International Journal of Engineering Systems Modelling and Simulation (58 papers in press)
Mathematical Modelling for Prediction of Mechanical properties of Abutilon Indicum fibre reinforced composite using RSM and ANN
by Doni Mohana Krishnudu, D. Sreeramulu, P. Venkateshwar Reddy
Abstract: Major application areas of natural fiber composites are found in packing industries. The focus of the present study is on natural fibre (abutilon indicum) reinforcement composite materials with filler and to predict their mechanical properties of Abutilon Indicum fibre reinforced composites. In the present study fiber content (wt.%) and filler content (wt.%) are taken as the influencing parameters and mechanical properties like tensile, impact and flexural strengths are taken as an output parameter. Various proportions of fiber content and filler content were designed as per L25 orthogonal array and their mechanical properties were found. A regression equation was developed using RSM for each mechanical property in terms of fiber weight and filler weight. Mathematical models are developed by using both RSM and ANN and predictive capability of these two techniques were compared based on its R2 values. From the R2 values it is confirmed that the predictive capability of ANN model is higher than the RSM model for the present study.
Keywords: Abutilon Indicum; Mechanical Properties; RSM; ANN.
Finite-element modelling of magnetic fields for superconducting magnets with magnetic vector and total scalar potentials using COMSOL Multiphysics
by Alexander Chervyakov
Abstract: The finite-element modelling of superconducting magnets is a resource-hungry and
challenging work. For these devices, the high-quality requirements for focusing fields are usually superimposed by complexity of the model geometries and nonlinearity of the magnetic materials. The precise field simulations could result therefore in substantial number of the degrees of freedom and, as a consequence, in significant usage of the computational resources. To achieve the acceptable accuracy with lower number of finite elements, the magnetic field distributions are calculated in terms of the magnetic vector potential (A-formulation) as well as the total scalar potential (V-formulation) with COMSOL Multiphysics and compared. For these calculations, we utilise the model of a superconducting dipole magnet recently designed for operation of the compact isochronous cyclotron SC200. The performance of the both methods is analysed in terms of accuracy and quality of the obtained fields as well as in terms of the computational cost.
Keywords: compact cyclotrons; superconducting magnets; magnetic fields; finite-element method.
Novel deep learning model for bitcoin price prediction by multiplicative LSTM with attention mechanism and technical indicators
by S. Aarif Ahamed, Chandrasekar Ravi
Abstract: Starting from the earlier civilisation to till date money plays a crucial part in the transfer of goods and services. With this digital world, the money also changes its faces from paper money to digital currency called cryptocurrency without any central bank which runs on top of the technology called blockchain. The trendiest cryptocurrency is bitcoin. Forecasting the daily price is a challenging task due to its nonlinearity. Most of the researchers tried to predict using various statistical and machine learning models which were not satisfactory because of its large dataset with more noise. The intention is to design a deep learning multiplicative long short-term memory model to estimate the price of bitcoin with an attention mechanism using technical indicators which gives better accuracy and a very less error rate. The proposed model is compared with some existing models say long short-term memory, peephole, gated recurrent unit and multiplicative long short-term memory on the presence and absence of technical indicators. The comparative result shows that the proposed model outperforms the existing models in terms of mean square error, root mean square error and mean absolute error when evaluated with two benchmark datasets.
Keywords: cryptocurrency; bitcoin price prediction; bitcoin; deep learning; technical indicator; attention mechanism; multiplicative LSTM.
Survey on personalised recommendation system development in big data analytics
by T.P. Ezhilarasi, K. Sashi Rekha
Abstract: In recent times, recommendation systems (RSs) are extensively used by diverse companies to raise offerings and profits with the target of providing specialised services to users. To meet the needs of the project, the objective is to provide a good idea of the RS in place for the support. To achieve this objective, this work provides a systematic survey of a personalised RS (PRS) based on the agricultural context for analysing the data. The results demonstrate that a combination of recommendations and data analysis is not extremely used for general practices. However, validation is regarded as infrequent in agriculture. From the investigators perspective, this review provides for an in-depth analysis of learning approaches and data analysis based on the preliminary need for agriculture needs. This leads to an extremely complex solution for identifying an effective solution that assists real agriculture for earlier predictions. Therefore, in this work, the author suggests the use of data analysis and PRS to facilitate implementation in various areas.
Keywords: personalised recommendation systems development; agriculture; big data analytics; BDA; systematic reviews; the optimal solution.
Solar power prediction in real time utilising supervised machine learning algorithms considering Madhya Pradesh region
by Sanjiv Kumar Jain, Kaustubh Yawalkar, Prakhar Singh, Advait Apte
Abstract: Solar energy holds the key for future electric power generation providing sustainable development to meet the requirement for long lasting demand in the future. As exhaustible sources are depleting constantly, we need to focus on non-exhaustible sources. The prediction of solar power output is critical to plan for the future operations and integrating the power grid with renewable sources. In this study, a dataset of climatic parameters that is, beam (direct) irradiance, diffuse irradiance, reflected irradiance, sun height, air temperature and wind speed for five years is used to predict the solar power output based on photovoltaic technology of Indore region (27.2046 N, 77.4977 E). The performance of the trained models is determined using statistical (mathematical) indicators. Amongst the machine learning algorithms, the best accuracy of 98% is achieved by random forest method for the prediction of solar power output for Indore region.
Keywords: machine learning; neural network; Poisson regression; random forest; solar power output; supervised learning.
Interactive search algorithm of artificial intelligence for household classification on smart electricity meter data
by M. Suresh, M.S. Anbarasi
Abstract: Smart grid (SG) is a future-generation power system commonly used to maintain electricity demand in a reliable and economic way using the latest information and communication technologies. It enables consumers and micro-energy producers to take a more active role in the electricity market and dynamic energy management (DEM).To maximise the accuracy of dynamic energy detection, research work designs an enhanced swarm-based big data analytics model for DEM in SG. In this paper, an artificial neural network (ANN)-based on classification model for predicting future power consumption is proposed. A novel bio-inspired optimisation namely interactive search algorithm (ISA) is used for optimising the weights of ANN.The results of the proposed model are compared with different performance measures to prove its efficiency. A detailed comparative results analysis takes place and the experimental results ensured the betterment of the proposed models over the state of art techniques. The compared results show the significance of the proposed model over existing algorithms.
Keywords: interactive search algorithm; ISA; classification; smart meter; artificial intelligence; IoT.
Detection of impaired objects in roadways using metaheuristic algorithms
by Sambandam Ramachandran Balaji, Karthikeyan Santhanakrishnan, Manikandan Radhakrishnan, Albert Mayan John
Abstract: Roads have become the most fundamental element in land transportation system. In the long run, some malformations will appear on the road, such as potholes and cracks. Since manual inspection is unpredictable, subjective and prolonged, we go for computer vision-based methods. Thus our work focused on the automatic detection of the cracks and potholes. For the detection process, we acquire the video, convert them into frames and use metaheuristic algorithms to implement detection of the roadway damages (i.e., cracks and potholes). The novelty of this approach lies in using texture-based features to differentiate between crack surfaces and intact roads. Three different metaheuristic algorithms are used to detect the crack and potholes. The performance of the algorithms is evaluated using the different parameters. Based on the performance, it is observed that grasshopper optimisation algorithm outperforms well for this application.
Keywords: crack detection; pothole detection; metaheuristic algorithm; particle swarm optimisation; PSO; whale optimisation; grasshopper optimisation.
Data protection and threat prediction in industrial IoT cloud environment
by Satish S. Salunkhe, Aditya Tandon, J. Thimmia Raja, S. Narayanan, Varsha ., D. Ramkumar
Abstract: There are several cloud computing data protection systems depending on accessibility control, attribute-based encryption (ABE), security and reputations, but still, it is disjointed and lacks a cohesive rationale. I-IoT systems including data protection and threat security metrics are discussed in this paper. I-IoT layered method that is general and extended, with privacy, security features and layers detection. To protect the users private data, security procedures and crucial management events are established amongst all the lower, moderate and higher levels. To permit data flow across the levels of the proposed cloud computed powered I-IoT paradigm, we built security certificates. The proposed systematic method eliminates potential security vulnerabilities and is also utilised in conjunction with the optimal security approaches to mitigate the security threats on cloud computed decentralised industrial IoT systems and devices. The assaults toward I-IoT systems are described in this paper, and a detailed examination of remedies to these attacks.
Keywords: data protection; industrial internet of things; I-IoT; cloud computing; end-to-end encryption; security protocols; threat protection.
Portable healthcare computing and clinical decision support system enabled by artificial intelligence
by Satish S. Salunkhe, Vinodkumar Jacob, Aditya Tandon, S. Jeevitha, Rakesh Kumar Arora, Shilpa Laddha
Abstract: In this work, we present a clinical decision support system enabled by the AI methodology including deep neural network (DNN) in association with IoT cloud for the forecasting healthcare of the patient and is analysed by considering chronic kidney disease (CKD) to provide the optimum healthcare services to consumers with its severity level utilising e-health apps. The proposed system gathers patient details from their IoT devices and their healthcare records from the UCI repository are being saved in the cloud aided in the customisation of health therapies for particular populations. In addition, the identification and severity of the disease are performed using a deep neural network (DNN) classifier. A particle swarm optimisation (PSO)-based feature extraction is utilised to increase the effectiveness of the DNN classifier. The standard CKD dataset is used to verify the proposed approach. The proposed approach is evaluated in terms of accuracy, sensitivity, F-score and specificity.
Keywords: artificial intelligence; clinical decision support system; CDSS; deep neural network; DNN; internet of things; IoT; particle swarm optimisation; PSO.
An energy efficient NOMA-based spectrum sharing techniques for cell-free massive MIMO
by Ch. Gangadhar, K. Chanthirasekaran, K. Ramesh Chandra, Aditi Sharma, M. Thangamani, Ponnusamy Siva Kumar
Abstract: In this paper, a hybrid spectrum optimisation strategy is introduced to enhance resource sharing capability concerning variable sum rate and minimum interference ratio of the secondary users (SUs) for cell-free massive multiple input multiple output (CF-MIMO) system and induce energy using non-orthogonal multiple access (NOMA) resource allocation-based algorithm. The proposed optimistic clustering method includes the secondary users (SUs) are placed in different radical coverage with fixed distance forming irregular cluster group size depend upon their interference degree and converges dynamic stability. The NOMA-RA algorithm enhances the variable sum rate by assigning different cluster groups, conquering with suitable sub-channels. The simulation parameters are normalised interference ratio, pilot insertion, received signal strength (RSS), successive interference cancellation (SIC) and uplink and downlink transmission power. As results, shows the significant improvement is achieved by available amount of SUs in terms of variable sum rate and minimum interference ratio as compared with CF-MIMO and NOMA, respectively.
Keywords: hybrid optimum spectrum sharing; CF-MIMO; cell-free; CF; NOMA-RA-based CF-MIMO.
Deep learning-based video coding optimisation of H.265
by C. Karthikeyan, Tammineedi Venkata Satya Vivek, S. Lakshmi Narayanan, S. Markkandan, D. Vijendra Babu, Shilpa Laddha
Abstract: Todays multi-media applications need high video quality with low bitrates. However, it is restricted in its capacity to provide higher quality than earlier coding methods. Deep learning (DL) approaches for video coding have shown compression capacities equal to or better than traditional methods, including high-efficiency video coding (HEVC) methods. The trade-off between compression efficiency and encoding/decoding complexity, optimisation for perceptual nature of semantic dependability, specialisation, and universality, the federalised layout of various deep toolkits, etc. remains unclear. HEVC encoding is more efficient than previous standards. Improved efficiency is driven by intra image prediction, which incorporates more prior directions (35 modes) than previous standards. Its high efficiency comes from balancing encoder complexity and dependability. Disadvantage #1: inclusion distortion rate (RD). This article presents DL, which uses a convolutional neural network to predict the best model with the least rate-distortion (RD) and further promotes study into deep learning video coding (DLVC).
Keywords: deep learning video coding; DLVC; high-efficiency video coding; HEVC/H.264; rate-distortion; rate-distortion optimisation; RDO.
A data transmission approach with energy reduction based on virtual machine migration technique in cloud computing
by H. Anwar Basha, R. Saravanakumar, K. Prabu, Divyendu Kumar Mishra, S. Narayanan, A. Samydurai
Abstract: To provide quicker data access, database centres use virtual machines (VMs) migration to maintain regular content pages in the necessary unit. Memory sharing without downtime is ideal for offline VM migration. However, it has several problems while migrating active VMs. To improve bandwidth availability and hardware stability, it is utilised in workload balancing, low energy retains, and dynamic VM resising. Thus, needless memory (dirty pages) moving leads in lengthy migration time and downtime. To minimise energy usage and the number of VM migration stages, we offer the NPA-FLI-EC. It combines neural prior prediction algorithm and fuzzy logic insertion of energy reduction on VMM method. Using NPA-FLI-EC, it may optimise VM placement and minimise connection loss on physical servers, while anticipating resource identification from each host reduces needless VM migrations. Thus, it allows for task diversification over multiple servers while saving 2/3 of total energy usage. It also saves bandwidth and improves energy efficiency by consolidating the number of VMs.
Keywords: virtual machine; VM migration; data sharing; evolutionary computing.
Software product line regression testing based on fuzzy clustering approach using distance method
by Ashish Saini, Raj Kumar, Gaurav Kumar, Satendra Kumar, Mohit Mittal
Abstract: Testing is a process that takes much time and effort in software companies. This becomes even more difficult and boring when it comes to testing a software product line (SPL). The SPL is a model in which multiple products from the same family are made simultaneously. Testing of all products is not possible. Hence a lot of testing methods have been given from time to time to test the product line, given by researchers based on contemporary conception. In the direction of testing product lines, this article has proposed a method, which used fuzzy C-means clustering with the Jaro-Winkler distance method. Variable features of the product form the basis for cluster development. The proposed method compared with other distance methodologies. After comparison, it concluded that the proposed method provides better results than other methods. This article has resorted to some product lines to compare the proposed methods.
Keywords: product line; software product line testing; fuzzy C-means; FCM; feature model; testing; software industries.
Visual exploration of fault detection using machine learning and image processing
by D. Vijendra Babu, K. Jyothi, Divyendu Kumar Mishra, Atul Kumar Dwivedi, E. Fantin Irudaya Raj, Shilpa Laddha
Abstract: The machine learning CNN method defect detection is highly reliant on the training data; thus, post-classification regularisation may significantly improve the output. The suggested fault detection process may perform well on demanding synthetic and actual information by using a practical synthetic fault system depending on the SEAM model. We further propose the visual exploration be made more reliable regarding fault tolerance. The visual exploration model is made up of three-phase namely visual identification and mapping, dynamic controller, and terminate criterion. The submap-dependent on visual mapping phase ensures higher mapping manageability, semantic classification dependent on active controller ensures continuous driving, and a new completion assessment technique ensures robust re-localisation under the terminate criterion. To preserve mapping and improve visual tracking, all the components are tightly linked. The proposed model machine learning CNN model is examined, and actual tests show fault-tolerance methods are proven to withstand visual monitoring and mapping failure situations.
Keywords: visual exploration; fault detection; convolutional neural networks; CNNs; image processing.
Analysis on road crash severity of drivers using machine learning techniques
by Mohit Mittal, Swadha Gupta, Shaifali Chauhan, Lalit Kumar Saraswat
Abstract: Traffic accidents are significant general well-being concerns, bringing a large number of deaths and injuries around the globe. To improve driving safety, the examination of traffic data is basic to discover factors that are firmly identified with lethal mishaps. In this paper, our main objective to evaluate the severity based on various factor to reduce the road accidents and enhance the safety. Therefore, a long range of factors are considered to evaluate severity into two type either fatal severity or non-fatal severity. Out of all factor, we have evaluated the top ten features that are most important with the help of CART, random forest and XGBoost algorithm. For prediction of severity, we have considered the logistic regression, ridge regression and support vector machine regression. The experimental results show that fatal severity is higher for fog weather condition, heavy vehicles such as truck, male drivers and old age drivers.
Keywords: injury severity; collision data; fatal accidents; machine learning.
A decision tree C4.5-based voltage security events classifier for electric power systems
by Sanjiv Kumar Jain, Shweta Agrawal, Prashant Kumar Shukla, Piyush Kumar Shukla, Anurag Jain
Abstract: Static voltage security classification has emerged as a potential field of research, due to large interconnections and more power demand. The paper presents a model to deal with static voltage security assessment problem through machine learning algorithm and decision tree C4.5. Using this algorithm, security classifications of power system operating states is achieved under vast load variations. N - 1 line outages contingencies are considered for the knowledge-base generation using the offline continuation power flow method. Mainly, the credible contingency cases are considered for security classification. The proposed approach is tested on IEEE-30 bus and IEEE-118 bus systems. The work will be useful for system operators in control decisions and prevent the occurrence of grid failure. Percentage classification accuracy of 100% is achieved for line outage nos. 8, 12 and 13 for IEEE-30 bus system and the accuracy is 98% for line outages no. 93 for IEEE-118 bus test system.
Keywords: artificial intelligence; decision tree; machine learning; power systems; voltage security.
Design and analysis of MIMO antenna for IOT applications
by R. Nagendra, S. Swarnalatha
Abstract: In this paper, a novel MIMO antenna of compact split ring resonator (SRR) type with high degree of isolation among four ports on the antenna has been proposed for MIMO multiband applications. It is a triband which operate at frequencies 1.44 GHz (-11 dB), 2.3 GHz (-10 dB) and 4.2 GHz (-245 dB), respectively. This proposed MIMO antenna is designed with meander line for better gain and split ring resonator (SRR) in the form of ring for high isolation between the antenna elements. A loaded stub is used to achieve the resonant frequency. The proposed MIMO antenna was simulated with IE3D EM software by mentor graphics and the features such as impedance view, lower return loss, gain, radiation pattern, VSWR, etc. are measured. The proposed antenna has better gain and provides higher isolation between multiple antenna elements.
Keywords: meander line; multiple input multiple output; MIMO; split ring resonator; SRR; IOT applications.
Rainfall prediction using ensembled-LSTM and dense networks
by Ujjwal Sinha, Vishal Thakur, Sammed Jain, M. Parimala, S. Kaspar
Abstract: Rainfall prediction has been of utmost importance in any country. The amount of rainfall in a particular region has been known to affect the growth in that area, especially in an agriculture-based country like India. This paper proposes a model which performs one step rainfall forecasting in the regions Ranakpur and North-Eastern states of Assam and Meghalaya based on time series data acquired from 1 and 75 weather stations in both areas, respectively. This model was chosen to be based on the LSTM algorithm which has proven to be better than existing rainfall prediction models based on linear regression, support vector regressors, artificial neural network, random forest and decision tree algorithms. The RMSE score of the proposed architecture for Ranakpur and North-East were 1.948 and 2.654 respectively, better than the algorithms used in comparison. The factors taken into consideration for while predicting the weather are - max temperature, min temperature, precipitation, wind speed, relative humidity and solar radiation.
Keywords: rainfall prediction; long short-term memory; LSTM; forecasting; weather; Root mean squared error; RMSE; precipitation; humidity; wind speed; time series.
Crowd management in public transport to ensure social distancing for prevention of spread of COVID-19
by K. Gerard Joe Nigel, J. Jenisha, R. Rajeswari, D. Pamela, P. Manimegalai
Abstract: Global shock from COVID-19 epidemic. Social isolation is becoming more crucial as this delicate condition spreads swiftly. Public transit must be enhanced to stop corona spread. An IoT method using LoRa technology might reduce overcrowding and disease transmission in public buses. Buses have LoRa transmitters and receivers. It is shown on the bus stop's LCD screen and announced over a speaker if the bus is within range of the receiver. An automatic door mechanism limits the number of people inside the vehicle. In the mobile app, the bus occupancy data is sent to Google Firebase. The app also indicates nearby buses, their occupancy, and their estimated arrival time. In certain cases, authorities may utilise this data to analyse and act. This simple technique would improve bus safety and contain COVID-19.
Keywords: COVID-19; ESP32; crowd management; bus management; IoT.
Research on seismic response of long-span continuous rigid frame bridge in degraded permafrost regions
by Jianning Li, Xuexian Sun, Ziqi Li, Wei Lu
Abstract: To study the influence of permafrost degradation on seismic response of bridge, a large-span continuous rigid frame bridge in warm permafrost regions of Qinghai-Tibet Plateau was used as research object, two permafrost degradation modes have been studied, Results showed that permafrost temperature has less influence on seismic response of rigid-framed pier but more obvious on movable pier under the overall warming degradation mode. However, the maximum bending moment of pile decreases and its position move down with temperature rise, while along transverse direction, the position all at top of piles. When under top-bottom degradation mode, only movable pier are significant affected in longitudinal direction. The maximum bending moment of pile increases and its position move down with degradation depth. In general, the adverse effects caused by permafrost degradation should be considered in piles, but for piers just consider it to a certain extenet.
Keywords: bridge engineering; permafrost degradation; seismic analysis; overall warming degradation; top-down degradation; seismic response.
Augmentation of predictive competence of non-small cell lung cancer datasets through feature pre-processing techniques
by M. Sumalatha, Latha Parthiban
Abstract: Non-small cell lung cancer (NSCLC) comprised of complex hidden and unknown data that is challenging for prediction at the earlier stage. The major objective of the research paper is to develop a novel preprocessing model based on minimisation of features and competency maximisation through feature pre-processing (FPP) to provide augmentation in predictive competence of NSCLC datasets. In Phase-I, the test for relevancy identified behavioural errors like null, empty and NAN values to reduce two features. In Phase-II, regression analysis was performed to find the relationship between features after which four features were removed. In Phase-III, cluster analysis is carried out to find the irrelevant features in the form of clusters and seven features are removed. The competency of NSCLC dataset before FPP showed more accuracy than after FPP with classifiers like simple tree, complex tree, linear SVM, Gaussian SVM, weighted KNN and boosted tree classifiers.
Keywords: non-small cell lung cancer; NSCLC; competency of prediction; relevancy analysis; regression analysis; cluster analysis; feature pre-processing model; feature pre-processing; FPP; competency analytics.
Rice plant diseases detection using convolutional neural networks
by Manoj Agrawal, Shweta Agrawal
Abstract: Rice is one of the main crops grown in India and it is complicated for farmers to accurately classify rice diseases manually with their imperfect information. Thus, the automatic recognition of rice plant diseases is highly desired. Many methods are available and have been proposed for the rice plant diseases detection. The latest advances indicate that the use of CNN models can be very beneficial in such troubles. In this paper we have explored and trained various CNN models with the unique combinations of training and learning methods to enhance the accuracy. The most advanced large-scale architecture, such as VGG19, XceptionNet, ResNet50, DenseNet, SqueezeNet, and CNN are implemented with the baseline and transfer learning methods. These models are trained and tested on data sets collected from various sources. Experimental results show that the ResNet50 architecture achieved the highest accuracy of 97.5% as compared to other CNN architectures and existing literature.
Keywords: convolutional neural network; CNN; deep learning; base learning and transfer learning; rice leaf diseases; VGG19; XceptionNet; ResNet50; DenseNet; SqueezeNet.
A triple band MIMO slot antenna with enhanced resonance for WiMAX, Wi-Fi and WLAN applications
by Beulah Jackson, P. Pattunnarajam, S. Asha, M. Bindhu, J.C. Elizabeth
Abstract: A triple-band multiple input multiple output (MIMO) slot antenna with /4 wavelength slots of varying lengths that cover the operating frequency ranges of 1.6-1.8 GHz, 3-3.3 GHz and 4.54.8 GHz is presented in this paper. To achieve isolation of greater than -50 dB between the ports, a simple decoupler network is built for a large slot and a series of tapered slots. The antenna coupling is reduced by varying the ground plane. The designed antenna is fabricated using FR4 dielectric substrate with appropriate dimensions to achieve radiation across the frequency band from 1.6-4.8 GHz with more fractional bandwidth. Return loss of less than -25 dB and VSWR of 1.1 are obtained in this antenna.
Keywords: multiple input multiple output; MIMO; defected ground structure; DGS; decoupler network; resonant frequency; VSWR.
XML document classification effectively using improved high-performance factor
by S. Sahunthala, Angelina Geetha, Latha Parthiban
Abstract: Nowadays, XML data plays in volume amount of business application. The real World Wide Web has more XML data in the website. The heterogeneous structure XML data classification is the challenging task in the research recently. Algorithms are available to classify the XML data by classification method. The performance is degraded in the classification XML document in the existing technique. In this paper, the machine learning technique tuning improved hyper parameter optimisation algorithm (TIHPOA) is proposed to classify the XML data. First the elements are extracted by using feature extraction vector space model. Then the XML data is classified using the algorithm of TIHPOA technique. The proposed model uses the improved hyper parameters to generate the better classifier than the existing classification approach. In existing approach, extreme machine learning (ELM), kernel principal component analysis (KPCA) and kernel extreme machine (KELM) and tuning swarm rapid swarm algorithm (TSRSA) methods are demonstrated. In this research the proposed model is compared with the existing model with various performance parameters.
Keywords: XML data; classification; feature extraction; TIHPOA.
Intensity modulation based optical fibre humidity sensor using Agarose Chitoson composite
by Yogesh H. Patil, Amrit Ghosh
Abstract: Optical sensors are widely used in chemical industries to monitor humidity and hence assure machine, product, and device quality. The measurable quantity is the vaporised water (H2O) fraction. Optical fibre-based HS have recently been developed for particular applications, with benefits and drawbacks in RH detection. A thin coating of hygroscopic material (Agarose and Chitosan) is deposited on the humidity sensors core. agrose (100%); 2) Agarose + Chitosan (20%) (80%); 3) Agarose 40% + Chitosan (60%); 4) 60% Agarose + Chitosan (40%); 5) Agarose (80%) + Chitosan (20%) (100%). This indicates that both Agarose (40%) + Chitosan (60%) and Agarose (60%) + Chitosan (40%) composite proportions achieve the required sensitivity of 0.34 nm/% RH. The POF-HS with hygroscopic thin film coating design suggests a viable RH detection approach. A low installation cost, small size, and a wide range of visible light intensity.
Keywords: relative humidity; RI; hygroscopic substance; Agarose and Chitosan; Abbes refractometer; plastic optical fibre; POF.
Assessment of mental workload using XGBoost classifier from optimised EEG features
by R.K. Kapila Vani, Jayashree Padmanabhan
Abstract: Cognitive workload evaluation is vital in any critical working environment for assessing the users mental status. Despite the fact that there are many methods for evaluating cognitive strain, the model that uses electroencephalography (EEG) data remains the most promising one. Brain related activities can be used to assess various mental states and also help us to determine mental effort. This study calculates the cognitive workload of people while performing multitasking mental tasks. Here the 'STEW' dataset is used to measure mental effort. 'No task' and 'simultaneous capacity (SIMKAP)-based multitasking activity' are the two tasks in the dataset. For the study we have chosen only the SIMKAP task dataset. The cognitive workload assessment from optimised EEG features using XGBoost classifier (CWAOEX) framework is proposed, in which numerous features from EEG brain signals are retrieved and the grey wolf optimiser (GWO) is utilised to select the best ones. The data is then categorised according to the best feature set. The XGBoost algorithm is employed in the classification step. The recommended method has the classification accuracy of 94.25 in categorising the workload as low, moderate and high which is better than the current methods.
Keywords: XGBoost; grey wolf optimiser; GWO; STEW dataset.
Study on generalisation of triple connected perfect dominating set in fuzzy graph
by T. Gunasekar, K. Elavarasan
Abstract: This study presents the idea of k-connected total perfect dominating set in fuzzy graphs in this study. We have generalised connected perfect dominating set to k-connected perfect dominating set. In fuzzy graphs, new concepts are compared with previous ideas. For a few of the important classes, such as cycles and trees are discussed. The vertex and edge k-connected perfect dominating number of fuzzy graphs are obtained. Maximal fuzzy bipartite part algorithm in fuzzy graph is also described.
Keywords: perfect domination; connected perfect domination; k-connected perfect dominating set and number.
Optical fiber-based refractive index measurement sensor
by Jayprabha Vishal Terdale, Amrit Ghosh
Abstract: Optical devices are increasingly widely used in surveillance, particularly in the food, beverage, and healthcare sectors. A precise refractive index profile is required to support maximum light intensity on the fibre material. But unpolished optical fibre has a difficult RI profile. Changing the solutions concentration allows for precise RI measurement. We present an intensity modulated fibre optic refractive index measurement using macro-bending and side polishing. RI conditions and the dipping solution concentration in the detecting region are more sensitive to probes with a polished surface (sucrose, glycerin, and ethanol). With RI 1.34-1.42, the POF performance sensing capability increases. This allows for physical variations in intensity at visible wavelengths at cheap cost and compact size.
Keywords: intensity modulation; macro-bending; transparent liquid; sucrose; glycerine; ethanol; U-shape polished optical fibre; refractive index sensor.
Renewable energy adoption in 3rd world countries
by Mujahid Tabassum, Saad Bin Abul Kashem, Md. Bazlul Mobin Siddique, Hadi Nabipour Afrouzi, Suresh Ponnan, Hafiz Zafarr Sharif
Abstract: Over the past decade, the global use of renewable energy sources for electricity generation has grown significantly. This rise in the use of renewable energy has mainly been driven by increased awareness of the effects of climate change and the need for a sustainable and cleaner future. Clean energy technologies are essential in reducing poverty, expanding rural development, and protecting health while encouraging sustainability and environmental quality. This research aims to analyse the current technologies being implemented in developed countries and find out the useable matching options for 3rd world countries on lower cost and strong sustainability. Furthermore, a low-cost and sustainable model for a selected 3rd world country is proposed, simulated in Simulink MALTAB to compare, and observe the output and efficiency.
Keywords: solar farm; renewable energy; climate change; sustainable model; simulation; PV.
An incremental learning on cloud computed decentralised IoT devices
by Satish S. Salunkhe, Aditya Tandon, M. Arun, Nazeer Shaik, Supriya Nandikolla, D. Ramkumar, S. Lakshmi Narayanan
Abstract: It is essential that IoT devices can constantly gather new ideas from streams of data independent of catastrophic forgetfulness. Although merely repeating all prior training samples can solve catastrophic forgetting issues, this method faces privacy problems, memory resources, as well as requires a lot of computational, making it unsuitable for limited-resources IoT devices. The proposed incremental learning for cloud computed decentralised IoT devices are developed comprises of constant upgraded information and task resolution model in this study. A neural network is trained and utilised to overcome this problem despite frequent disconnectivity or resource outages without losing a lot of progress using cloud computing. Several research experts have frequent disconnectivity issues regarding cloud computing frameworks because of the platforms free membership. Identical difficulties can be seen when working on a localised computer, where the machine will run out of resources or power at times, forcing the researchers to retrain the systems.
Keywords: incremental learning technique; cloud computing; decentralised IoT devices; internet of things; IoT.
CAD-based automatic detection of tuberculosis in chest radiography using hybrid method
by M. Mercy Theresa, A. Jesudoss, P. Pattunnarajam, Sudha Rajesh, Jaanaa Rubavathy, A. Raja
Abstract: Automated processes are essential in medical imaging to identify anomalies. This study uses chest radiography (CXR) for CAD analysis, which is indicated for about 90% of TB patients. Even when it is cost effective, certain reasons are difficult to pinpoint. Input CXR lung field segmentation, highlights from the segmented lung region, and TB detection calculations. Plans call for three phases: To segment well, this step employs a deformable active contour model. These two parameters are used to assess the algorithms segmentation output. Its now time to extract and optimise features Hybrid multiresolution extracts the features. Various transform coefficients were statistically analysed to obtain a feature collection. The final stage is to classify lung anomalies using MSVM and KNN for three publicly available datasets. The classification performance of the JSRT, Montgomery, and Shenzhen datasets is assessed. The recommended method identifies pulmonary TB 96.5% of the time.
Keywords: computer-aided diagnosis; CAD; chest X-ray; CXR; lung segmentation; novel active contour model; hybrid multiresolution approach; feature extraction; classification.
Effective allocation of resources and task scheduling in heterogeneous parallel environment
by Kirankumar Kataraki, Sumana Maradithaya
Abstract: Scientific studies on a large-scale are being performed in cooperation with teams located around the world. Each team exchanges data and conducts experiments using dispersed resources. Therefore, scientific data is duplicated and stored in geographically dispersed places. These data are incorporated in application processes which ease the operation and maintenance of applications in distributed/parallel computing systems. A workflow management system must take advantage of the presence of different data sources and distributed/parallel computing resources provided by platforms such as grids and clouds to effectively execute these procedures. This paper thus expands an existing workflow architecture and provides better planning algorithms for resources management. It begins with a comprehensive study of the planning techniques used in the past as the basis for parallel systems. The work offers a method that includes information management components and assesses its practical viability by utilising resources to operate several real-world applications.
Keywords: distributed; scientific; parallel; resources.
Artificial intelligence enabled additive manufacturing system using 5G and industrial IoT
by Rudresh Deepak Shirwaikar, Aditya Tandon, K. Sathesh Kumar, M. V. Aditya Nag, Bobin Cherian Jos, Bos Mathew Jos
Abstract: There are numerous types of additive manufacturing equipment, ranging from basic RepRap machinery to sophisticated fused metal depositing systems. Lightweight items, lower errors, minimum tool costs, optimum components usage, and feasible manufacturing method are depending on field implementation. Smart agents powered by AI can help minimise the number of people needed to raise AM productivity with the simultaneous improvement in source usage. The present state of AI-enabled AM item enhancement is explored in this paper. Artificial intelligence (AI), digitalised reality, internet of things (IoT), blockchain, driverless cars, and future innovations are all dependent on 5Gs lightning-fast connectivity and minimal latency. The launch of 5G is much more beyond a generational shift; it ushers in a new period of opportunities for the IT sector. Based on the original objectives and expectations of both domains, the goal of this article is to develop a method that integrates AI with additive manufacturing systems.
Keywords: artificial intelligence; additive manufacturing; 5G; industrial internet of things; I-IoT.
CFD analysis of direct-operated poppet relief valve under different parameters and its structure optimisation
by Yong Sang, Pengkun Liu, Xudong Wang
Abstract: This paper investigates the dynamic characteristics of the direct-operated poppet relief valve (DOPRV) with different parameters and improves its dynamic characteristics by modifying the core. First, a simplified schematic of the valve is introduced; mesh elements, SST k~w turbulence model and corresponding time step size are selected by comparing the simulation results. Then, simulation is conducted under different nominal diameters, and a suitable range of diameter is determined. Further, an improved structure of the DOPRV is introduced, and the dynamic characteristics of the DOPRV under improved structure are stimulated under different annular clearance, spring stiffness, and pressure jump. The simulation results show that adding annular clearance can largely enhance the dynamic performance of DOPRV and the optimal annular clearance for the 20 mm diameter DOPRV in this paper is 0.2 mm.
Keywords: direct-operated poppet relief valve; dynamic characteristics; structure optimisation; computational fluid dynamics; CFD.
Automatic refilling vending machine using Amazon DRS AWS
by R. Suresh Kumar, P. Manimegalai, S. Abishak, N.K.S. Hariharan, D. Pamela
Abstract: Presently, the existing replenishment device may be handiest systematically reviewing the inventories. Whereas, there are ongoing vending machines that may not be able to provide or discover whether the substances are doing nicely to be replenished or no longer. This effect is an inefficient replenishment policy and there will be regularly inventory-out among the products. This undertaking intends to provide the automatics replenishment of smart merchandising gadget with the intention to provide and support inventories to the administrator. Amazon sprint replenishment service (Amazon DRS) permits while the vending device is about to run out of the stock/product, it automatically locate orders on Amazon, administering which of the supplies may be wished for replenishment. By executing this venture, inventory out of the vending machines may be prevented and a green manner of the replenishment system can be carried out.
Keywords: vending machine; Amazon dash replenishment service; Amazon DRS; ESP32; internet of things.
Analysis and implementation of SQL injection attack and countermeasures using SQL injection prevention techniques
by A. Jesudoss, Theresa M. Mercy, A. Christy, M. Maheswari, M. Selvi, V. Ulagamuthalvi
Abstract: SQL injection attack is the most critical and very common attack to security of web applications. The paper analyses the vulnerabilities that arise due to SQL injection attacks and presents a consolidated prevention techniques which consider all vulnerabilities and identifies the SQL injection attacks. It also provides appropriate solution for safeguarding against SQL injection attacks. While being cost-effective, these prevention techniques are also easy to configure, administer and implement. Experimental results have proven that these prevention techniques efficiently identify and protect against SQL injection attacks. The prevention techniques discussed in this paper have been implemented and tested effectively. The results of testing are satisfactory.
Keywords: SQL injection; malicious input; validation; web application firewall; injection attack.
Design and evaluation of a deep CNN algorithm for detecting farm weeds
by Balachandra Pattanaik, Areej Malibari, M. Kumarasamy, V. Nagaraj, M. Gopikrishnan
Abstract: Weeds are unwanted plants that grow with crops and usually removed by spraying herbicides or by manual labour. Herbicides being sprayed mostly do not reach their target because of the focus on a very wide area. This also tends to harm the environment, and other living organisms. Manual labour is time-consuming and expensive and it is continuously managed and monitored. The autonomous robotics and image processing, tasks can be completed with precision and ease in agriculture. With image processing, plants and weeds can be classified. Methods like scale invariant feature transforms (SIFT), speeded-up robust features (SURF), and ensemble learning, neural networks can be incorporated into identifying the difference. We can easily classify weeds and crops from images of plantations leveraging machine learning algorithms, artificial vision analysis systems, among others. Deep learning methods like convolutional neural network (CNN), rectified linear units (ReLU) and SoftMax (for classification) are focussed in this paper.
Keywords: image processing; deep learning; convolutional neural networks; CNN; rectified linear units; ReLU; weed detection; automation.
Optimised dual hybrid energy storage systems for EV powertrain based on modified genetic algorithm
by Mukil Alagirisami, Balachandra Pattanaik, Ramesh Redrouthu, Chandu V.V. Muralee Gopi
Abstract: EV has generally been recognised as a viable substitute for internal combustion engine-powered vehicle. The EV is capacity and lifetime of the energy storage system, leading to decreased drive range of the vehicle and rise in price. To overcome these drawbacks, dual-HESS is introduced in the EV. In this work, batteries and ultracapacitors are utilised as HESS. In this work, fuzzy control is also implemented, which is accountable for splitting the energy among the front and rear wheels units in a far more appropriate manner in order to meet the needs of greater performance. Utilising the MATLAB/Simulink software, the entire framework was built with the FTP-75 (urban), US06 (maximum speed as well as required acceleration) and HWFET (highway) driving cycles. When contrasted to a corresponding existing EV deployed with a solo HESS unit, the suggested dual-HESS architecture enhanced the driving range by 145.15 km as well minimising the HESS mass by 23.93%.
Keywords: electric vehicles; GOA; GA; HESS; power management control; PMC; SoC; FLC.
Robust control of frequency considering operations of AC microgrid in islanded mode
by Sanjiv Kumar Jain, Shweta Agrawal
Abstract: Modeling and control of an autonomous microgrid with both governable and non-governable sources is discussed in this paper. Power sources utilised in microgrid are solar, wind, fuel cell (FC) and diesel based generator (DG). Batteries, flywheels (FW), and aqua electrolyses (AE) are used as energy storage elements. Recently, due to the increased penetration of non-inertial distributed generations in the microgrid, frequency deviation has become a matter of concern. If the loading conditions and active power generation changes unexpectedly, then the power system frequency would deviate largely. The output of the flywheel, battery, diesel generator, FC, and aqua electrolyser is regulated using the proportional-integral-differential (PID) controller. In this paper, a novel mathematical framework is proposed to tune PID controllers. For zero frequency deviations in steady state, optimum values of proportional, differential, and integral gain coefficients are proposed. The importance of the proposed scheme is to achieve better steady state frequency response.
Keywords: diesel generator; frequency regulation; microgrid; PID control; renewable energy sources.
Analytical characterisation of 3D nano scale ultra-thin film surrounding gate MOSFET
by Aditya Agarwal, R.L. Sharma, Prashant Mani
Abstract: This paper is about the characterisation of 3D nanoscale surrounding gate metal oxide semiconductor field-effect transistor (SG MOSFET) and drains current model has been developed for the proposed device. The expressions for transconductance and subthreshold swing were also obtained. The drain current modelling reflects enhanced current and suppressed output transconductance. The expression for threshold voltage is also taken in the current work. The various technology parameters were taken, and the work function difference has also been presented. The results show that the surrounding gate MOSFET device offers improved performance over other devices. Simulation results verified using 3DTCAD Silvaco device simulator.
Keywords: threshold voltage; sub-threshold voltage; surrounding gate MOSFET; short channel effects; SCEs.
An efficient fruit quality monitoring and classification using convolutional neural network and fuzzy system
by K. D. Mohana Sundaram, T. Shankar, N. Sudhakar Reddy
Abstract: Fruit quality monitoring in agro industries is carried out by people who may deviate from their responsibility due to tiredness, illness, or personal reasons. So, an automatic quality assessment system is proposed based on convolutional neural network (CNN) and Mamdani fuzzy logic that estimate quality of a Persian Lemon. The proposed CNN was trained with the transfer learning method and the results obtained were compared with previous works. The proposed CNN achieved 94.79% accuracy in the validation process which is 13% higher than the existing architecture. The proposed fuzzy logic classified each lemon in three ranges based on rules customised for the estimation of fruit quality standards.
Keywords: fuzzy systems; transfer learning; convolutional neural network; CNN.
Machine learning-based financial analysis of merger and acquisitions
by S. Kalaivani, K. Sivakumar, J. Vijayarangam
Abstract: Stock market analysis and forecasting is one of the most sought-after areas of study. As anyone who has observed stock market movements even as an outsider knows very well the enormous amount of risk involved with numerous factors affecting it, its study is quite an intriguing and interesting one, let alone a profitable one. So, it is imperative we look for prediction tools to help us through the process. As we dwell into already available tools in the fields of economics and statistics, we can sense a need of innovation from other evolving domains and the immediate one is the field of machine learning. This paper is a stock market price forecasting one, using neural network model, employed on financial data concerning pre and post-mergers of companies. We have collected data of pre-merger and post-merger states, formed a neural network model to fit it and used the model to forecast. The predictions were reasonably accurate.
Keywords: neural network; financial forecasting; merger; acquisitions.
Insights on future employment and required technical skills pertaining with Oman
by Sanjai Gupta, Mohammed Faisal
Abstract: Unemployment is one of the major problems of any country and a challenge that the graduates lack in required skills and knowledge that employer's need. Joblessness rate in Oman arrived to a stagnant state since 2017 at 3.1% (Oman Unemployment Rate-19912019 Data, 2019). Every year Oman is producing many graduates, which are lacking in required skills and knowledge to get suitable job. In this research, we have extracted the jobs and the skills from Oman job portals using RapidMiner, finding more demanding skills in the fields of engineering, information technology and sales, mapping the skillset with the college outcomes. The research possibly identifies the areas to enhance the employability skills and job profiles to meet the job requirements of in Oman. These inputs can further help the concern authorities to decide upon training the employable candidates to bridge the gap between employers expectations and the college produce (Rehman, 2014).
Keywords: skills; skillset; employability skills; text-mining; RapidMiner; mapping skills; 4th industrial revolution; Oman jobs; Oman.
A modified hidden Markov model for outlier detection in multivariate datasets
by G. Manoharan, K. Sivakumar
Abstract: The processing of data is an essential part of any field. More than 80% of the study effort is focused on collecting meaningful information from the vast amounts of data available. However, in order to minimise calculation time and improve accuracy, it is necessary to keep track of any unused, redundant, or irrelevant data in the dataset. Because its tough to build up a data warehouse to separate homogeneous data, it will be inefficient and inappropriate in terms of deployment costs and performance metrics. Meanwhile, handling heterogeneous data consumes more time to process due to uneven data samples and missing data. Thus, identifying the data class and balancing the data is critical for improving the performance of classification models. Outlier detection is the process of detecting irrelevant, missing, or unequal data samples in a large database. The goal of this study is to employ a modified hidden Markov model to find such outliers in a big dataset. This method improves classification model performance while also reducing computation time and increasing classification accuracy. The proposed model is experimentally verified and compared with prominent existing technologies such as random forest and decision tree models.
Keywords: outlier detection; hidden Markov model; HMM; classification; support vector machine; SVM; random forest; RF; decision tree; DT.
Effect of semi batch and fed batch addition of glucose on alkaline protease production: a multi-objective optimisation approach
by Anitha Mogilicharla, V. Swapna, Rajasri Yadavalli
Abstract: Alkaline protease is one of the important enzymes in many industries. In this effort, semi batch addition and fed batch addition of glucose have been considered for maximisation of protease concentration in minimum fermentation time. The kinetic model of the process is validated with the experimental batch and fed batch addition of glucose from the open literature. A theoretical study has been conducted with such a validated model to check the effect of protease concentration on the semi batch addition of glucose. Based on this, multi-objective optimisation studies have been done for the simultaneous minimisation of fermentation time and maximisation of protease concentration with the relevant constraints. The elitist non-dominated sorting genetic algorithm (NSGA II) has been utilised for this purpose. The additions of glucose in semi batch mode show the potential increasing of protease concentration in at a less fermentation time as compared to the batch experimental data.
Keywords: protease; semi-batch addition; fed-batch; NSGA II; multi-objective optimisation.
HCP miner: an efficient heuristic-based clustering method for discovering colossal frequent patterns from high dimensional databases
by T. Sreenivasula Reddy, R. Sathya, Mallikhanjuna Rao Nuka
Abstract: This paper presents an efficient heuristic-based clustering method for colossal frequent patterns discovery from the high dimensional databases (HCP miner). The HCP miner avoids exhaustive level-wise pattern tree traversal and quickly mines colossal patterns from the high dimensional databases. To achieve this, our approach constructs the sub-patterns using a lattice array and applies the binary clustering over the sub-patterns initially. While constructing the sub-patterns using a lattice array, it uses the support values. These sub-patterns are explored as conditional patterns by estimating core patterns using heuristic measures to minimise the searching time during the database scan. Finally, colossal cluster is constructed from which colossal patterns are discovered. We perform the experiments on various high dimensional databases using different performance metrics. Our experiments shows that, the proposed HCP miner achieves prominent and efficient results for mining. In addition, these analysis of results reveals that the HCP miner algorithm outperforms with CoreFusion, colossal pattern miner (CPM) in diverse aspects.
Keywords: data mining; big data; frequent pattern mining; FPM; high dimensional database; FIMI dataset; backtracking search.
Machine learning and image processing technique to describe outdoor scenes for visually impaired people
by S. Pavithra, V. Prabhakaran, T. Helan Vidhya, D. Gururaj, P. Shanmuga Priya
Abstract: It is estimated that there are about 280 million visually impaired individuals in the globe, who are unable to see and experience the world in the same manner that a normal human being does. Using digital image processing and voice processing, we want to assist visually impaired individuals in interacting with the actual world by narrating the descriptions of a scene in front of them, as described in our article. In addition to this, our gadget serves as a personal assistant by keeping the user up to speed on the latest developments.
Keywords: blind; assistive device; image processing; speech processing; image recognition.
Predictive analysis of smart agriculture using IoT based UAV and propagation models of machine learning
by M. Kumarasamy, Balachandra Pattanaik, Jaiprakash Narain Dwivedi, B.R. Ramji, Muruganantham Ponnusamy, V. Nagaraj
Abstract: Every year, unfavourable weather conditions cause many crops to fail. Every time, over 12 million dollar losses are recorded. This article provides a proper background for delivering the yield's current state. The project proposes to employ IoT-based unmanned aerial vehicles (UAVs) and tensor-flow machine learning to estimate crop yields. This framework enhances agricultural yield accuracy by using unmanned aerial vehicles (UAVs) (UAV). The IoT-enabled UAV module captures data and texts it to the farmer or rancher. Data cloud storage. This server uses MQTT for safe data transmission. The cloud server leverages UAV for continuous surveillance and harvest forecasts. Predictive analysis using propagation model has an accuracy of roughly 85% compared to real-time analysis for the same crops at the farm.
Keywords: predictive analysis; unmanned aerial vehicle; UAV; smart agriculture; machine; learning; internet of things; IoT.
Ultra-low latency communication technique for augmented reality application in mobile edge computing
by S. Narayanan, Rakesh Kumar Arora, Sanjeev Gangwar, J. Pradeep Kandhasamy, T. Ratheesh, K. Murali
Abstract: In wireless communication, ultra low-latency communications (ULLC) services offer short packets that may coincide alongside enhanced mobile broadband (eMBB) services which send lengthy packets. In a mobile edge computing architecture with eMBB and URLLC services, we explore how to evaluate latency and enhance shorter packet offloading methods. Using eMBB and URLLC functions, we examine how to evaluate latency and enhance shorter packet offloading methods in mobile edge computing (MEC) approach. In the MEC system, a server called processor sharing (PS) is utilised to reduce computation delay for short packets by distributing the servers whole processing power equally to all packets. The server ignores long packets in favour of shorter ones. With a small packet size, a closed-form formula for the complementary latency distribution function may be developed. To minimise the short packets end-to-end (E2E) latency, offloading probabilities are adjusted based on the dependable demand. In tests with short and long packets, the processor sharing server outperforms 2 types of first-come, first-serve servers.
Keywords: mobile edge computing; MEC; processor sharing server; ultra low latency communication augmented reality.
Image quality estimation based on visual perception using adversarial networks in autonomous vehicles
by D. Vijendra Babu, A. Umasankar, K. Somasundaram, C.M. Velu, A. Sahaya Anselin Nisha, C. Karthikeyan
Abstract: To improve autonomous cars, the dynamic systems method is re-enacted. Due to the unreality of the sensors employed in vehicles, human creation of the surrounding environment and objects is necessitated. We propose a novel efficient method for generating accurate scenario sensor data using limited LIDAR and video data from an autonomous vehicle. A new SurfelGAN network recreates realistic camera pictures to recognise the cars and moving objects in the scenario. The suggested approach uses real-world camera image data from Waymo Open Dataset to evaluate actual scenarios for autonomous vehicle movement. A new dataset allows for simultaneous analysis of two autonomous cars. This dataset is used to test and explain the proposed SurfelGAN model. GAN is the greatest technique for capturing realistic pictures. The machine generates precise sensor data that is used to identify obstacles, cars, and other moving objects in the route of an autonomous vehicle. The autonomous car approaches the destination by recreating a surfel scene. Pictures are collected using semantic and instance segmentation masks.
Keywords: generative adversarial networks; GAN; visual perception; image quality assessment; IQA; autonomous vehicle; SurfelGAN.
Scalable image compression mechanism for surveillance video summary
by T. Venkata Satya Vivek, Manoj Kumar Gupta, J. Pradeep Kandhasamy, Renu Kachhoria, Santwana S. Gudadhe, S. Lakshmi Narayanan
Abstract: The use of large-scale video surveillance systems is widespread in important areas such as home and public safety. Recognising and evaluating appropriate security measures is critical since these systems are vulnerable. A clear movie requires good compression. Lossy image compression may decrease the amount of bandwidth needed for picture transmission and the amount of storage available to a device, improving network performance. Neural networks have thrived in image processing thanks to deep learning. We present an image reduction technique based on semantic analysis based on the degree of human attention to each region of the picture. After evaluating the semantic images using a convolutional neural network (CNN), a compression bit-allocation algorithmic technique is used. This technique enhances video surveillance visual quality while keeping the same compression ratio.
Keywords: convolutional neural network; CNN; image compression; recurrent neural network; scalable image; video surveillance.
Bayesian-based binary compression with bandwidth optimisation for UAV aerial images
by Pankaj Agarwal, Sapna Yadav, J. Pradeep Kandhasamy, A. Balaji, S. Markkandan, D. Vijendra Babu
Abstract: This article proposes a new Bayesian-based binary compression model for UAV aerial pictures. This technique utilises inter-signal correlations to extract several sparse signals simultaneously. BKF-based approach employs both intra- and inter-signal correlations. The Bessel K-form (BKF) also features a higher zero peak with longer tails. Consumers may use UAV-borne base stations for temporary or emergency services. The effectiveness of low-bandwidth wireless Bayesian UAV communication BS still a challenge. This studys aim is to enhance UAV-BS spectrum usage while maintaining user fairness. Through aerial picture quality, we propose adjusting the distribution of bandwidth, power, and UAV-BS trajectory to capture the object image. The proposed method outperforms other approaches in aerial picture detection. To get high quality aerial images, Bayesian-based binary compression lowers picture size and minimises noise. The advantages of UAVs using the Bayesian approach have spurred research interest in novel communication systems.
Keywords: Bayesian method; binary compression; bandwidth optimisation; UAV aerial images; Bessel K-form; BKF.
Hysteresis controlled single phase-VIENNA rectifier fed DC drive system with enhanced response
by B. Manimaran, R. Rani Hemamalini, Ramareddy Sathi
Abstract: This work deals with the forming, enquiry, strategy and simulation* of proportional resonant-proportional resonant (PR-PR) and hysteretic-controlled two loop single-phase VIENNA rectifier fed DC drive system (SPVRDDS) using MATLAB Simulink. VIENNA rectifier with low THD is proposed for the control of DC drive. HC is suggested for closed loop SPVRDDS. Simulation is done for PR (proportional resonant controller) and PI-HC (proportional integral-hysteresis controller) controlled two loop systems using VIENNA rectifier fed DC drive and the outcomes are evaluated. The assessment is done in terms of time domain parameters like settling time and steady state error. The endings of SPVRDDS represent grander concert of HC controlled two loops VIENNA.
Keywords: THD; PIC; PRC; Swiss-rectifier; PWM-control.
Implementation of wearable device for upper limb rehabilitation using embedded IoT
by M. Veeresh Babu, V. Ramya, V. Senthil Murugan
Abstract: In this paper, designed a low-cost upper-limb rehabilitation device that includes sensors and in-built technology that allows for accurate movement evaluation and mussel force. The function modules designed encompasses a multiple mechanical structure, sensor in addition driver circuits, database, as well as interactive interface. In mechanical structure design, 3D printing is used for obtaining some key components. The proposed upper-limb rehabilitation has advantage of low step angle, holding torque, pressure and angle monitoring, assist or opposes upper limb mussel motion with low cost. The data storage and analysis is obtained using windows application, and control implementation is obtained by DSPIC30F4011 and stepper motor. The suggested upper-limb rehabilitation scheme is integrated and tested; the necessary parameters are verified.
Keywords: wearable device; IoT; upper limb; 3D printing.
Privacy-preserving with data optimisation in social networks using ensemble algorithm and K-neural network
by P.S. Arun Karthi, S. Sathiyabama
Abstract: A huge collection of data is being produced for each second due to advance technology development and its innovation in the social media. Monitoring the system and network, securing lines and servers are come to end by using various mechanism. The data accuracy has been increased by the K-nearest neighbour (KNN) classification. A deep learning technique which is a neural network is used for the detection of attacks being done by hackers or fraudulent users. The proposed model uses a programming language called Python which has packages of Scikit-learn, Tensorflow and Seaborn. We have also developed and found the accuracy rate gets increased by the deep learning model and so the attacks made on the social networks have been avoided as much as feasible.
Keywords: social networking; data mining; intrusion detection system; IDS; data optimisation; deep learning techniques; clustering-based IDS; ensemble algorithm; KNN classification.
3D segmentation of brain tumour
by Rudresh Deepak Shirwaikar, Kuthika Ramesh, Abu Mohammed Faisal, M. Jeshwanth, Aditya Raghav
Abstract: Brain replacement is a grouping of abnormal cells in the brain. Brain tumours may be malignant or non-cancerous. Glioma, meningioma, and pituitary are common brain tumours. MRI scans may detect these cancers at various stages. There are many ways to classify and extract features from MRI brain tumour pictures. The CNN image classification approach accurately detects early stage tumours. Explore the 3D CNN using brain tumour segmentation. Training several CNN architectures to compare their performance and design collects local and contextual data. One of the drawbacks of 3D design is memory use. 3D convolutions are computationally intensive and have exponential parameters. However, if correctly done, automatic identification of crucial traits without human supervision is conceivable. It is tremendous computational efficiency makes it the most often used design. Our main objective is to optimise memory consumption and processing to detect brain cancers in 3D MRI data. The 3D CNN architecture removes brain tumours first, then feeds them to a pre-trained featured extraction for CNN model. Using these extracted features to choose the best features using correlation-based output. This is done through feed-forward neural networks.
Keywords: tumour; CNN; segmentation; U-net architecture; accuracy; encoder; decoder; training; validation; dice score.
A scheme based on ECDSA and its implementation for information security
by G. Mallikharjuna Rao, K. Deergha Rao
Abstract: Cryptography methods are means of securing digital data on a network. Digital signature on a document today is trendy in the digital world for authentication, authorisation, integrity, and non-repudiation. The elliptical curve digital signature algorithm (ECDSA) has been implemented and proved that it requires a key size of small length as compared to the RSA. However, its implementation for image and audio security is lacking in the literature. Hence, this paper has proposed a scheme based on ECDSA for text, image, and audio security over networks. Further, the proposed method is implemented on text, image, and audio using both LabVIEW software and myRIO hardware and verified for authentication.
Keywords: digital signature; ECC; elliptical curve digital signature algorithm; ECDSA; cryptography; security.
Study on optimisation of seismic performance of special-shaped column structure in residential buildings based on BIM technology
by Peng Zhang, Li Wang, Jing Zhong
Abstract: In order to solve the problems of poor seismic performance of special-shaped column structure of residential buildings after traditional optimisation methods, a seismic performance optimisation method of special-shaped column structure of residential buildings based on BIM technology is proposed. Building visual information base model is constructed by BIM technology; The variation of parameters of special-shaped column structure in residential buildings is analysed; considering the economic benefits of special-shaped structure of residential buildings, build a comprehensive two-way driving structure; obtain the displacement parameters of the special-shaped structure of residential buildings, analyse its multi-degree of freedom system and equivalent single degree of freedom, determine the displacement degree of the special-shaped column structure of residential buildings, and complete the optimisation of the special-shaped column structure of residential buildings. The results show that the maximum displacement curvature ductility coefficient is about 4.5 and variation range of energy dissipation coefficient is 0.03~0.04.
Keywords: BIM technology; residential building; non-standard pillar; Rhino parametric model; seismic performance optimisation; special-shaped column structure.