paper proposes a compact, dual-band circularly polarized monopole antenna. Here, a rectangular strip with an I-shape is utilized as radiating element and 50 Ω microstrip inset line feeding is used to trigger the essential vertical and horizontal component for achieving circular polarization. The asymmetrical coplanar waveguide (CPW) fed ground plane is used for obtaining axial ratio below 3 dB within the impedance bandwidth region. The size of the antenna is 32.62 × 22.69 mm2 (0.674 × 0.468 λ02). The proposed antenna generates dual impedance bandwidth of 6.22–8.29 GHZ and 15.99–18.58 GHz which are resonating at 7.25 and 17.28 GHz, and they are 28.55% and 14.98%. The dual circularly polarized ARBW ranges from 7.68 to 8.08 GHz and 15.11–16.19 GHz which are resonating at 7.84 and 15.65 GHz, they are 6.12% and 6.90%, and they are LHCP radiation patterns. One can employ the lower resonant frequency band for broadcasting and fixed mobile application, and higher resonating band can be used for broadcasting and fixed radio-location application.

Mehtaz Marin, Reshmi Dhara, Sanoj Mahato
Massive MIMO Systems Precoder Design Under Various Environments

The day-to-day increased usage of wireless-connected devices like smart phone, tablets, and laptops leads to the increase in data traffic demand in communication. The upcoming generation like 5G- or 6G-based communications system supports to serve large number of users communicating simultaneously for data transfer with less energy consumption-based additional core requirement. The concept of introducing the base station with thousands of antennas and parallel serving of multiple users simultaneously structure is achieved in the promising technologies called massive MIMO system. The system energy efficiency and radiation energy under concentration are made minimized into smaller area; it enhances the massive antennas systems. This paper discusses the performance of massive MIMO system under various precoding techniques like MRT, MMSE, Rayleigh-based ZF channel fading system under complete state and incomplete state in the transmitter in MIMO System. The results address the achievable sum rate and minimized bit error under various scenarios.

R. Srividhya, M. Anto Bennet
A Miniaturized Modified Koch Fractal Antenna with Partial Ground Plane for Wideband Application

This work describes a diminished, altered Koch fractal antenna for wideband applications with enhanced linearly polarized impedance bandwidth. The implemented antenna is designed from the first iteration to the third iteration. A partial ground plane has been used on the third iteration to achieve wideband characteristics. To improve the impedance bandwidth, the length and width of the ground plane are varied. Finally, a modified partial ground plane with a notch on the leftmost side is proposed to introduce capacitance, resulting in an ultra-impedance wide bandwidth through ground-to-radiator coupling. The number of resonances increases with increasing iteration. This extends the physical length at the radiator's perimeter, allowing the radiator to resonate low-frequency causes without increasing the radiator's size. However, when the number of iterations increases beyond three, the radiator design gets increasingly complicated, with no substantial improvement in the number of S11 < −10 dB bandwidths or frequency response for that area. The third iteration is selected as the best design for the proposed antenna. Ansys HFSS is used to design the antenna. The proposed antenna has a resonance frequency of 3.0 GHz and a calculated impedance bandwidth of 2.62 GHz to 4.48 GHz. The antenna used can help WiMAX and S-band wireless communication applications.

Tejaswi Kumar, Reshmi Dhara, Sanoj Mahato
Comparative Study on YOLOv2 Object Detection Based on Various Pretrained Networks

Object detection is an important topic in computer vision. You only look once (YOLO) is a fast and effective object detection algorithm. In this work, various object detection models are constructed by using different types of pretrained networks and YOLOv2 detection layers. The pretrained network portion of the model is used for the feature extraction and the YOLOv2 portion for the object detection. Training, testing, and validation of these object detection models are performed with a common vehicle dataset. A comparison is presented in terms of precisions and model sizes for the various detection models at the end of the research. The YOLOv2 detection model based on MobileNetv2 has the highest precision of 81.64% among the eight models. The ShuffleNet-based detection model has the smallest model size of 6.16 MB which is suitable for platforms with limited resources.

Richard Ningthoujam, Keisham Pritamdas, Loitongbam Surajkumar Singh
Data Transmission and Optimization of Energy in Smart Campus Using LoRaWAN Industrial IoT Technology

Our idea is to build a Long-Range Wide-Area Network (LoRaWAN)-based smart campus monitoring system, in order to replace the current technology of short-range communication devices. This paper mainly focuses on transmitting the data collected from end nodes or wireless sensors like Digital Humidity Temperature Sensor–11 (DHT11), Ultrasonic Sensor, Passive Infra-Red Sensor (PIR), and Ambient Light Sensor (ALS), etc., to the network server using gateway, end nodes. The end users make use of this data for monitoring and controlling several appliances. The network interface is implemented by LoRa which solves communication failure problems and saves energy during data transmission with 100 percent accuracy. This LoRa supports moderate data rates which reduces energy consumption in turn optimizes the energy utilization in the campus. This paper explains how the sensor data is transferred from the sensor using the LoRa technology.

Ramasamy Mariappan, Ch. S. V. N. S. L. Amulya, M. Yogisri Vasanthi, Pothapu Aditya, Ch. Sai Manohar
An Implementation of Differential Difference Voltage Difference Transconductance Amplifier (DD-VDTA) and Its Application as a Dual Output Integrator

This research article brings a design of Differential Difference Voltage Difference Transconductance Amplifier (DD-VDTA), which enables the Differential Difference property in a traditional VDTA block. It can be achieved by adding Differential Difference Amplifier (DDA). The verification of the DD-VDTA block is first verified with both DC and AC characteristics using PSPICE simulation. Moreover, the applications of this block is extended for the design of lossy inverted and non-inverted integrator and lossless integrator. The workability test of the DD-VDTA-based dual output inverted and non-inverted integrator is well simulated in PSPICE.

Prerna Rana, Ashish Ranjan
A Compact Dual-Element MIMO Antenna with High Isolation for Wideband Applications

A two element Multiple Inputs Multiple Outputs (MIMO) antenna which possesses very low mutual coupling with a compact size of 0.32 $${\uplambda }_{0}$$ λ 0 × 0.36 $${\uplambda }_{0}$$ λ 0 has been proposed for wideband application. The MIMO antenna involves two antenna elements with I-shaped isolation structure between them. The antenna has achieved impedance bandwidth ranging from 3.15 GHz to 8.5 GHz which covers C-band (4–8 GHz), WLAN (5.1–5.8 GHz), ISM band (5.2/5.8 GHz) and 5G (3.3–5 GHz). The results of simulation show that the antenna has isolation greater than 20 dB throughout the operating frequency band. The antenna uses FR4 as substrate with thickness 1.6 mm. The antenna has a simple planar structure for which the design is easy with simple process involved.

Mahd Azharuddin, Deepak Kumar Barik, Kalyan Mondal, Lakhindar Murmu, Tapan Mandal
Design of GWO-MBIMC Controller to Stabilize the Frequency of Microgrid on Real-Time Simulation [OPAL-RT OP4510] Platform

The frequency of an independent microgrid (MG) system is attempted to be controlled in this work utilizing an IMC-based PID controller and Droop-based controllers (DBC) or modified bias (MB) controller with Grey Wolf Optimization (GWO) in extremely unfavourable conditions. Frequency deviation occurred in the system because of mismatch between Actual Generated power and total load demand. To address this issue the MB & IMC-based gain parameters of the microgrid are taken into consideration for finding the limits of control parameters for GWO, which will make sure that the search points produced by GWO are robust and stable. In real-time simulation environments using the digital simulator OPAL-RT, simulation of the proposed (GWO-MBIMC) controller-based MG model has been appropriately tested. In various real-life situations, the system performance is obtained using the GWO-MBIMC controller and system outputs are closely investigated. The performance of proposed controller is compared with that of PSO-MBIMC, MBIMC, and MB-LDR controllers. The result shows that, in forms of peak deviation, time of settling and a smaller no. of oscillations, the proposed GWO-IMC controller provides the most optimum adaptive output.

Badal Kumar, Shuma Adhikari, Nidul Sinha
Improved Polar Extensions of an Inequality for a Complex Polynomial with All Zeros on a Circle

If $$f\left( w \right)$$ f w is a polynomial of degree $$m$$ m having all its zeros on $$\left| w \right| = \rho$$ w = ρ , $$\rho \le 1$$ ρ ≤ 1 , then Govil proved that $$\mathop {\max }\limits_{{\left| {\varvec{w}} \right| = 1}} \left| {\user2{f^{\prime}}\left( {\varvec{w}} \right)} \right| \le \frac{{\varvec{m}}}{{{\varvec{\rho}}^{{\varvec{m}}} + {\varvec{\rho}}^{{{\varvec{m}} - 1}} }}\mathop {\max }\limits_{{\left| {\varvec{w}} \right| = 1}} \left| {{\varvec{f}}\left( {\varvec{w}} \right)} \right|.$$ max w = 1 f ′ w ≤ m ρ m + ρ m - 1 max w = 1 f w . In this paper, we proved refined extensions of this inequality in polar derivative under the same hypothesis.

Kshetrimayum Krishnadas, Barchand Chanam
Universe with Power Law Expansion

This paper presents Inflationary scenarios in $$f(R, T)$$ f ( R , T ) gravity using power law, with scalar curvature $$R$$ R and energy momentum tensor trace $$T$$ T . Power law expansion of scale factor is used to determine precise solutions to field equations (FE) of Bianchi type-I (BT-I) model. Expansion rate of universe changes with time, fluctuating between an infinite pace at starting and a fixed rate at later epochs. The current model’s energy conditions have been established, and the results show that it is consistent with the data. The strong energy condition (SEC) is failed which implies the accelerated phase of the universe. The model’s geometrical and the physical behaviours have been discussed.

S. Surendra Singh, Nikhil Swami
An Optimal Fourth-Order Iterative Method for Multiple Roots of Nonlinear Equations

In this paper, we are presenting an iterative scheme for solving nonlinear equations having multiple roots. The newly developed scheme is an improvement of a method for simple roots and it satisfy the Kung-Traub conjecture, so it is optimal. The weight functional approaches used to develop the method. We have analysed its convergence order and proved it. The methods are numerically compared with known methods in terms of the convergence behaviour of convergence, it shows that the developed schemes are superior to existing methods.

Waikhom Henarita Chanu, Sunil Panday, Shubham Kumar Mittal, G Thangkhenpau
A Dual-Band Circularly Polarized with Large Impedance Bandwidth Planar Monopole Antenna for Wireless Application

The outlining and design of a dual-band planar monopole antenna is presented in this work. A novel structured monopole antenna is designed and analyzed for dual-band applications. The proposed work covered dual circularly polarization (CP) bands along with broad impedance bandwidth (IBW). The designed antenna consists of a half-moon, full moon slotted rectangular modified radiator with partial ground plane. By fine-tuning, the slot size and ground plane with desired CP radiation are found at desired impedance bandwidth. The physical dimension of the antenna is given by $$0.49\lambda_{0} \times 0.40\lambda_{0} \times 0.02\lambda_{0}$$ 0.49 λ 0 × 0.40 λ 0 × 0.02 λ 0 . The −10 dB IBW is 11.01 GHz (3.59–14.6 GHz) with center frequency of 9.1 GHz having parentage of IBW 121.05% is conformed. The dual 3 dB axial ratio bandwidth (ARBW) of 2.24 GHz (3.65–5.89 GHz) and 0.7 GHz (9.05–9.75 GHz) are achieved. The proposed antenna is very simple to design, and it can be applied for 5G application, S-band radar, and maritime radio navigation.

Deepak Kumar Barik, Mahd Azharuddin, Kalyan Mondal, Lakhindar Murmu, Tapan Mandal
FPGA-Based True Random Number Generator Architecture Using 15-Bit LFSR and ADPLL

A true random number generator (TRNG) based on a linear feedback shift register (LFSR) and an all-digital phase-locked loop (ADPLL) is a type of hardware-based random number generator that uses the principles of both LFSR and an ADPLL to generate random numbers. We extend an approach that uses on-chip jitter and metastability state in this study. 15-bit LFSR with ADPLL-based TRNG (15-LAT) architecture is created with the help of ring oscillators, flip-flops, ADPLL, 15-bit LFSR, and other physical devices that generate various sources of entropy. When compared to other existing TRNG designs, we established that our novel design lower the complexity as well as decreased the power usage (0.072W) due to the use of minimum FPGA hardware resources. This architecture offers better performance over conventional TRNGs and opens up possibilities to design a more efficient, low-power producing highly reliable TRNG. The architecture was tested on an Artrix-7 FPGAs evaluation board with positive results. A digital storage oscilloscope (DSO) is used to record the resulting pattern and FFT pattern of the corresponding wave. We provide the results of statistical analyses performed on the output bit sequence generated by our design using the MATLAB tool. The NIST SP 800-22 assessment demonstrates that the output TRNG bitstreams are unpredictable and stochastic, suggesting that the suggested architecture is better suited for cryptographic applications.

Huirem Bharat Meitei, Manoj Kumar
IoT Based LPG, Smoke, and Alcohol Detection System with Automatic Power Cut-off

Since the dawn of the industrial age, fossil fuels have driven the wheels of human civilization and have directly and indirectly fueled all major technological advancements of the modern age. Liquefied petroleum gas is a primary fossil fuel that is used as a source of energy in industries, automobiles, and for domestic purposes. In spite of it yielding high calorific value, less smoke, and less soot, it is highly flammable and can burn far away from the source of leakage. The risks of severe accidents, fire, and suffocation can be avoided by implementing a smart gas leakage monitoring and notification system. This paper outlines the design of a smart gas leakage detection system that can detect, alert, and regulate gas leaks automatically. As IoT has become one of the most important technologies of the twenty-first century that can easily communicate with other IoT devices over a wireless network. So, we have developed an IoT-based LPG, smoke, and Alcohol detection system that can cut the power supply automatically to ensure more safety and surety.

Dwarakanath Dey, Saikat Datta, Subhojit Datta, Souptik Das, Tanusree Dutta
A Dual-Band Dual-Polarized Coupled Asymmetric T-Shaped Monopole Antenna for Linear and Circular Polarization Applications

A single feed dual-band dual-polarized coupled asymmetric T-shaped microstrip antenna is presented in this paper. The designed antenna involves a coupled asymmetric T-shaped radiating patch that generates mutual coupling to attain wide circular polarized band at higher impedance bandwidth (IBW). A square wide slot with an angle of 45° rotation generates low resonating frequency. In order to maximize the axial ratio (AR), the ground plane and substrate are truncated or cut off from its two opposite right corners. By applying these structures, good circularly polarized performance can be obtained in higher frequency region. The simulated IBW at lower and higher frequency regions is 2.17–2.66 GHz, 20.28% and 7.15–7.78 GHz, 8.43% which are resonating at 2.41 GHz and 7.46 GHz frequencies. The simulated 3 dB axial ratio bandwidth at higher resonating frequency is 7.38–8.3 GHz, 11.73% at center resonating frequency 7.84 GHz. Right-hand circular polarization (RHCP) characteristics are exhibits by using this design. The highest peak gain is 4.31 dBi at 7.4 GHz. The maximum radiation efficiency of this implemented antenna is 95%, and throughout the impedance bandwidth region, the radiation efficiency is greater than 85%. The proposed design in this paper has multiple applications like it can be used for at lower resonating frequency linearly polarized band WiMAX, WLAN application, and higher resonating frequency band can be used for C-band: broadcasting and fixed mobile communication application.

Abinash Kumar Singh, Reshmi Dhara, Sanoj Mahato
New Derivative-Free Families of Four-Parametric with and Without Memory Iterative Methods for Nonlinear Equations

In this paper, we develop new derivative-free four-parametric families of with and without memory iterative methods for determining the roots of nonlinear equations. The family of without memory methods has convergence order eight and supports Kung–Traub’s conjecture. It is then extended to obtain the family of with memory methods using the four parameters as accelerating parameters without the need for extra function evaluations. As such, the convergence order increases from 8 to 15.5156 for the family of with memory methods. Analysis of convergence and numerical experiments are carried out on some nonlinear functions to validate the theoretical results and also to demonstrate the effectiveness and applicability of the proposed families of methods.

G Thangkhenpau, Sunil Panday, Shubham Kumar Mittal

Computer Science and Engineering

Meetei Mayek, Hindi, and English Text Detection from Natural Scene Images Using YOLO

Detecting text in a natural scene image is a crucial computer vision task. This paper presents a multilingual text detection system for detecting Meetei Mayek, Hindi, and English text. The proposed work will serve as a prerequisite for developing a natural scene character recognition system for the mentioned languages. With the evolution of deep learning models, You Only Look Once version 7 (YOLO V7) object detector has been used for detecting text from natural scene images along with language identification. The mean average precision for individual language text detection has been computed. Moreover, the mean average precision with images having more than a single language has also been calculated and the results are appreciable.

Chingakham Neeta Devi, Nella Kartheek, Bokka Purna Manikanta, Motha Yasaswini Saisree, Manjeet
Digit Recognition of Hand Gesture Images in Sign Language Using Convolution Neural Network Classification Algorithm

Sign language is a language used by differently abled persons like deaf and muted people. Hearing impaired people use this language to communicate with normal people. If normal people don’t understand sign language, it is difficult to fill the bridge gap. In this manuscript, a novel approach is developed to recognize the digits of hand gesture images in sign language in order to fill the gap between normal people and deaf, dumb people. In this article, the digit dataset from digit 0 to 9 is considered and taken from Kaggle datasets. The database consists of hand gesture images from 0 to 9 digits, and each digit is having 500 sample images. Convolution neural network algorithm is applied to train the given hand gesture images of database. The evaluation matrix which is considered in the analysis is Recall, F1-score, Precision and Accuracy. More than 90% of accuracy is acquired in the experiment.

M. Navyasri, G. Jaya Suma
Design and Evaluation of Speech Processing Systems for Meetei/Meitei Mayek

The paper focuses on Manipuri, an under-resourced language, and is written in a way that will help eager researchers build a robust speech processing system. To improve the quality of speech processing and language processing systems, the tools and models need to be properly designed and evaluated. Various metrics are considered in the areas of syllabification, speech recognition, speech synthesis, machine translation, etc. Multiple works have demonstrated that authors value accuracy, precision, and scores. Tests are designed to assess data perplexity, naturalness of speech, and intelligibility. Furthermore, the challenges and difficulties associated with the development of various systems are compared. The evaluation metrics, both subjective and objective, are thoroughly discussed, as are various modifications designed to address specific issues arising during data collection, design, and evaluation, mainly in speech recognition and speech synthesis.

Hoomexsun Pangsatabam, Yambem Jina Chanu, Naorem Karline Singh
Congestive Heart Failure Prediction Using Artificial Intelligence

One of the main reasons people for hospitalization of adults over the age of 65 is heart failure (HF) or congestive heart failure (CHF). CHF affects millions of people globally and is among the primary causes of death. Heart failure means the heart is doing less work (pumping less blood) than usual. It may be due to increased pressure in the heart, which gradually leave the heart too weak hence a slower rate of blood flowing through the body. Before it’s too late, heart failure can be prevented by analyzing and controlling the conditions that can cause it. In this work, we use Cardiovascular Health Study (CHS) dataset and compare five different machine learning techniques to predict congestive heart failure (CHF). For feature selection, we employ the decision tree (DT) C4.5 method and the Predictive Mean Matching (PMM) method for missing data imputation. From the different ways applied, our proposed method gives the optimal result.

M. Sheetal Singh, Khelchandra Thongam, Prakash Choudhary
Intelligent Speaker Identification System Under Multi-Variability Speech Conditions

Speech is the natural source of information for human identification in most biometrics, forensics, and access control systems. Mismatch in speech data is one of the biggest challenges preventing speaker identification systems from being employed in real-world scenarios. This research explores how intelligently speaker identification tasks are affected by degraded speech. Our preliminary investigation into mismatch effects in conversational style telephonic speech conditions utilizing the IIT-G database includes mismatch in sensor, environment, language, and conversational style. Convolutional neural networks (CNNs) have surpassed traditional techniques in speaker identification (SI) systems in recent years. This paper proposes a novel architecture based on a VGG-like network for an end-to-end speaker identification system. The proposed architecture outperforms the statistical methods with an improvement of 7% accuracy in identifying the speakers. The results show that the suggested approach is more accurate than state-of-the-art speaker identification techniques and notable performance deterioration compared to the matched scenario.

Banala Saritha, Tungala Thiru Venkata Naga Manoj, Sachin Kumar Sharma, Rabul Hussain Laskar, Madhuchhanda Choudhury, K. Anish Monsley
Artificial Neural Networks and Enhanced Adam Optimization for Effective Wi-Fi Intrusion Detection

In recent years, wireless network expansion has been astounding. The upsurge in the mobility of smartphones and wireless self-contained devices through Wi-Fi network access. These devices have evolved into quintessential electrical gadgets. As wireless networks have gained popularity, they have become increasingly susceptible to attacks. Achieving a high rate of detection, accuracy, and the least false positive result is imperative for network intrusion detection systems. We present a method for training artificial neuron networks with an enhanced Adam optimization (EAO) algorithm to detect intrusions against Wi-Fi networks efficiently. The validation of the method’s performance was done by comparing its results with those of other optimization algorithms and machine learning techniques on a publicly available Aegean Wi-Fi intrusion dataset. With its outstanding performance, this Wi-Fi intrusion detection system, which utilizes artificial neural networks and the enhanced Adam optimizer, offers a superior alternative for safeguarding Wi-Fi networks.

Lenin Narengbam, Shouvik Dey
Computerized Sensing of Diabetes Retinopathy with Fundus Images Using CNN

In the current scenario, artificial intelligence performs a great job to detect and classify numerous diseases, one such being diabetic retinopathy. It helps to detect various problems and figure out different problems. It provides cheaper and better results for the identification as well as screening of retinal disease. There are several eye problems like macular degeneration, glaucoma, cataract, and diabetic retinopathy (DR). DR is one of the broad causes of unordinary visual impairment. But, in recent eras, convolution neural networks or CNNs have given the most excellent execution in image classification in different to conventional or past models. Thus, in this paper, we look at the utilization of convolution neural organize usefulness for the location of diabetic retinopathy with the assistance of colorful fundus pictures from verified data of Kaggle with an exactness of nearly 98%.

Waseem Khan, Khundrakpam Johnson Singh
Intrusion Detection System Using Supervised Machine Learning

The whole world is joining the Internet because our world is moving toward digitalization. Internet/networks are significant today on the planet; information security has turned into a pivotal area of study. As the number of users of the Internet is increasing, it has become a challenge to provide that network security. Nowadays, the improvement of organization security is subsequently featured. Assurance of the network permits the accidental impedance to a structure to arrange and stay away from it. Intrusion detection system (IDS) is one of the most fundamental security devices for the overwhelming majority of security concerns existing in the present digital network world. IDS is developed to inspect the framework applications and organization traffic to uncover dubious exercises and issue an admonition in the event that it is found. To create a more perfect system, countless strategies are accessible in AI for intrusion detection. The main purpose of this paper is to perform anomaly-based intrusion detection systems and apply different machine learning algorithms techniques to the dataset and then compare and estimate their performances. In this paper, we use the KDD’99 cup dataset (KDD cup 1999 data. Retrieved from . Accessed on 7 Nov 2021) and Pearson’s Correlation method to select the important features from the dataset and remove those features that are useless in finding the accuracy. The preprocessed dataset was tried with the models (Decision Tree, Support Vector Machine and Logistic Regression) to get the noticeable outcomes, which prompts expanding the expectation precision. Machine learning methods, namely Decision Tree, Support Vector Machine, Logistic Regression, are used. The examination provides a predictive computational methodology to boost intrusion detection in the Network Traffic Data along with applying different approaches for the appraisal of the best accuracy from machine learning. The aftereffects of various order algorithms analyzed finally by utilizing the proposed dataset have been introduced in the paper for their functionality.

Shubham Kumar, Khundrakpam Johnson Singh
Examining Bioactivity of Medicines in Twenty-First Century Smart Society 4.0: An Approach with ML and DS

Humans and all living beings have different combinations of structural and chemical patterns that act for the smooth functioning of the body. When it does not perform naturally, drugs are needed that are combinations of different chemicals and have a biological function to act upon the target. Drugs can be discovered by finding the target and its chemical pattern, and then, drug detection techniques and machine models using ml have been developed to detect the correct drug for a particular target. The composition of the chemical is important because it has the ability to attach to another compound with the same structure. This created a dataset that was used to construct a machine learning model that predicts and identifies hits by examining the biological activity of various ligands.

Rohit Rastogi, Yash Rastogi, Saurav Kumar Rathaur, Vaibhav Srivastava
A Review on Automatic Assessment and Detection of Pathological Speech

Communication through speech is an essential requirement for social participation. It is a way through which people exchange information, emotions, ideas and feelings. Hence making it the most popular form of communication. Speech processing (SP) research has primarily focused on modelling the activity of voice sources and the activities in the oral cavity. However, SP inherently corresponds to other physiological activities, such like respiration and heart activity, which changes due to several reasons like neurological speech disorder or motor speech disorder (e.g. Dysarthria), mood/emotion, neurodegenerative diseases (e.g. Parkinson’s disease), social setting, environmental setting (e.g. loud/quiet environment). These changes, in turn, can influence speech communication. Changes in pathologies impair the speech construction process, thus producing decreased speech intelligibility and communicative ability. Automatic pathological speech (APS) assessments are indispensable in assisting the clinical diagnosis, treatment and management of speech disorders. In contrast to clinicians’ subjective and time-consuming auditory-perceptual analyses, such automatic assessments provide accurate, objective and practically usable assessments. APS distinguishes between normal and pathological speech and speech intelligibility assessment (SIA) estimating intelligibility as the ratio of the number of words correctly comprehended by human listeners to the total number of words. In this paper, discussion about the techniques used in extracting features of pathological speech (PS) along with the classical machine learning approaches or deep learning approaches used for APS detection. Comparative analysis on various datasets is done for comprehending the anatomy of PS. Brief discussion on APS and SIA will be given. As these are two significant automatic analyses for a possible future tool that can be used for medical purposes.

Ashita Batra, Pradip K. Das
Proposing ML Approach for Detection of Diabetes

The current paper is a work in the field of Machine Learning by the authors. In the past three years, the authors did a lot of work in the field of Machine Learning. In the Literature Survey of the Current Work, we will observe the outcomes found during the research work. Machine Learning is a current trend that has evolved in almost all fields of Engineering and Technology. In the current work, the authors made work in the field of Health Sector. The work comprises the implementation made in Python for detection of diabetes. Experimental dataset is taken from a standard repository available, and using the Python libraries inroads is made to make a detection of disease that is diabetes. In the paper, we will observe several graphs and comparison chart of the various ML approaches and propose the best ML technique for the detection of diabetes. The paper basically aims at finding a technology that is going to be using the current technology to take over the problem of diabetes. This paper is going to provide a way for the researchers to work in the field of computer-oriented generation of applications that suit the Medical Domain.

Vaibhav Kant Singh, Nageshwar Dev Yadav
Comparative and Preventive Analysis of Dictionary Attacks

Hacking is one of the most widespread issues that the general public faces today. Hackers essentially use some social engineering techniques, combined with the publicly available information, to crack open the social media accounts, banking accounts, or cloud storage accounts. As a result, security and protection becomes the main motive for organizations running these platforms. The severity of threats linked to such incidents is tantamount to those linked with data leaks, security breaches, and security lapse events of their likes. Among others, dictionary attacks, a strategy employed for password cracking by trying every common-use word present in an extant wordlist, are the most common. Even though nowadays password strength classifiers are being used extensively almost everywhere which tends to keep these attacks at bay, they are still prone to hyper-personalized dictionary attacks as people are inclined to choose a password which revolves around their personal details itself. Consequently, the necessary awareness for the prevention of dictionary attacks is called for and is provided by performing an analysis of the time taken by different types of dictionary attacks and what type of password or otherwise is safe or not.

Sanat Shourya, Ilayaraja Venkatachalam, Harpal Patel, Manit Mittal
Task Decomposing Optimization in Wireless Sensor Network

Numerous techniques available in the area of wireless sensor network to optimize the energy requirement. Several software and hardware approaches are existing for same purpose. For minimizing the energy, bandwidth and maximizing the security, shortest route is followed by Greedy and many hybrid heuristic algorithms. By proper management of load balancing, a significant life years and performance can extend in WSNs. Load management is one of the other very precise fields in literature. Transport layer computational loading ensures accurate delivery with sufficient aspects of energy and memory in WSNs. Balanced loading is primarily focus rather than equal distribution among all nodes. Load balancing improves the application responsiveness and life time. Complicated task performance is creating a very heavy load in WSN and resulting complex computation, large communication overhead, large amount of information, and high energy requirement. In this paper, we tried to decompose the complex tasks into divisional parts for smooth operation and transmission among nodes in WSNs. This paper considered advance surveillance management as complicated task and tried to decompose in some easy tasks as layering systems. Three layered task subdivision is providing better energy consumption. More sub-layers can also be increased in development of efficient load management.

Arpana Mishra, Rashmi Priyadarshini, R. M. Mehra
LMSF: Lightweight Minimal Scheduling Function for 6TiSCH Networks

Link scheduling is a core component of 6TiSCH networks, specifying which link is allowed to transfer packets at what time and in which channel. For 6TiSCH networks, the IETF recently published Minimal Scheduling Function (MSF). The 6TiSCH network employs the IEEE 802.15.4 Time Slotted Channel Hopping (TSCH) as MAC layer protocol which instructs how to utilize a communication schedule for communication in the network. However, TSCH does not define how the schedule is built and managed. In this regard, MSF provides instructions on managing the communication schedule and adapts the schedule according to the traffic demand. However, MSF updates cell scheduling by unity even when traffic demand changes massively, and it follows a reactive strategy to act upon cell scheduling. As a result, 6P negotiation overhead is high, and it experiences longer delays. This paper proposes to use well-established Poisson distribution for traffic estimation. Then, based on its estimation, the 6P negotiation is adjusted so that a node can directly adapt to high variations in traffic demand. The proposed Lightweight Minimal Scheduling Function (LMSF) is evaluated with the 6TiSCH simulator. LMSF significantly reduces 6P negotiation overhead. Also, it consumes less energy and shows better latency and reliability performances than the standard MSF.

Karnish N. A. Tapadar, Priyanshu Singh, Manas Khatua
An Efficient Content-Based Image Retrieval Using Threefold Technique

Managing large amounts of information in the digital world requires an efficient and effective image retrieval system. Content-based image retrieval (CBIR) has garnered tremendous attention and effort over the past few years. Several CBIR systems have been proposed using both onefold and twofold approaches. This paper introduces a new threefold content-based image retrieval system (TfCBIR), which is built on a threefold approach and consists of two modules. As part of its first module, TfCBIR analyzes images to extract color, texture, and shape information. The second module includes three steps. The first step involves comparing the color feature space of all the images with the query image to find the most similar $$S$$ S images. In the next step, the shape feature space is compared with the image query for the $$S$$ S images obtained in the first step to retrieve the closest $$S1$$ S 1 images. Finally, in the third step, the texture feature space is compared with the query image for the $$S1$$ S 1 images found in the second step, and the $$R$$ R most relevant images to the query image are obtained as output. The proposed TfCBIR system has been shown to outperform existing state-of-the-art CBIR methods.

Nepoleon Keisham, Arambam Neelima
A Study of Various Audio Augmentation Methods and Their Impact on Automatic Speech Recognition

Automatic speech recognition transforms a spoken utterance into its corresponding textual form. The scarcity of annotated speech data hinders the development of any language’s ASR system. The quantity and quality of training data that are provided have a direct correlation with how well these systems perform. This is further excruciating for regional low-resource languages. The development of a credible speech corpus for such a language is time-consuming and expensive. To alleviate this problem to some extent, data augmentation is generally employed to enhance the generalization ability of the model. This paper examine several currently used techniques, ranging from raw data augmentation to frequency domain augmentation, such as time masking, peak normalization, and frequency masking. We analyze and contrast their benefits and drawbacks in terms of a few key factors such as execution time and computational accuracy. We also look at the applicability of a few techniques that have been shown to improve speech recognition through audio data augmentation.

Naorem Karline Singh, Yambem Jina Chanu, Hoomexsun Pangsatabam
A Hybrid Federated Reinforcement Learning Approach for Networked Robots

Federated learning (FL) evolved as a game changer to ensure the privacy of user-sensitive data, by using locally available models, rather than data, for aggregation at a central server. While in a centralized FL approach, network connectivity or server failures could bring down the system, its decentralized counterpart can circumvent this issue, but at the cost of increased learning times. With FL being used in the domain of networked robotics, a combination of centralized and decentralized approaches can prove to be a safer and more viable option. In this paper, we present a mobile agent-based Hybrid version of Federated Reinforcement Learning (HyFRL), where the learned models, viz. Q-tables, are aggregated and shared among a set of connected robots inhabiting different environments. Multi-robot experiments performed using several networked instantiations of Webots®, an open-source robot simulator, reveal the efficacy of this hybrid version over its centralized and decentralized equivalents.

Gayathri Rangu, Divya D. Kulkarni, Jayprakash S. Nair, Shivashankar B. Nair
Proposing ML Approach for Detection of Lung Cancer

In this paper, we will observe the usage of machine learning algorithms for the detection of lung cancer. In a country like India and especially in a city like Bilaspur or State like Chhattisgarh, the condition to take care of disease like lung cancer is very poor. The people who get this disease have to go to big cities for their treatment and in state like Chhattisgarh people are also not very rich. If the disease is detected at a concluding stage, the chance of saving the person is very less. We in the current work taking a sample experimental dataset from a standard repository the description of the fields and other factors of which are described in the paper have made an implementation in Python for detection of the disease. We, in the paper, made consideration of various algorithms on the dataset and got different values for the different assessment parameters. Based on the comparative analysis and looking to the utility that ML algorithm that turns out to be the best, we propose the usage of ML for detection of lung cancer. The literature survey part comprises of various work done in the field and data regarding the research in the field. In the paper, we are trying to propose ML approach and the current computing technology available to handle the problem of lung cancer. Generally, we are more concerned for the people of our state where medication facility is still not that good and early detection could help in saving life. In the current work, we made a utilization of Python which is a great data analysis language. Currently, people around the universe are working in this tool to have their work done in the desired way. In the current paper, we will look into the implementation of several machine learning algorithms. The algorithms that we observe in the implementation have proven their feats in the past and that’s why had been considered in the current paper. We will also have a look on the lung cancer paradigm which is the subject of implementation in the current paper. The paper is nicely drafted and is having introduction, problem statement, literature review, implementation, and result and conclusion as the main sections of consideration. The paper covers a wide spectrum of papers in which it made a good detailed survey of the various computer generated work for the solution of problems of various types. The paper is altogether a very nice work that could be handy for the research that is going to be put up in the time to come. We will observe a good detailed work on lung cancer and to continue in the journey will see a lot on Python and the various machine learning algorithms.

Vaibhav Kant Singh, Nageshwar Dev Yadav
Predicting the Heart Attacks Risk Using Artificial Neural Networks

Coronary illness is a dangerous infection that enormous populace of individuals around the globe experiences. When considering demise rates and enormous number of individuals who experiences coronary illness, it is uncovered how significant early analysis of coronary illness. Customary method for conclusion isn’t adequate for such an ailment. Building up a medicinal analysis framework dependent on AI for forecast of coronary illness gives more precise finding than conventional way. In this paper, a coronary illness expectation framework which uses artificial neural system backpropagation calculation is proposed. Thirteen medical highlights were applied as input for the neural system with cross validation-10 got the accuracy of 80%.

Rayi Naveen Kumar, Mullapudi Navyasri
A Printed Character Recognition System for Meetei-Mayek Script Using Transfer Learning

This paper presents a printed character recognition system for Meetei-Mayek script of Manipuri language. The input to the system is downloaded and cropped images containing printed characters of Meetei-Mayek. The images go through pre-processing which involves thresholding and Morphological Transformations. Segmentation of characters has been carried out using image processing techniques like dilation, thinning and text properties like height, width, and aspect ratio. After segmentation, the images are converted into three channeled black and white images. The work comprises development of standard database for printed characters for Meetei-Mayek. The classification task has been carried out using transfer learning using pre-trained VGG-16, VGG-19, and ResNet152-V2 convolutional neural networks (CNNs). The recognition results have been compelling, and VGG-16 has achieved better classification accuracy in comparison with other pre-trained CNN—VGG-19, ResNet152-V2. The classification accuracy obtained by VGG-16 is 99.27%.

Vishwaksena Vishnu Simha Dingari, Ganapathi Kosanam, Devi Sri Shankar Chavatapalli, Chingakham Neeta Devi
An Efficient Intrusion Detection System Using Feature Selection and Long Short-Term Memory (LSTM)

As the number of services available over the Internet increases, the network infrastructure becomes more vulnerable to malicious cyber threats. To address such issues, various techniques have been adopted in order to mitigate such malicious activity in the network. A hardware or software tool called an intrusion detection system (IDS) can watch a network for fraudulent attacks that could cause the system to malfunction. A number of deep learning and machine learning methods are being used in order to demonstrate their effectiveness in detecting such attacks. In this study, a feature selection technique and a deep learning approach based upon long short-term memory (LSTM) classifier are used to detect attacks. The proposed system was implemented on a benchmark dataset, the CICDDoS2019, and it was observed that the model achieves the best accuracy while training and testing. Additionally, a comparison with other related works shows that the model is more effective when encountering attacks.

Hidangmayum Satyajeet Sharma, Khundrakpam Johnson Singh
A Review on Speech Biomarkers for Obstructive Sleep Apnea(OSA)

Obstructive sleep apnea is well known as OSA. It is a breathing disorder that happens when we sleep, as it blocks significant parts of the upper airway(UA). The person becomes breathless or choked, resulting in loud snoring and dizziness throughout the day, even after proper sleep. If proper attention is not given, it may lead to high blood pressure(BP), heart strokes, etc. Mostly affecting people who are obese. Sleep apnea can be detected by analyzing the person’s snoring pattern, which helps diagnose the disease’s severity. Studies have been conducted on various databases which are publicly available in order to make a comparative analysis amongst various machine learning algorithms, iVectors, or supervectors. Along with the various feature extraction techniques which are already exploited by the various state-of-the-art algorithms. In this paper, we discussed several classical machine learning algorithms used for detecting OSA will be comprehensively reviewed whilst discussing the constraints that arise when machine learning algorithms are exploited. Since the emergence of wearable technology, alternative methods of diagnosing OSA have been investigated, including home sleep tests.

Himanshu Sharma, Pradip K. Das
Real-Time Object Detection for Unmanned Underwater Vehicles Using Movidius Neural Compute Stick

Deep learning-based object detection methods have demonstrated promising results in controlled environments. Due to specific limitations, these methods are insufficient for unstructured environments, such as underwater exploration or espionage using unmanned underwater vehicles (UUV). Poor image quality is the primary issue limiting UUV’s functionality. Underwater objects are typically small and blurry as a result of severe noise that confuses the detectors. This study presents a graph convolution neural network (GCNN)-based real-time object tracking system in an unstructured environment. On the algorithm side, object’s regions are categorized based on the pixel population density. Resultantly, the regions are disintegrated recursively into an image patch tree. Moreover, the surfaces are subsequently and proficiently stored, later retrieved as patches. To keep the power consumption low, a PYNQ Z2 FPGA is used as development platform in connection with a Movidius NCS stick for the outsourcing of the GCNN. The proposed technique is generic and reducing high computational complexity and power consumption.

K. Amal Thomas, Soumyajit Poddar, Mourina Ghosh, Amitava Nag
Dysarthric Speech Characterization and Classification Based on Affinity Propagation

Millions of people are estimated to suffer from speech impairments originating from various causes, including neurological disorders, brain damage, and physical conditions. Dysarthric speech is hard to recognize and understand because of its enormous variability and low intelligibility rate. In today’s high-tech environment, speech recognition system has achieved more than 95% accuracy, from which dysarthric speakers are far behind and excluded from the state of art escapade. In this paper, we propose a model to narrow down this gap and help them to get recognized by deep feature analysis and informative feature extraction techniques to get a better characterization of dysarthric speech. Linear Prediction Cepstral Coefficient (LPCC) is used as a feature extraction technique and a pre-clustering-based algorithm, Affinity Propagation (AP), is used to select the best features for a speaker based on the locality of the speech signal. SVM is used to verify the selection of features. UASpeech digit dataset is used for the experiment, and the result is consistent. A significant amount of data is required to build a conventional speech recognition system, but some fields, like medically challenged speech recognition, lack sufficient data, that’s why some optimized or alternate techniques are used.

Komal Bharti, Sandeep Agri, Pradip K. Das
Securing Secret Information

The art of information concealment has become a crucial subject since information security has grown to be a primary concern in the online world. Using cryptography and steganography plays a significant role in secure data transfer. Steganography is a covert writing technique; the message is contained behind a cover medium. Cryptography encrypts messages to hide their contents. Two-dimensional bar codes called QR codes can store text strings. They can encode data both vertically and horizontally, encoding more information. By combining the ideas of cryptography, steganography, and QR codes, an innovative method for covert communication is proposed in this paper. Statistical and security analyses of the proposed method show that the proposed method provides better imperceptibility, security, and robustness.

Zeba Shamsi, Laiphrakpam Dolendro Singh
A Comprehensive Study of DDoS Attack on Internet of Things Network

In relation to each IoT layer, this paper examines DoS/DDoS attacks as well as security measures. It demonstrates how different vulnerabilities in every tier can be used by attackers. The IoT network is made more secure by the discussion of network security potential solutions. We must address security challenges for all of the various layers, not just one, if we want a solid and safe organisation. In other words, merely protecting the application layer won't stop hackers from accessing the network layer. Perception layer devices are flexible and simple to use, which helps to save expenses. Because of this, the perception layer is the one that is most susceptible, and identifying skills requires extensive investigation. Even though there are several DoS/DDoS preventive mechanisms identified in the literature, they still require a lot of study and development. IoT applications have caused an industry-wide dynamic transformation. In order to create a unified solution for a variety of scenarios including heterogeneous devices, networks, and protocols, and it is imperative to leverage technologies like Machine Learning and Artificial Intelligence. Additionally, users of the applications need to understand the significance of utilising secure credentials and passwords as well as regular software updates. In this paper we have analysed various datasets available for intrusion detection along with parameters like detail classification of Data Pre-processing methods. The most often utilised metrics are precision, recall, F1 value, accuracy as well as false-alarm rate (FAR).

Nitin Anand, Khundrakpam Johnson Singh
A Matrix Factorization Algorithm for Movie Recommendation

In this paper, we build various movie recommender system ranging from simple recommender systems to hybrid recommender system. We used content as well as a collaborative filtering technique. To establish the similarity relation between users or movies we have used cosine similarity measures. We analyse the used of user-user similarity matrix as well as a movie-movie matrix in building our recommender system which plays a very crucial role in determining the accuracy rate and computational time consumption. In content-based recommender system, for deriving similarities between movies from their description corpus we have used the Term frequency-Inverse document frequency measures. Next we move on to build collaborative filtering using various models like K-nearest neighbours, K-nearest neighbours with mean, Singular value decomposition and Singular value decomposition ++ and compare them in terms of their accuracy in Root Mean Square Error with the least time consumption. Prior to all, we utilize the Internet Movie Database weighted rating formula to build a simple recommender system.

Disinlung Kamei, Khundrakpam Johnson Singh
Recent Advances in Electrical and Electronic Engineering
Bibhu Prasad Swain
Uday Shanker Dixit
Copyright Year
Springer Nature Singapore
Electronic ISBN
Print ISBN