Proceedings of Fourth International Conference on Computing, Communications, and Cyber-Security

With the boom of industrialization and the escalation of world population, there is a significant decline in the land suitable for farming and an increase in demand for food. Hydroponics, a branch of agriculture, facilitates the growth of plants in water without the use of soil. This offers a lifeline of sorts by making vertical farming possible. Hydroponics gives us the freedom to wholly control the parameters that affect plant growth, and if done correctly, it can be beneficial both commercially and environmentally. Manual monitoring of all the variables of plant growth is difficult and can lead to growth failures. This paper proposes how the Internet of Things (IoT), a revolutionary technology can be integrated with hydroponics to automate the process of climate monitoring and control. A practical model is implemented in this paper where IoT sensors are used to control the climate conditions and ambient lights of the hydroponic farm. Sensors monitor the climate and light conditions throughout the day and automatically maintain suitable growth conditions.

Unmanned aerial vehicles (UAVs) are small aircraft that are guided autonomously, or by remote control, or both. Generally, these UAVs are used for providing aerial views when tackling complex scenarios in dangerous locations where human intervention is not possible. Internet of Drones (IoD) is a new term that has been created for separating general IoT devices like smart refrigerators, bulbs, cars, etc. from special use-cases drones. IoD is an infrastructure designed to provide access and control of UAVs over the cellular or internet via the ground station. The use of insecure drones in the civilian domain risks the general population to a very high privacy breach, as well as other domains such as policing, medical support, public safety, and industrial delivery system. Therefore, the concept of cybersecurity issues, privacy risks, and vulnerabilities introduced in the UAV ecosystem is important and is explored through this work. This research paper deals with UAV infrastructure and lays out a plan for the IoD communication architecture based on QUIC messaging protocol rather than TCP over the 5G cellular network. According to our research, over 87% of the UAVs are vulnerable to at least seven types of attack vectors and their variants that we have mentioned. Only high-end UAVs that are used for military reconnaissance and recovery operations are thoroughly secured to evade basic attacks but are still vulnerable to sophisticated attacks.

There are a variety of mishappenings that may occur in the network, including explosions, network failure, and battery optimization. Due to the fact that the nodes in an ad hoc network do not remain in the same location, it is very difficult to determine which node in the network is at fault or which node was affected by the event. This study uses artificial bee colony (ABC) optimization to find the various pathways. A neural network is used to perform both the mitigation and identification processes. The ABC algorithm is used for event management throughout the course of the route discovery process. It is proposed that a technique is developed for the generation of an ad hoc network for the purpose of identifying route mechanisms that include identification and the movement of nodes within the coverage region. As a result of this research, an improved technique for the route-finding process has been provided. Using an ABC algorithm and a Levenberg–Marquardt neural network, the faulty node can be diagnosed with this technique.

With ever-increasing smart applications in diversified domains, ranging from home automation, smart cities, industrial applications, military applications, remote sensing, etc., number of wireless sensors, wireless modules, and processing units have multiplied many folds from thousands to zillions, so has the requirement of power backup and energy harvesting strategies. Concerns about efficiency of energy usage have become a prominent issue in the networks deployed at sites for communication. With the 5G in sight, a slew of some base stations and the demand for energy-efficient system design and administration is increased by a plethora of linked devices. Awareness about energy harvesting has increased in fifth-generation wireless networks due to extensive consumption of energy in smart applications which have grown exponentially in last decade. Energy harvesting technique is one potential and probable solution for enhancing the lives of networks as well as devices. Till date, several researches have been carried out by many researchers for use of techniques for harvesting of energy in fifth-generation wireless networks communication, and these are still at very primary stage and lot of room for improvement there. The purpose of this work is to present the advantages of energy harvesting technology in fifth-generation network communication. Work presented categorizes the available literature on harvesting of energy in fifth-generation networks for energy harvesting devices, energy conversion methods, phases, resources of energy, and propagation medium of energy. Work concludes with the challenges and opportunities ahead with energy harvesting techniques and outlines research gaps with future directions.

The development of microfluidic biochips has been one of the fastest-growing study fields in recent years. Microfluidic biochips are guiding the miniaturization of laboratory-based bioprotocols on a tiny chip. All bioprotocols, with the exception of dilution, include a stage called “sample preparation” that allows for the blending of several reagents in a specific volumetric ratio. It calls for mixing and storing a number of reagent fluids for real-time implementation, which offers a cheap and dependable method for on-chip sample preparation. Concentrate on minimizing reagent consumption, reducing fluid waste, achieving an accurate volumetric ratio, maximizing resource utilization, minimizing dilution time, and minimizing cycle count during sample preparation because many scheduling algorithms have been developed for performing sequences of mixing operations. To obtain a fixed concentration value, some of the reactants are blended and diluted in the appropriate collection. In practice, a reservoir switching operation takes longer than a mixing operation because it involves unloading, washing, and loading. The reservoir-constrained optimal scheduling (ROS) algorithm proposed a workaround to resolve the switching operation. This algorithm is used to speed up reservoir changeover, although more storage units are needed as a result. The cost of chip manufacture limits the number of storage units in real-time implementation. Hence, we propose a time quantum-based scheduling algorithm for digital microfluidic biochips to meet the storage constraint while reducing the switching count. The time it requires to execute a dilution operation is symbolized by the term “mixing time,” and the term “tree” represents the preparation of a sample. The proposed algorithm focuses on the mixing graph and scheduling algorithm.

The conventional agricultural sector is declining. In the meantime, due to the prevalence of modern technology which has increased, smart agriculture has evolved in farming in response to rising global population and ongoing industrial revolution. The rapid growth of the Internet of Things has captured the interest of every sector (IoT). The Internet of Things (IoT) is reshaping several sectors, including the agricultural sector. The increased need for agricultural expansion and modernization is a direct result of the high demand for food. The Internet of Things (IoT) is a set of programming tools that we can use to implement a wide variety of strategies for the modernization of agricultural practices. Researchers and research institutes from all around the world are competing with one another to supply surplus Internet of Things technology to stakeholders in the agriculture industry to lay the groundwork for a clearly articulated goal in preparation for the time when Internet of Things technology will be widely used. This review states a methodical overview at the impact that the Internet of Things plays in agriculture as well as how it is facilitating new approaches to increasing agricultural productivity for the benefit of farming in the future. Graphic representations are employed for illustration of several ways in which the Internet of Things might be applied in the sector of agriculture, as well as its role in “smart agriculture in the modern world.” In addition, a smart agricultural architecture model has been presented, which collects data from farms and fields by using a variety of sensors in various locations. A conclusive outcome is also presented.

In this paper, supercontinuum (SC) generation is demonstrated using tetrachloroethylene (C2Cl4)-filled photonic-crystal fibers of hexagonal pattern based on fused silica glass. Here, the air holes of the inner ring have been infiltrated with tetrachloroethylene (C2Cl4) to get better results. It is seen from the simulation results that, by applying an optical pulse of pump power of 16.66 kW at 1.56 μm wavelength in a 50 mm long PCF with the help of flat and anomalous dispersion, wideband SC spectra from 0.7 to 3.1 μm can be generated. The proposed design technique has potential to be applied in the fields of biological imaging and optical coherence tomography.

The huge increase in the count of devices and the connections among them has birthed the Internet of Things (IoT). The capabilities of wearable health devices are immense, and their potential has especially become apparent during the Covid-19 pandemic. The physiological changes recorded using the activity tracker bands, and smartwatches can be modeled as a data format that supports semantics, i.e., Resource Description Framework (RDF), to help enable interoperability. The objective is to observe and analyze this monitored data to establish how prediction models based on it can be utilized for early identification of infection (pre-symptomatic detection) and effectively raise alerts for early isolation. In this paper, sensor data of heart rate, step count, and sleep hours from users’ Fitbit trackers have been used for showing how Covid-19 infection can be detected early than by using symptoms’ data.

The 5G network is anticipated to be more densified in the future, containing numerous heterogeneous cells. Managing the heterogeneous networks (HetNets) becomes challenging and almost unattainable. Self-organizing networks (SONs) are needed to ensure flexiblity and automatic deployment and maintenance of the 5G networks. Automated cell outage detection is a prominent research focus since self-healing SON solutions execute compensation processes to mitigate network disruption. This study presents a Fisher’s discriminant analysis (FDA) to obtain feature vectors with lower dimensionality, which are suitable for hidden Markov model (HMM). The proposed FDA-HMM automatically predicts the present status of 5G base stations (BSs) and determines a cell outage. The proposed FDA-HMM outage detection scheme’s performance is compared with existing algorithms such as the conventional HMM, support vector machine (SVM), and random forest. The results of simulation indicate that the proposed FDA-HMM algorithm effectively detects cell outage with 97.02% accuracy as compared to the exisiting supervised learning methods.

Swelling urban population and increasing the number of vehicles with finite road capacity account for a major section of air pollution in cities and towns, causing problems like the smog that lead to serious health issues. Exhaust gases from vehicles contain dangerous gases that poison the surroundings we live in as well as cause climate change. Congestion in traffic leads to a higher impact on the health of individuals as they remain in polluted conditions for a longer period which increases morbidity and mortality of travelers near high-density roads. As the environmental pollution problem has aroused more and more attention from the public, there is an increasing need to control and manage vehicles traffic, and also inform citizens regarding the ambient air quality to reduce the risk of health problems. This paper presents a cloud-based framework for effective air quality monitoring using machine learning. In the proposed approach, time series analysis is performed on the Spatio-temporal data that contains values of different gases and calculated air quality index (AQI) values concerning time and space. The experimental results indicate the efficacy of the proposed framework and prediction model. The LSTM-based proposed approach achieves 99% AQI prediction efficiency that helps to recommends pollution less route. Finally, the paper is concluded with issues, challenges, and future scope.

With advancement in the growth of network technology, Unmanned Aerial Vehicles (UAVs) attained a new position in imparting its functionality in the diversified domains such as military and civilians through providing a set of services such as data transfer, photography, construction, monitoring and surveillance et al. UAVs provides such services and functionalities with the help of drones. In all domains, Drones help in assisting the human beings but there are several challenges that needs to be tackled, major one is the communication. Drones can be connected in adhoc mode with the help of advancement in the functionality of microprocessors that helps in providing intelligent autonomous control of various systems. The distinguished features possessed by drones are: dynamic behavior of node mobility and network topology, variable network performance, flight range, autonomous and remote operations, fast data delivery and cost effective. In this paper, we present the literature study of all the work related to the UAVs in the field of Drones, several challenges faced in the field of UAVs, applications of Drones, advantages and disadvantages, simulator tools, future work and conclusions. All list of services and functionalities for the Drones applicable in the field of UAVs. We also summarize the most promising Drone applications and outline characteristics of Drones. The paper conclude by providing the future work towards promising directions.

By adjusting its electric and magnetic properties, the reconfigurable intelligent surface (RIS) can control how electromagnetic waves propagate. The characteristics of radio channels can be altered by using these surfaces in a particular application. Cognitive radio (CR) is a revolutionary method that allows a transceiver to find vacant radio spectrums. The transceiver then chooses a free channel rather than using the ones that are in use at the moment. The radio frequency (RF) spectrum may become more productive as a result of this. The RIS-based CR system may make use of both technologies’ advantages. The aim of the paper is to evaluate the energy efficiency (EE) performance of RIS-enabled CSS. The EE is the all-inclusive statistic that takes into account all system characteristics and represents the CSS system performance. A system model that establishes the EE of CSS with RIS support has been put out. The EE performance has been examined in relation to the number of sensor users, the detector threshold, and the sensing time. The findings indicate that there is an optimum number of reflecting elements at which maximum EE is achieved.

The paper look into the non-orthogonal multiple access (NOMA) for a network including terrestrial and satellite communication both. The users can access the satellite through base stations. The intra-cell interference, inter-cell interference, and cross tier interference influence terrestrial communication. The work provides a full coverage to terrestrial users through BSs or through satellite. The poor channel users or to the users to whom BSs are not able to provide coverage and request coverage from satellite via BSs. The user grouping is done between a far user and near user. The simulations are done to evaluate the performance under varying number of subchannels of BS and BSs.

In this era of fake digital content, deep generative models have lately demonstrated outstanding outcomes in a variety of real-world applications like open access to large-scale public databases. Moreover, we can generate high-resolution and diverse samples with the help of advanced deep learning techniques, particularly Generative Adversarial Networks (GANs). This results in the development of remarkably realistic fake content, causing significant concern and spreading distrust in multimedia content with related societal impact. Hence it arises the urgent need for automated methods to detect fake multimedia generated by artificial intelligence. Although many face editing algorithms appear to produce realistic human faces, closer inspection reveals aberrations in specific domains that are typically invisible to the naked eye. These deep learning-based contents are knowns as deepfakes. There are four broad categories of deepfakes that are as follows: photo deepfakes, audio deepfakes, video deepfakes, and audio–video deepfakes. This paper provides a review of the existing generation and detection methods of the various deepfake contents. It also provides a detailed comparison of the objectives, methodology, and algorithms proposed in various studies by different researchers in recent years. Finally, the paper concludes with the notion that one should minimize the restrictions and bottlenecks that are experienced by the existing methods by proposing more advanced techniques of detection.

Cybersecurity is a collection of technologies, policies, and operations that are used to protect networks, devices, software, and data against intrusions, damage, malware, viruses, hacking, and security breaches. The protection of the customer’s assets is the primary objective of data security in banks. As more people become cashless, additional acts or transactions go online. Transactions are carried out using electronic payment methods like debit and credit cards, which require cyber complete security. In this paper, we are proposing a framework for KYC based on biometric retina recognition which is more secure than the existing framework. During the onboarding process, our proposed framework employs the unique digital identity of each customer and lessens the chance of digital identity theft. Retinal biometric recognition may be implemented in ATMs in the future to guard against fraud risks such as cash trapping, card skimming, bogus assistance, eavesdropping, shoulder surfing, fake PIN pad overlays, hold-ups.

The crop disease and the plant disease are a major issue affecting the food quality and quantity in agriculture. The lack in detecting the crop disease impacts the crop yield and the farmers’ income. Early detection of leaf diseases will prevent and control the diseases in the initial stages. The objective of the proposed work is to develop a smart phone-based plant diseases detection system using deep learning techniques. This system aims to provide information to the farmers about to the identification of the bacterial, fungal and pest diseases and provide management and control especially of the paddy plant. Paddy is indigenously grown in delta region of Tamil Nadu. Millions of hectares of paddy fields are infected annually by deadly diseases, and crop loss may be as high as 75%. The proposed work uses deep learning-based MobileNetV3 architecture to detect the leaf diseases from images to reduce the major crop losses. The performance improvement with 93.75% accuracy is achieved through the proposed model.

RFID (radio frequency identification) is a technology that enables the connection of physical objects to the Internet of Things (IoT) by using radio waves to transmit data. RFID can connect any physical object to the Internet of things. RFID can connect any physical object to the IoT. When RFID is widely used and adoption rates rise, security and privacy issues become inescapable. Unauthenticated tags and readers may also send out false or malicious messages, which can be a security risk. It is important to address these issues in order to ensure the safe and secure use of RFID technology. RFID authentication is needed for IoT. We assess the security of the hash-based and ECC-based RSEAP2 authentication mechanism. RSEAP2 contains serious security weaknesses such as mutual authentication, key extraction challenges, session key agreement, and denial of service threats, availability issues, according to our security analysis. As a result, RSEAP2 in VCC is vulnerable. Furthermore, the computation cost consumption was $$\approx 22.0617$$ ≈ 22.0617 , the communication cost at the sending mode was 672 bits, and receiving mode was 512 bits.

In recent years, the use of the Internet has been proliferating. Organizations use social media platforms to promote their products and find people’s opinions about their products. People share their experiences, views, and thoughts on social media platforms such as Twitter, Facebook, LinkedIn. Business organizations analyze the posts created by the users about their products through sentiment analysis. India is multilingual, so people use multiple languages to post their opinion, e.g., English and Hindi. Code-mixed language is a way in which people use words of different languages to express their thoughts. There is limited research done on code-mixed languages such as Hinglish, even though a great deal of effort has been devoted to assessing the sentiments of a single language. A comprehensive review of the Hinglish dataset is offered in this study. In this research, we first discuss several datasets used by the researchers to tackle this problem and then perform some extensive investigation to look more closely at them. Then, we examine several efficient approaches for categorizing sentiment for the Hinglish dataset.

This paper considers the possibilities of using Java Server Pages technologies as an alternative to the existing platforms on Hypertext Preprocessor (PHP) basis. Data relating to relative productivity were collected and processed. There had been also analyzed the statistical data on the trends for supporting different programming languages considering the technologies with the significant potential. The paper also presents the short discourse of the unit, chosen as such an alternative JSP with the example of the simplest program. There had been given the approximate (not the full one) list of possibilities of the above platform to demonstrate its capacity and the potential in the development of the powerful web applications. The given examples and analytical issues allowed to make a conclusion about the application of JSP as an alternative to PHP.

Fog computing originated from cloud computing. It provides a distributed environment for providing services to mobile users having heterogeneous configurations. Microservice is an efficient mechanism to subdivide the tasks of an application in a distributed environment. Load balancing in microservice environment is a challenge with heterogeneous fog devices and end devices. As it is important to make sure neither of the fog nodes should be overloaded nor underloaded to utilise the potential of the fog environment. In this paper, a novel approach for handling the load in a microservice-based healthcare application in fog environment has been presented. Load balancing is performed on the basis of parameters which are getting updated after the successful execution of requested microservice which will serve as a feedback mechanism to handle failures in microservices while ensuring load balancing in fog environment. The proposed technique, when Simulated on iFogSim2 simulator resulted in considerable better throughput and reduced execution time which are key parameters for determining the performance of load balancing algorithm.

Cloud computing has been developed as a front-line platform for consumer and commercial potential. Additionally, it enables users to access data and apps from any location. Businesses will use cloud services to borrow storage as well as other operations, which significantly lowers the cost of technology. In this aspect, resource allocation is more important to assure the efficiency of the cloud service. The primary goal of this study is to suggest a new resource allocation model that uses optimization to help maximize sales while minimizing cost. An improved k-means clustering with an optimal tuned centroid is used to cluster all workload (the jobs that have been allotted). Here, tasks are clustered as per the consideration of QoS (trust) and task execution time. Subsequent to this, the resource allocation is done optimally under the consideration of power utilization effectiveness (PUE), CPU usage, and execution time, respectively. For solving the optimization problem specified, this work proposes a new self-upgraded coot optimization (SUCO) algorithm. Finally, the proposed SUCO algorithm was evaluated over the other models.

In this paper, we propose an innovative method XGBoost to enhance the company’s engagement level with the customers based support with machine learning concepts. The trade-off between the customer needs and providing services towards them plays a major role in the service management sector. So, it is necessary to understand the customer needs, and offering better services plays a major task in today’s Internet world. Most companies cannot find the exact root cause of the minimal number of users on buying/reviewing the particular product. Also, the subscribers’ usage patterns may vary dynamically on a time-to-time basis. Most companies will earn millions and millions of revenues depending on the subscriber base and the level of engagement of their subscribers. So, it becomes necessary to evaluate the parameter for companies’ fall and make forward the progression towards the next level of enhancement for earning a good amount of revenue. Thus, the XGBoost classifier gives the idea behind the focus on the improvement of the shortfall of subscribers in the industrial management system.

Cleaning is critical to each location for a smooth surroundings which ends up in germs-loose surroundings. At instances, it appoints people for cleansing and will pay coins, and a number of the time cleanings are wanted in areas in which the presence of a dwelling being is unstable so it could not relegate humans in every spot. A few spots have a large ground vicinity in that spot for cleansing purposes, it wishes a couple of person so it required a few tactics to make up for those issues. In the headway of science, a robotic comes into the light, and computerization is an first-rate solution to this issue. This paper intends to foster an independent robotic that may flow with out ceaseless human direction. While doing marketplace studies, we got here to realize approximately some technical factors which may be improved. We additionally took a survey from a big network to apprehend their problems at the same time as cleansing in order that we are able to enhance the proposed version and make it user-friendly. The programmed cleanser robotic incorporates low-energy devouring digital elements, and it could paintings at low force. We proposed a self-ruling ground-cleansing robotic; this is operated through the Internet of factors and Raspberry Pi programming. The novel concept we have got used is to make power reusable and cooperate with different cleaners the use of numerous technologies. The renewable power-based “CleanO” will assist us to decrease the time and power and provide a residue/germs-loose climate.

The global “pandemic” of 2020, “COVID-19,” left us all enclosed within our homes for over a year, with 3 to 4 waves adding up to a total of over six million deaths, forcing us to learn to operate from home. The elderly ditched their walks in the park, the adults realized to work and network from home, and the children quit going out to play at school. The school where they learned not only academically but also experienced holistic development. Adults have the patience and the maturity to understand the need to stay indoors to avoid a fatal viral infection. Still, children aged 2–5 years were compelled to remain home without correctly understanding “why.” This paper is a bibliometric survey of these children’s challenges in the last two years. It analyzes the research by the geographical locations they belong to, domains, influential authors, organizations, and funding agencies. The material was obtained from the data from 2020 to 2022 from extensive Scopus databases. “Bibliometric analysis” is referred to as the analytical statistics of content that is widely published as articles, papers presented in renowned conferences, and reviews for a better understanding of the impact that the particular publication has had in the research domain all over the world. Further, for visualization analysis, freely available tools like “GPS Visualizer” are used alongside “Gephi,” “VOS viewer,” “ScienceScape,” and word cloud as well. The visualization enables an easier comprehension of holistic perspectives.

Digital learning is a significant part of today’s world. Due to the accessibility of high-speed Internet connectivity, users can acquire information on vast concepts known to humankind through the web on their laptops, desktops, or smartphones at any time or location. Digital-based learning is also more accessible for people with disabilities and can provide them with equal access. For this purpose, the Higher Education Government of India initiated Digital-based learning, which aims to enhance students’ soft skills, provide e-resources and promote technology awareness. This paper discusses the overview of digital learning, digital resources, and the Indian Government’s Higher Education Digital Learning initiative and its benefits for students, research scholars, and educational institutions. This kind of work showcasing different digital learning platforms and their aids is not done earlier. This paper provides information about the resources and the digital platforms and their uses, which would be useful for the trainers and the students.

Autonomous vehicles will fundamentally change the future mobility of people and goods. However, there are many factors that raise the concerns of the residents of the United Arab Emirates towards adopting autonomous vehicles. The distrust of the people is based on factors such as safety, trust, privacy, accessibility, ethics, and other inherent concerns that are difficult to quantify and model. In this paper, we have captured data using an online survey designed for residents of the UAE in order to understand their concerns and the corresponding factors. We address the adoption of AV in the UAE as a classification problem, in which machine learning algorithms are applied to classify the participants of the survey into categories/classes representing their willingness to purchase an AV. The applied machine learning techniques include: Ada-boost, KNN, Neural Networks, SVM, Decision Trees, Random Forest, Naïve Bayes, and Gradient Boosting. We provide a comparison between the applied machine learning classifiers to come up with the best classifier model, which will represent the prediction model for the adoption of future mobility technologies for the UAE public. Our results demonstrate the potential of machine learning to accurately forecast AV adoption in the UAE based on user data.

The technological advances converge with digital transformation in order to improve our everyday life. Cultural heritage is one of the application areas that may take advantage of this progress, for example, museums, bringing several new experiences to their visitors. However, several digital solutions return static and simple data and contents that are often uninteresting to the visitors nor promote their integration with a museum visit. This paper proposes an augmented reality framework, as a multimedia content and information transmission solution, in the indoor environment of a museum, focusing on a dynamic approach, which increases the immersion of the visitor into the visit storyline. The framework architecture was developed as a solution to a real problem, stated at the Foz Côa museum, in Portugal. An augmented reality contextualization was carried out on that museum, as well as a comparative study between two augmented reality tools, in order to determine the most appropriate to use. Also, the accuracy of the developed framework was tested, under different conditions, in the real museum environment, as an experimental result of this study

Many industries use recommendation systems (RS) to identify product recommendations when users actively participate on e-commerce sites. Recently, massive growth in both goods and consumers has faced serious challenges. Numerous websites present the consumer with numerous options at once, creating a lot of confusion. In addition, finding the right product or active user is an essential part of RS. Products are already recommended based on consumer preferences and sociodemographic trends. A hybrid action-related recommendation based on K-Nearest Neighbor Similarity (HAR-KNN) combines the ease of hybrid filtering with the development of feature vectors to improve the user behavior matrix. To categorize attributes, it uses both quality and quantity classifiers. Additionally, the proposed methodology overcomes shortcomings in earlier approaches to evaluating user preference for goods and feature analysis. The SOM AND KNN classification technique has been approved for the purpose of locating information about user behavior online and in real time for a specific user group containing a large amount of data in relation to the commonalities among many users and target users. A test result is evaluated by using highly predictive metrics such as precision (P), recall (R), F, as well as mean absolute error (MAE), mean square error (MSE), and root mean square error (RMSE).

Short Message Service (SMS) acts as one of the media of communication for the mobile device system due to rapid increase of mobile users. During disasters, the government provides rescue agencies for the people to deal with emergencies. Since there exist critical positions during the time of disasters, the rescue agencies cannot categorize the messages based on emergency. So, a hybrid system of the SMS spam filtering method is used to classify messages like spam, ham, alert1, alert2 and alert3 using machine learning algorithms such as Naive Bayesian Algorithm and Decision Tree Algorithm. Since it is a hybrid system significant messages are considered effectively by the rescue agencies and help the victims.

Lung cancer is one of the biggest threats to mankind. The number of patients who died from lung cancer is too large compared to the total number of cancer diagnoses. Lung cancer is uncontrollable cell proliferation in the lungs and can be recognized as a nodule, which might be benign or malignant. A nodule is a white-colored area on the lungs that can be seen on an X-ray or CT scan image. With the advancement of technology, many interdisciplinary domains are working together. Technologies such as machine learning (ML) have greatly assisted in lung cancer diagnosis using an imaging modality. An X-ray and/or CT scan image is used as the input for ML techniques. It is processed using image processing techniques, and the findings are then classified using ML algorithms that are presented in this study through a pipeline of ML. According to the specifications of the ML pipeline, several intermediate processes such as image preprocessing, lung segmentation and enhancement, nodule detection, feature extraction, and classification were also briefly described with their techniques. The entire study investigates various ML approaches used in lung cancer diagnosis using X-ray and CT scan images.

As malaria is a fatal disease that can occur anywhere, prompt diagnosis is crucial to stopping the disease in its tracks and reducing its overall impact. To better foresee future malaria epidemics, a hybrid machine learning model was created in this research. The model performance can be improved using various strategies, such as ensemble methods or fine-tuning the hyperparameters. Choosing a proper ensemble technique impacts the model accuracy. The performance of the ensemble model was measured using several well-known machines learning algorithms, such as Decision Tree, Support Vector Machine, Naïve Bayes, K-Nearest Neighbors, and Random Forest. This methodology’s stacking strategy of ensemble technique allowed the integration of five separate algorithms. Compared to other machine learning classifiers, the results obtained using this ensemble method is superior with accuracy 92.8%, AUC 91%, recall 100%, precision 85%, F1-score 91%, specificity 100%, macro-average 94%, weighted average 98%, and error rate 5.4%. The relative importance of each variable and the degree of connection between them in explaining malaria prevalence are calculated. The results indicate that the hybrid approach is valuable for anticipating malaria outbreaks.

Yoga, in the Western culture, is considered a form of posture-based physical activity which helps relieve stress and relax your muscles while increasing flexibility. On the other hand, traditional yoga is focused on meditation and released from worldly attachments. It was first brought up in Rigveda with references in the Upanishads. Yoga originated in India, more than 5000 years ago. While some poses in yoga are simple to understand and perform, some poses require precision in the angle at which your body stays to avoid injuries. In this paper, we propose a system to detect the yoga pose of an individual and help them with the correct pose, if wrong. There are several key points detection algorithms that can be used such as OpenPose, PoseNet, and PifPaf. The key points extracted from the video captured are passed to the system, to calculate the angles made by several joints. These angles are used to check whether the posture is correct or not.

The current COVID-19 disease outbreak has been quite difficult and challenging for human society. Early diagnosis of the virus in people and quarantining them are now vital due to the virus’ quick spread around the globe. Currently, the most widely used method for testing for COVID-19 is RT-PCR. Though it is widely been used, its accuracy is not as desired. From CXR images, we have proposed using neural networks to forecast a patient’s COVID-19 infection status. The COVIDX CXR dataset has been used to train our model. To use the model with ease for the general public, we have developed a web application using Flask for the backend and HTML, CSS, and JavaScript. Using this web application, the user can get the COVID report in a few seconds.

In this modern era, with online shopping on the rise, many e-commerce websites and a myriad of different online websites have increased online payment methods and modes, unfortunately increasing the risk of credit card fraud. Hundreds of thousands of people fall victim, each year. This will be susceptible to increase. To avoid massive losses, financial institutions must strengthen their detection and privacy measures. This paper aims to provide a novel approach to recognizing credit card fraud transactions by using machine learning techniques like kNN, decision tree classifier, Random Forest regressor, and further training datasets with AdaBoost and Bagging classifier which increases the accuracy for identifying the fraudulent transaction. Experimental analysis of the proposed approach and findings are based on a legitimate dataset, containing information on European cardholders.

A common opinion of stock market in society is that the stock market is either insecure to invest in or troublesome to trade, so many people are disinterested. The stock market is a marketplace that facilitates the acquisition and sale of business stock. The stock index has its unique value on each stock exchange. The index is the average value calculated by aggregating the prices of several stocks. The forecast of the entire stock market is time-varying and depends on the stock price movement. The seasonal variance and steady flow of any index helps both professional as well as amateur investors understand and decide whether to invest in shares and the stock market. Individuals and businesses alike can be affected significantly by the stock market. As a consequence, accurately anticipating stock movements can lower the risk of losing money while increasing profits. To address these issues, time series analysis will be the most effective tool for predicting the trend or even the future. This article represents the comparison of LSTM, ARIMA and SARIMAX models for stock price prediction. It is observed that error for ARIMA model is less as compared to SARIMAX model.

The thyroid gland, a butterfly-shaped nodule in the front of the neck, can develop thyroid carcinoma. Compared to other carcinomas, thyroid carcinoma is one of the most common endocrine carcinomas. Thyroid carcinoma may not create symptoms at first, but it can cause soreness and enlargement in the forward facing of the neck as it spreads. In this study, first thyroid disease is identified and later-on on the basis of symptoms such as Neck Swelling, TSH, Tumor, Neck Pain, Anxiety Thyroid Carcinoma is identified. Over the past few decades, patients with thyroid disease and carcinoma have been on the rise. Thyroid disease is the second most prevalent malignancy; the majority of instances in the general population is benign and malignant in thyroid cancer. As clinical detection of thyroid cells necessitates the utilization of various features in a variety of scales, a traditional method of extraction of features may not provide adequate results. The goal of this paper is to use ant colony optimization algorithm to reduce the number of features scores and to predict accuracy on thyroid carcinoma from the symptoms of thyroid diseases.

To investigate the feasibility of determining tropical cyclone (TC) intensity using satellite photos, a deep learning convolutional neural network system is employed. This paper also demonstrates how to determine a cyclone’s strength using an image processing method. To identify cyclone components, a convolutional neural network-based method is employed. Predicting and determining the intensity of flash floods are essential to lowering the number of fatalities and property damage caused by tropical cyclones (TC). This study describes a method for evaluating TC strength from satellite images. The convolutional neural network model is utilised for cyclone detection and characterisation, whilst AlexNet is used to extract features, build models, and forecast cyclone conditions. The findings imply that our model qualitatively developed a perspective resembling that of subject-matter specialists. Additionally, we use the convolutional block attention module (CBAM), which is based on 3D, to simulate visual attention in order to improve the model’s focus on the crucial channels and primary cloud structure. According to the analysis from the study, the suggested model’s root mean square error (RMSE) is 9.5 kts, which is lESS than both the classic deep learning (DL) method of intensity estimate and the advanced Dvorak technique (ADT) by 9.3% and 27%, respectively.

Educational data mining and learning analytics are contemporary research disciplines that can provide interesting and hidden insights into the effectiveness of different learning styles, course complexity, learning content difficulties, and learning design issues. However, these two emerging research disciplines do not deal with the initial phases of data ingestion, preparation, and transformation, because the researchers often expect data to be available, grouped, and cleaned. Therefore, we aim to explore the possibilities of big data processing in education from the data engineering point of view. Further, we analyse a referenced data infrastructure model and discuss its appropriateness for developing an ML platform for learning analytics and educational data mining research at the university. As a result, we propose the ML platform for learning analytics research and emphasise the importance of suitable data infrastructure selection, as well as the impact of the individual steps of the data engineering life cycle, on the quality of the learning analytics model.

The majority of critical injuries caused by road accidents occur in developing nations, where they are a major global concern. Due to these traffic incidents, many individuals have lost loved ones. Consequently, a system that has the potential to save lives is needed. The approach identifies key contributing factors for accidents or establishes a connection between accidents and different causes of accidents. This study suggests a system for predicting accidents that can assist in analyzing potential safety concerns and foretelling whether an event will happen or not. In order to identify key qualities and forecast accident risk in single- and multiple-crashes, this research provides a hybrid feature selection-based classification-based machine learning (ML) method. It took into account classifiers like the random forest, k-nearest neighbor, and XGBoost. A few evaluation criteria, including precision, root mean square error, accuracy, recalls, and receivers operating output characteristics, were used to assess these classifiers. When the accuracy of the ML algorithms was compared, the accuracy of the proposed model attained as 99.78%.

Intelligent reflecting surfaces (IRSs) are gaining interest due of their potential coverage and spectral efficiency advantages. However, there are a number of issues that must be resolved in order to materialise these surfaces in practise. A collaborative design for the BS and IRS beamformers is necessary, and the IRS components must be easily modified accordingly. Channel information is crucial to the accuracy of beamformer design deep learning (DL) is a data-driven technique that is essential in overcoming these difficulties. Deep learning is resistant to data flaws and environmental changes because of its faster computation speeds and model-free structure. DL has been demonstrated to be effective at the physical layer for active/passive beamforming, channel estimation, and IRS signal recognition utilising supervised, unsupervised, and reinforcement learning architectures. The methods for creating deep learning-based IRS-assisted wireless systems are summarised in this article. It gives insight to the different deep learning techniques like reinforcement learning, supervised learning, and federated learning which could be used along with IRS.

Prediction of invoice payment date through machine learning in the B2B marketing is an important factor that affect the business dealings between the companies and changes the complete direction of the business. If the seller company finds the predicated payment date of the buyer company is getting highly delayed from the due date then seller company may not sell that product to that company. So, in this way, payment date prediction plays a very important role in B2B marketing. In this study, we explore how machine learning (ML) can be used to develop models for predicting whether newly created bills will be paid, enabling customized collection activities specific to each invoice or customer. Our models can accurately forecast whether or not a bill will be paid on time and also give estimates of how much time will be lost. Our methods are demonstrated using real-world transaction data from several firms. Finally, simulation results compared with other state-of-the-art approaches.

Many industries have now adopted the predictive maintenance approach to increase the life span of the equipment. So as a part of predictive maintenance, early estimation of faults in the motor is a pivotal component as it helps in many ways, such as minimizing time, cost, and efforts for equipment maintenance. However, it is challenging to find the data on the faulty condition of the electric motor, so based on hypothetical points of faulty conditions are added. Data is used to monitor conditions generated by different sensors deployed on an electric motor. This paper mainly focuses on early fault detection for rotating machinery onboard ships’ motors using two other methods of fuzzy logic and K-means. The fuzzy technique uses a rule-based approach, and K-means uses an unsupervised approach for the early detection of faults in the motor. In the proposed work, the fuzzy logic technique takes vibration and voltage as inputs to the fuzzy inference system. And, K-means takes input features extracted by the feature extraction technique to monitor the motor condition. Experiments show that fuzzy logic can readily determine the condition of the electric motor with only two inputs, and K-means can easily differentiate clusters of motors under different conditions with the help of features extracted using PCA. The proposed work has identified the conditions of the motor, such as good, acceptable, monitored closely, and unacceptable, which helps in early fault detection.

One of the AI applications that has attracted the most attention from researchers around the world is the recommendation system. Since the beginning of the Internet era, recommendation systems have been widely incorporated into our daily lives. It is still difficult to give new users the proper recommendations. This paper suggests a novel and sophisticated movie recommender model to address this problem and to suggest highly rated movies to the new users. Moreover, we have applied a number of supervised machine learning techniques, including as KNN, DT, RF, GNB, and LSVM, to evaluate the performance of the proposed model in terms of precision, recall, F1-score, and accuracy. In simulation, all the techniques are performing well on the proposed model and on the given datasets, however, the LSVM and GNB is proving higher accuracy along with precision, recall, and F1 measure.

Stock price prediction is a vital part of the life of any person involved in the financial market. Predicting trends is a topic that is increasingly gaining popularity among researchers, investors, and traders alike. However, numerous external factors affect the volatility of stock prices, thereby making the prediction extremely complicated. Being able to make accurate predictions is of paramount importance to maximize the profitability of trading in stocks. Machine learning algorithms can train and improve their performance individually and autonomously. This characteristic is beneficial in the financial domain, where large amounts of historical data must be understood. There are some underlying patterns in the historical data that humans may be unable to identify, and this is where machine learning models become very relevant. This paper provides a comparative analysis between three machine learning algorithms—linear regression, support vector machines (SVM), and random forest as tools to perform predictions on stock prices. These algorithms all use a set of eleven technical indicators as input features to the models and perform technical analysis to make the predictions. The paper also provides the result of incorporating sentiment analysis into the prediction models.

Change is constant and reforms in systems are cyclic in nature, every domain is subject to the sequential time interval-based cyclic process, revolutions happen as the need arises resulting in changes in design, thought process, technology, algorithms, infrastructure, and applications. Every industrial revolution may have given global education patterns a new dimension. Primary, junior, middle, and senior grades, as well as undergraduate, postgraduate, and doctoral degrees, are the several levels of education offered in India. Students generally are less aware about domains and scope each program offers at degree level, while opting a particular program, parameters like specializations, scholarships, faculty resource, minimum eligibility criteria, placement records, and technical assistance. Work employs opinion mining-assisted mechanism to address the above-mentioned issue. For the problem discussed, a self-dependent automated opinion mining strategy is required to resolve this issue. The paper presents an attempt to find a fuzzy-supported pattern that will help the students choose the best program alternative for their better improvement. The study suggests a data analytics methodology for the student’s educated, intelligent decision-making while selecting the best program. A candidate should consider a variety of weighted criteria while choosing a certain domain as they contribute to the selection process. A SWARA-assisted approach has been employed to assess the viability of the presented method and to determine weights of criteria depending on experts’ choices. Data from a college of education, Bilaspur, Greater Noida, was utilized in the research project. An empirical case study was carried out on data set of 800 students selected from diversified courses like B.Ed., BCA, M.Ed., B.Sc. (Phy, Chem. Bio), LLB, BA, and MA. The organization focusses specifically on graduate courses after 10 + 2, and the data was acquired employing a questionnaire. Database used was heterogeneous and was optimum for the analysis owing to the diversity. The degree of utility for the proposed analysis was estimated (θi), and the priority order $$({\mathcal{S}}^{*})$$ ( S ∗ ) was calculated. The degree of utility for ɸ2 was 97.4%, ɸ3 was 91.2%, and ɸ1 was 83.2%. It was determined using approximated values that aspirants preferred the parameters for opting program in the following order ɸ2 > ɸ3 > ɸ1 > ɸ5 > ɸ6 > ɸ4.

Adverse weather conditions like rain, fog and storm cause degradation in the quality of an image. Computer vision operations such as detection, classification and various monitoring of objects activities are adversely affected due to image degradation. Hence, image enhancement is an important pre-processing step. We propose a method for the elimination of rain streaks from rain-affected images by making use of conditional generative adversarial networks. We have used structural similarity index measure and peak signal-to-noise ratio as evaluation metrics evaluate the model. We have tested our model on one synthetic and four real-world datasets and compared the performance with other state-of-the-art methods. We have obtained a generated image that has a close resemblance with the ground truth images.

Content-based video retrieval is one of the foremost research problems in the fields of artificial intelligence, computer vision, digital image processing, and natural language understanding over the past decade. The exponential development of video retrieval has created several research problems and opportunities for the design and develop a variety of real-world applications such as face video retrieval/recognition, video tagging, video annotation, crime investigation, Web albums, and health care. Most of the research is to develop efficient content-based video retrieval techniques from large video databases. Even though a large number of video retrieval techniques have been increased, there are no universally accepted techniques available. The challenge in content-based video retrieval is to build a rapid solution for high-dimensional indexing, which is essential for constructing large-scale content-based video retrieval systems. This study presents an emphasis on the role of content-based video indexing and also highlights the challenges in designing content-based video retrieval and recognition systems.

Alzheimer’s disease is a neurological condition that gradually reduces brain size and destroys brain neurons. The most common varieties of dementia, Alzheimer’s disease limits a person’s ability to operate independently and is described by a steady deterioration in mental, cognitive and social abilities. Due to the severity of moderate cognitive impairment, AD diagnosis is frequently challenging at an early point. Nevertheless, treatment is likely to be successful at this stage. This raised concerns about the early diagnosis and treatment of AD. Deep learning approaches are used, including fastai and InceptionV3 to identify the most accurate markers for predicting AD.

Steganography is the art of hiding secret messages within a non-secret object. It helps hide information in images such as graphics interchange format (GIF) images, text, audio, and video. In GIFs, animated GIFs are a common form of media. GIF images are mainly classified into two types: GIF87a and GIF89a. GIF87a is a type of palette-based image, and GIF89a is the improved version of GIF87a because the GIF89a image has more than one frame. Furthermore, it allows the GIF image to play simple animations. In addition, message hiding in palette-based graphics can be accomplished in two significant ways: firstly, by embedding messages into the frame, and secondly, by embedding messages into the palette. The primary goal of this paper is to review and describe numerous methodologies for hiding data within the GIF images and make a detailed comparison of these methods.

Whenever any data is embedded inside any image, then there is always some distortion caused, which caught the attention of attackers. As a consequence of this, there is always a need to reach a balance between the image quality and the embedding capacity of the picture. The researchers working in the field of data hiding have two basic objectives: one is to boost the performance of the suggested algorithms, and the other is to give a better capacity for embedding. This survey paper focuses on various traditional interpolation methods and interpolation-based data hiding techniques for the past few years. Also, the goal of this survey is to offer an in-depth evaluation of current up-to-date approaches, as well as their benefits and drawbacks if any. This survey paper concludes with some research recommendations that can be used in the future to enhance both embedding capacity as well as image visual quality.

Reversible data hiding (RDH) is one of the most crucial issues in the field of information security in the modern world, as it enables us for securing communication by hiding secret data in a cover media. With the help of this, it also lets us extract the secret data without losing any original data at the receiver side, i.e., called as lossless recovery. RDH is very suitable for medical images for example MRI scans, CT scans, and in military applications. In the past few years, several RDH approaches have been developed, for example, difference expansion, histogram shifting, prediction error expansion, and interpolation. This review paper focuses on different reversible data-hiding strategies that are basically based on interpolation techniques in order to get a good visual image quality and embedding capacity, which have been attempted by many researchers to improve it at the same time.

One strategy to prevent COVID-19 as a result of the pandemic is to use a face mask. Although many countries have made it a requirement for citizens to do so, the majority of people continue to disobey this order. In the current situation, police frequently check for face masks in public locations and fine anyone found not to be wearing one. The existing system identifies the person wearing the mask or not, but our proposed method identifies the person those who are not wearing the mask properly and also identifies the multiple person wearing the mask or not. On the other side, some governments have implemented technology to identify individuals wearing face masks and communicate their information to a patrol team so that they may catch them. The proposed model identifies individuals who are in the public without face masks. The proposed approach is able to identify these persons using facial detection technologies, and the data is then combined with a database of public identity information to gather information about the individual and deliver the fine amount to his home address and mobile numbers. Using the convolution neural network (CNN) model, we have identified people wearing and not wearing masks. When compared to many other algorithms, CNN can more precisely recognize data down to the pixel level. In the implementation of our model, the activation functions for the hidden and fully connected layers, respectively, were rectified linear unit (ReLU) and softmax. Two convolution layers with 100 filters each were used. The 91.21% accurate Cascade classifier is used to recognize faces. Adam is the optimizer, and cross-entropy serves as the loss function. More than 1500 images were used to train the model, which comprises classes with and without masks. The public will start wearing masks in public areas because of the terror that this AI-based mask recognition system instils in them, helping to stop the spread of diseases that are otherwise beneficial to society.

In the era of upgrowing technologies with automation, building a smart city is the need of the country. Goal of smart city is to develop smart technologies that can be helpful to the day-to-day routine life. At that time, surveillance plays an important part in development of smart city. In surveillance systems, vehicle surveillance is very needy and important for intelligent transportation system. To develop a system for vehicle monitoring, vehicle speed measurement is very useful parameter that can be useful to detect over speed in speed limit area or detection of possibly accident due to over speed. To reduce the cost of Speed gun or sensors, computer vision technologies are very useful to develop a system for vehicle speed measurement that takes CCTV footage as input. To calculate a speed efficiently, tracking of vehicle from video is crucial part. The paper proposes a system for tracking a vehicle based on optical flow and Shi-Tomasi corner detection. Additionally, the edge-detected frame after thresholding is applied for Shi-Tomasi corner detection that derives a novel approach to track a vehicle precisely.

Mammography is used as a primary method for the X-ray imaging of breast regions. These mammogram images are utilized by radiologists for localization and prognosis of the mass regions present in the breast region, as either malignant or benign. We present a similar approach in our work, where a two-stage system is proposed for the localization and classification of breast mass regions using the mammogram images. First, these mammograms are passed through an image processing pipeline for the initial processing. Secondly, these processed images are fed into the proposed two-stage system for segmentation and classification. For the segmentation stage, we use the UNET segmentation architecture with EfficientNetB0, ResNet50, and MobileNet encoders without any pre-trained weights. For the classification stage, we use the VGG16 and ResNet50 pre-trained models for our task where we feed in the segmented region of interest of tumors as input and the output of the model being the pathology of the tumor. The results obtained show good accuracy in determining the pathology of the mass region in the mammogram images, with results obtained at low latency with good precision, recall, specificity, and sensitivity rates.

The early determination of skin sore illnesses is exceptionally challenging for individuals living in rustic regions because of inaccessibility of qualified dermatologists. In this situation, the dermatologists can analyze skin sore illnesses via cautiously looking high quality at anyplace. Further, the AI-based programmed analytic framework might help essential wellbeing experts for fast and precise detection of certain skin disorders. Consequently, there is a requirement of artificial intelligence-based medical image processing and examination of skin lesion images to enhancing their perceptibility characteristics. The use of image processing in biomedicine for diagnostic purposes is a non-invasive method. The potential for automatic image analysis approaches that offering quantifiable data on a lesion that can be utilized clinically and as a standalone early warning tool is pretty high. High quality digital photographs of melanoma skin lesions have been researched as a potential early skin cancer detection method without the need for skin biopsies. A strong artificial intelligence-based software application for skin lesion identification and detection will provide a better classification scheme and maybe enhance the automatic diagnosis of skin lesions. In this review article, we have analyzed and reviewed various malignant and non-malignant skin lesion detection techniques for human skin image.

This analyst is particularly intrigued by multi-input (MI) converter framework as opposed to individual converters for sustainable sources. The singular converter brings about a huge volume of the gadget in the framework and high use. The specialist planned another novel multi-input KY boost converter for environmentally friendly power with a high-power factor and less distortion. KY boost converters with multi-input data sources have been demonstrated to be the most steady, time-domain, and ripple free. These hybrid frameworks pick the DC-coupled framework. Sun and wind are nonlinear sources due to their climatic circumstances. This incorporated framework is shrewdly controlled under the management of FLC. The PWM of multi-input KY converter switches is changed utilizing FLC. The MPPT utilizes fuzzy to expand the productivity, intermingling speed, quick reaction, and supply the continuous power given to the load. When there is no supply from the input side, the DC bus maintains a constant DC bus voltage, and the storage device manages the power. A dynamic way of behaving is tried and checked with the assistance of MATLAB recreation. The entire framework is rebuilt with the assistance of MATLAB/SIMULINK. Energy management is carried out with the assistance of a hybrid renewable energy sources, PV, and wind system, which has been integrated with an energy storage system.

Data security is the process of safeguarding digital information, such as that contained in a database, from malicious entities and uninvited human behaviour, such as a cyberattack or a data breach. Since most medical records are currently housed on an antiquated digital file management system, malicious actions have been committed frequently due to our government's lack of interest in the healthcare industry, putting patients’ health at risk and causing financial loss. Hospitals and the insurance industry are at great danger of financial loss as a result of these frauds. In this paper, we discuss a framework that uses a method that primarily verifies the records to guard against malicious activity. If malicious activity has so far taken place, it will be detected using machine learning (ML) technology, maintaining the confidentiality, integrity, and availability (CIA) of the EHR. The primary goal of our framework is to provide a secure environment for EHR and give the investigators leads that they can follow up on and perhaps recover from, recuperate from, or refer to the proper authorities or agencies.

Cloud computing is a technique that connects a large number of systems with either public or private network, providing dynamic scalability for data storage, building applications, and other services. The consumers are charged only for the resources they consume. Cloud computing provides enormous applications and benefits in the IT field, but all the facilities need to be completely secured and protected. Removing security risks and maintaining privacy are essential for securing cloud computing. Several encryption algorithms are used for encrypting data in the cloud. Homomorphic encryption is an advantageous technique in cloud computing because it allows the cloud service providers to work on encrypted data. Thus, it ensures the privacy of sensitive data and denies access to data by cloud providers. Nevertheless, this homomorphic encryption technique has a problem with key sharing and management. Hence, to provide efficient key management, the technique of particle swarm optimization is used along with homomorphic encryption. Data to be encrypted is taken as input and treated as the initial population. PSO generates the key for encryption; then, the data is encrypted using homomorphic encryption. We have used MATLAB to implement this proposed encryption algorithm. Performance is analyzed on the basis of the time taken to execute and the utilization of resources. The observations made suggest that the enhanced algorithm executes well and provides better results as compared to the existing algorithm. It shows less time taken to execute and requirement of less number of resources. Hence, the proposed scheme ensures cloud security efficiently.

The Internet of Things (IoT) in health care has remained elusive, mostly because of the threat of unwanted access to protected health care because of a weak wireless channel. Furthermore, administrators cannot employ sophisticated and resource-intensive security measures on IoT devices due to their limited computing and storage capabilities. In addition to this, the ad-hoc nature of such networks makes them even more vulnerable to security threats. To address these issues, our research proposes an authentication protocol that uses a minimally intrusive data security approach for communicating data in real time. In the proposal, a secure authentication and key agreement mechanism for the Internet of Medical Things (IoMT) is presented. The proposed protocol employs zero knowledge proof (ZKP), Ayushman Bharat Health Account (ABHA) number, biometrics, symmetric cryptography, message digest, and other techniques to achieve the protocol’s goal with low storage, computing, and communication costs. The proposed protocol is evaluated using the Automated Validation of Internet Security Protocols and Applications (AVISPA) tool, and it is observed that it protects data anonymity, confidentiality, integrity, and safety from major cyber threats.

In India, the crime against women is increasing at a startling rate. The National Commission for Women reported that there was a 46% rise in crimes against women in the beginning of the year 2021 as compared to 2020. This pressing issue needs to be addressed and in order to do that crime data needs to be collected so as to understand and analyze the crime rate and make predictions by using the existing information so as to take some proactive measures to reduce the rate of crime against women in future. Machine learning models become an effective mechanism to predict the rate of crime in India for a variety of crime categories by doing analysis of patterns followed in a particular crime or a category, criminal-centric areas, and the comparative study of all those categories of crime. A dataset has been created by collecting data from Kaggle and NCRB, the government website of crime data collection. Various machine learning algorithms have been applied on the data after preprocessing, and performance of all these models is evaluated and compared using various metrics.

This paper proposes an image forgery localization algorithm based on extracting texture features from fixed size image blocks. The computation complexity of the block-based methods depends on the number of blocks, which in turn depends on the image size. Higher the block number more is the complexity. Thus, in an attempt to improve the complexity, this paper proposes a technique based on discrete wavelet transform (DWT) of image. The single level DWT of image yields four sub-band images of which the LL sub-band is considered for further processing in place of the original image, as it carries the maximum spatial information. While the size of the LL band is one fourth the size of the original image, the proposed algorithm when applied to the LL band image yields reasonable results.

In the current modern world of computation and technology, the Internet plays a very important role. Majority of the things on the Internet work on the concept of centralized networks which offer good accessibility and security along with innovations. But, centralized networks are vulnerable to a fair share of cyberattacks and security issues. Decentralized networks form the basis of blockchain technology. Blockchain offers a simple solution to ever-increasing cyber and security attacks on networks by removing the concept of a sole point of failure from networks. The advantages of decentralized networks are that it keeps more information from becoming a separate process. As a result, it is quite difficult for malicious hackers to steal or misuse information from the network. The work presented is an effort to secure digital currency transactions for online auction, and application developed is a secured decentralized platform using blockchain. Model comprises DApp development, a Non-Fungible Tokens (NFTs) Marketplace which is a platform to acquire NFTs, sell NFTs, display NFT status, store them, and even mint new ones. Work done also employs technologies like IPFS, hardhat, Ethereum smart contract functionalities, Next JS and other technologies for implementation of DApp. Application developed provides an extremely secured environment with non-fungible tokens’ (NFTs) for implementing online auction and exchange of digital currency with decentralized blockchain at the backend. Application was developed on Solsea platform, virtually deployed, and tested with 22 bidder’s input. On auction time completion, prototype provides the address of the leading bidder and in addition also provides the margin by which the leading bidder claims the bid. Platform designed is a solution that is secure, reliable, and transparent technique for online auctions equipped with multipurpose smart contract that can address to the broader requirements of auctioneers and bidders.

Recently, a paradigm shift is observed toward Industry 5.0, where tasks (processes) are automated at massive scales. This shift has initiated modern developments in artificial intelligence (AI) to support a plethora of applications like manufacturing, health care, vehicular net- works, and others. However, owing to the black-box nature of AI models, the research has shifted toward the proposal of novel techniques that aim toward the explainability and validity of these AI models. Thus, explainable AI (XAI) has become a norm in modern applicative domains, and the study of its frameworks and tools has become the buzzword among researchers. Thus, the paper intends to present the key concepts of XAI and aims at improving the model transparency. The survey systematically untangles the key concepts of XAI and presents a solution taxonomy in different applications. Modern XAI techniques are classified as self-explanatory, visual-based-model-agnostic, global surrogate, and local surrogate-model-agnostic. We also cover the tools and frameworks of XAI and discuss the open issues and challenges in practical realization. Thus, the survey intends to arm AI practitioners to design optimal solutions to realize XAI in practical use-case setups.

Recently, with the massive exchange of data over Internet of Things (IoT) ecosystems, attacks surfaces have also intensified. In IoT, connected devices share data over open channels and thus are highly vulnerable to security and privacy attacks. Botnet-based attacks have been found to have a significant effect on the network-based system. Thus, in this paper, we present a scheme AutoBots, which differentiates the normal and anomaly behaviour of IoT devices among the connected network. To exploit this, we consider diverse parameters like network behaviour profiles and apply autoencoders to classify and detect anomalous traffic from normal traffic. We used the BASHLITE and MIRAI IoT botnet setup and trained our network with the N-BaIoT dataset that has both benign and malicious network traffic. We compared our scheme for metrics like attack detection time, attack detection with respect to hourly traffic, deep residual accuracy, and residual loss. The presented results signify the efficacy of the proposed scheme against conventional bot-detection schemes.

Keeping in mind to preserve our environment from harmful gases emitted by the vehicle, which leads to increase in green-house effect. Researchers have found out the way to design a vehicle which is free from contamination and zero emission, which makes it eco-friendly to the environment. Due to this concern, leads to the development of a fuel cell electric vehicles which is powered through hydrogen. In this, fuel cell (hydrogen) acts as a main power source, which generates electricity generally using oxygen from air and compressed hydrogen. This electricity is then supplied to an electric motor which drives the wheel and have an electric battery as a secondary power source which is used to store the excess energy during braking and assist driving. A vehicle component is modelled in MATLAB/SIMULINK software, and its performance of vehicle is verified by ADVISOR GUI model.

The world that we live in today is flooded with information from sources that sometimes are true and sometimes are not. Just a simple change of narrative on the same thing might drastically change people’s opinions. With fabricated media already rampant, a new player has emerged in the synthetic media industry which goes by the name of “deepfake” which is a type of fabricated video made using deep learning, accurate to such an extent that it is sometimes impossible to differentiate the real video from the fake. The paper evaluates information from various sources and then expresses in a non-complex manner, the technical details of deepfakes, it’s formal definitions, an overview of the impact that it has had on the world till now, the technical process of making it also explaining the technical concepts in the process, and lastly the chapter talks about how some technical indicators that we can notice to spot deepfake videos, their place in today’s world and their legality is also talked about. This paper serves as the review of the technical details that we have about deepfake vides and then expresses the essential technical details about it in an organized, articulate, and an easy-to-understand manner. For people interested in making their own deepfakes, the chapter serves as an efficient starting point, by explaining the procedure in a simple yet technical way, and by explaining the tools that are required to make deepfake videos and how to use them. The paper might also serve as a way of spreading awareness about deepfakes and what the average person can do from his or her side to identify deepfake videos.

Long-term research has been done on anomaly identification. Its uses in the banking industry have made it easier to spot questionable hacker activity. However, it is more difficult to trick financial systems due to innovations in the financial sector like blockchain and artificial intelligence. Despite these technical developments, there have nevertheless been several instances of fraud. To address the anomaly detection issue, a variety of artificial intelligence algorithms have been put forth; while some findings seem to be remarkably encouraging, no clear winner has emerged. In order to identify fraudulent transactions, this article presented Inspection-L architecture based on graph neural network (GNN) with self-supervised deep graph infomax (DGI) and graph isomorphism network (GIN), with supervised knowledge methods, such as random forest (RF). The potential of self-supervised GNN in Bitcoin unlawful transaction detection has been demonstrated by the evaluation of the proposed technique on the Elliptic dataset. Results from experiments reveal that our approach outperforms existing standard methods for detecting anomalous events.