Data Science and Network Engineering

In this digital era, people are cheated in multiple ways by sending fake messages. Without realizing its impact, they respond to the links the cyber frauds share. This immediate reaction to the fraud messages makes people lose their balance in bank accounts or fall into some other horrible events. These types of fake or spam messages have to be identified earlier before they come to users’ Inbox. This paper proposes a Spam message filtering model that extracts significant hand-crafted features and is classified using machine learning algorithms. This research collects 7700 short messages in Dravidian languages like Tamil, Kannada, Telugu, and Malayalam and creates an optimal Spam-Ham filtering framework. Experimentation has also been carried out with a benchmark dataset for performance comparison regarding accuracy, precision, recall, and F1-score.

In this paper, we focus on how a probabilistic mixture of normalizing flows can be fitted. In the literature, there are (at least) four approaches that do not necessarily provide an actual implementation of the method. These four algorithms are gradient ascent maximizing the log-likelihood of the data, (soft) expectation–maximization, hard expectation–maximization, and gradient ascent maximizing the evidence lower bound. Our contribution or the novelty of the paper can be described as follows: we (re)implement each method, we create a software program that encompasses all these four implementations, and we compare those on toy datasets and image datasets on which we fit a mixture of masked autoregressive flows. The non-linear flexibility is shown in the plots. The metrics and the running times are reported. There is not necessarily a certain training algorithm to be preferred, although there are some advantages and disadvantages for each. The code is available at https://github.com/aciobanusebi/training-algs-for-mnf .

A person's mood can be changed by music, which impacts both the body and emotions, so our project aims to describe a method for music recommendation based on facial expressions. Many music applications are based on a user’s browsing history but this work presents an idea on the music recommendation based on users’ facial expressions that give outstanding results. It has got a unique ability to lift one’s mood. The work’s overall concept is recognizing facial expressions and efficiently recommending songs. The proposed model will be both time and cost-efficient. This yields better performance and computational time accuracy and reduces the design cost.

Sales forecasting is a decisive task for businesses, as it enables them to make important decisions about production, inventory, and marketing strategies. Time series analysis is a tackle for sales forecasting, as it allows us to analyze and model data based on time-dependent patterns. In this paper, we explore different time series analysis techniques and their application to sales forecasting. We use a real-world sales dataset (retail) to demonstrate the use of various time series techniques such as decomposition, auto-correlation, and lag features. This report presents a solution for a case study in which we forecast the sales of retail stores. It supports strategic decisions on three levels: the featuring of data, decomposing the data, and applying the models. We also discuss the significance of feature engineering in time series analysis and demonstrate the time series features such as lag, date time, and windowing (rolling means). Then, we compare the performance of different time series models, such as naive (persistence), Moving Average, ARIMA, and SARIMAX. We conclude that time series analysis techniques are used correctly and can handle powerful tools for businesses to make accurate sales forecasts and make informed decisions.

Keystroke dynamics, sometimes known as typing biometrics, is an automated method of recognizing or verifying a person's identity based on the style and rhythm of their keyboard strokes. It alludes to the precise timing data that shows when each key was pressed and when it was released during keyboard typing. Keystroke Dynamics assists in identifying a specific person's hand biometric template. This research uses a wide range of dwell time and flight time attributes to ascertain the hand posture of a specific person. To determine a user's hand posture at any given time, an Android application was created to record about 13 distinctive and unquestionably important attributes. Through this application, information was gathered by having users participate in a typing session. A variety of keystroke-related data, including Pressure, Finger Area, Uptime, and Downtime for each key, and motion-based data, including RawX, RawY, GravityX, GravityY, and GravityZ, were collected. Additionally, to achieve higher levels of accuracy, multiple Machine Learning models were used including ensemble classification methods like Bagging and Boosting to achieve conclusive results. It was observed that the Random Forest classifier obtained the highest accuracy score of 97.30%. The model was integrated with a mobile application and was utilized to identify the hand involved in the typing process. This work can be extended to include the field of surveillance, Multi-factor Authentication (MFA), and to help improve the one-hand mode layout.

Adolescents frequently encounter news reports multiple times daily, which can induce feelings of anxiety, stress, and fear when they come across stories about crimes. Studies indicate that young people tend to replicate behaviors and attitudes they observe in the news, making them vulnerable to becoming numb to violence and increasingly prone to violent and aggressive conduct. Prolonged and repeated exposure to such events may have serious consequences, including fear, insensitivity, and behavioral changes. So, it is essential to have a system that can approximately classify safe and unsafe news for teenagers and only the safe news is visible to them. We address this challenge of text extraction and classification from News Headlines using well-known statistical measures and machine learning (ML) models. In this proposed system, we compare Linear Support Vector Classifier (LSVC), Logistic Regression (LR), Multinomial Naïve Bayes (MNB), Random Forest Classifier (RFC), and Decision Tree Classifier (DTC) algorithms in which LR outperforms the other algorithms.

The optimization problems can be solved by using population based heuristic search techniques namely Particle Swarm Optimization (PSO) and Genetic Algorithm (GA). One of the drawbacks of standard PSO was to prematurely converge on local optimal solutions. In this work, we have used Turbulent Particle Swarm Optimization (TPSO) instead of standard PSO due to the drawback mentioned above. Here, different operations of Genetic algorithm were included to obtain a good solution. In this paper, we would like to compare the results of both algorithms. Experimental results were examined with functions which were function maximization and results show that the Turbulent PSO outperform the GA.

Computer vision has emerged as an important subject of study, with several practical applications in a wide range of domains. OpenCV, a widely used framework, has played an important role in allowing computer vision tasks. This study presents an AI-driven Python implementation of the OpenCV framework to expand its capabilities. Data collecting and pre-processing, feature extraction, model selection and training, and Python-based system implementation are all part of the proposed system. A thorough examination of system performance indicates its advantage over competing approaches. The system's flexibility enables it to handle a variety of computer vision tasks, such as gesture recognition, face detection, and object recognition. This paper advances AI-powered computer vision systems by providing significant insights into the implementation of OpenCV with Python. The focus of future work will be on improving system accuracy and broadening its functional range.

The food marketplace needs a quick and reliable system for tracking and assessing the freshness of meat products. However, meat experiences a quick process of freshness deterioration, which leads to bacterial growth. As a result, the need for a reliable and quick way of monitoring and evaluating meat deterioration is growing urgent. By Considering these aspects, this paper proposes a Novel Fifteen Layered Deep Convolutional Neural Network (15L-DCNN) to predict the freshness state of meat with maximum accuracy. The model utilizes the Meat Freshness Image Dataset extracted from the KAGGLE machine learning repository. The Meat Freshness Image Dataset comprises three meat state classes, Fresh Meat, Half Fresh Meat, and Spoiled Meat, with 2269 meat images. The Meat Freshness Image Dataset have been subjected to data augmentation and performed with four operations: Random horizontal flip, Random vertical flip, zooming, and rotation. After data augmentation, the dataset ends with 6000 images. The Meat Freshness Image Dataset was splitted into 4800 training images, 600 validation images, and 600 testing images. The Meat Freshness training Images were subjected to the proposed 15L-DCNN and the same dataset was applied to EfficientNet, DenseNet, and ResNet Large models for evaluating the efficiency metrics. Python was adopted for the execution of NVidia Geforce Tesla V100 GPU workstation with 100 training iterations for a block size of 64. Experimental results show that the proposed model 15L-DCNN shows a maximum accuracy of 98.85%, Precision of 98.33%, Recall of 98.25%, misclassification rate of 1.15%, and FScore of 98.24% when compared with another convolutional neural network.

Autonomous vehicles are becoming more common in various industries, but the use of autonomous maritime vehicles is still being studied. This is because controlling these vehicles requires making important decisions about design, propulsion, payload management, and communication systems, which can lead to errors and collisions. One major challenge is detecting other ships and objects in real-time to avoid collisions. Recently, deep learning techniques based on convolutional neural networks (CNNs) have been developed to help with this challenge, such as YOLOv8 (You Only Look Once) and EfficientDet. This paper examines how these methods can be used to detect ships. We trained and tested these two models on a large maritime dataset. On examining the performance of the two models, we have compared the working of both.

Although surveillance cameras are used in ATM cells, we face some problems of robbery and theft at ATMs due to lack of security; however, the monitoring capacity of law enforcement agencies has not kept pace. ATM spoofing attacks can be carried out to break or damage the ATM by stealing the machine and taking cash from the ATM. To reduce this problem, we arm the ATM with a camera module mounted in the room to perform continuous video observation. The camera detects the human and his activity in the ATM and attempts to breach the ATM. It detects unusual activities and immediately sends an alert notification to the police. Therefore, the system handles the application developed to automate video surveillance and detect any potential criminal activity at ATMs. Therefore, in this work, abnormal behavior is observed using CNN and RNN in surveillance videos. These algorithms can be used to recognize faces, detect and track camera movements, and detect and identify the action required to prevent such activity.

Text summarization, a technique in Natural Language Processing, helps in summarizing documents like news articles, legal documents, essays and more. The content may be comprehensive and redundant. A summary gives an insight of the document. Text summarization is broadly classified into two categories—extractive and abstractive summarization. Abstractive summarization uses deep learning techniques to generate summary, just as humans generate summary using their own words and sentences. Extractive summarization highlights information based on some features or technique used to identify the importance of the sentence from the source document. Methods used for extractive summarization include ranking algorithms, sentence scoring, sentence similarity and so on. In this paper, a framework for extractive text summarization using features extracted from a Tamil document has been proposed. The summarizer is based on Fuzzy logic inference engine. The framework describes the modules involved in the generation of Extractive Text Summary for a single document.

TheAndrew, B. increase in the availability ofSandeep, D. data on the Internet in the past years has created an enormous amount of data and research in the field of Artificial Intelligence and Machine Learning. With the advancement inRobert, L. technology, computational power has also increased dramatically in the past few years, and this has led to more and more advancements in ArtificialAlexander, G. Intelligence research and its applications. Mizo language, which is a low-resource language, also tends to emerge in recent years along with these advancements and with the help of news articles collected from the two biggest news outlets for the Mizo language namely Vanglaini and The Aizawl Post, an approach to news classification based of their category was done in this paper. This paper tested several machine learning methods using supervised classification techniques and got the highest accuracy among other low-resource languages in most of the models tested and among which Multinomial Naive Bayes classification gives an accuracy of 96% and is the highest when compared to the other models.

The availability of information and news over the Internet is exploding. In this context, text summarization is becoming very important since it gives a good overview of the content. Also, it saves time by exposing the most significant information at a glance. Summarization techniques are very vital in extracting this useful information from lengthy text. In this work, a novel architecture for abstractive text summarization architecture, BASiP, has been proposed, which effectively generates a summary from the given text. The base model used for summarization is BART. The proposed architecture is compared with the existing work. It is found that BASiP performs well in terms of the ROUGE score. Also, a case study is given at the end to show the efficiency of BASiP, in generating a meaningful summary.

Natural remedies are less expensive, non-toxic, and associated with negative side effects. As a result, their demand is rising, particularly for herbal-based medicinal products, health products, nutritional supplements, and cosmetics. Threats from leaf diseases exist to the global agricultural industry's economic and production status. The need for farmers to protect agricultural products is reduced by the ability to find illness in leaves utilizing Deep learning (DL) and Machine learning (ML). Our approach involves a combinations method for the diagnosis of flora illness. In our suggested method RESNET-50 is employed for extracting the deep features and Random Vector Functional Link (RVFL) is employed for the classification. To look at the efficiency of the suggested RES-RVFL model, its categorizing performance is contrasted with Support Vector Machine (SVM), Decision Tree, Random Forest and K-Nearest-Neighbors (KNN). The findings showed that RVFL is very suitable for classifying leaf diseases, with a disease classification accuracy of about 94%. The fact that this result highlights the significance of early detection and naming of flora diseases for justifiable cultivation and food security is very positive. Our research has a solid foundation, thanks to the Plant Village dataset, and our findings add to the body of knowledge on applying deep learning and machine learning to identify plant diseases.

Food production through Agriculture plays an important role in keeping the world’s population hunger-free and nations economically secure. The continuous change in land minerals, weather situation, and pesticide usage affect the yield of the crops. Farmers can choose successful crops for the season with the help of machine learning algorithms used for crop yield prediction. In this study, we forecasted agricultural production using numerous kinds of machine learning models while considering several factors that affect crop yields, such as rainfall, temperature, and pesticide use. By merging multiple separate model predictions, ensemble machine learning models improve the performance of the machine learning models. We have worked with individual models and ensemble models like SVR, RandomForestRegressor, LinearRegressor, and DecisionTreeRegressor to predict crop yield and found an ensemble solution that combines the strengths of both the stacked generalization model and the gradient boost algorithm which can provide improved accuracy and robustness in crop yield prediction. According to the findings, the ensemble solution provided an R2 score of 98 percent, which is higher than the R2 scores of 96 percent obtained using the Decision Tree Regressor and 89 percent obtained using the Gradient Boosting Regressor.

There is a lot of research going on in the agriculture business right now to create new medications or insecticides to preserve crops. However, this leads to the blind use of insecticides to crops without identifying insects based on their benefits. In the realm of agriculture, there are two sorts of insects: pests and non-pests. Pests are known to harm crops or degrade the environment in which crops thrive, but non-pests may hunt pests, which is beneficial and accomplishes the work without the need of pesticides. The objective of this work is to use the best model for the object detection. This work uses YOLO v7 model as it stands to be one of the best models crossing Mask R-CNN. The model helps in recognizing the pests more accurately and distinguishing them from regular insects. YOLO v7 has enhanced the model by obtaining higher accuracy and reducing the mean square error. The significance of the model lies in achieving the accuracy and thus the model could act as a tool for the farmers to take necessary action. The performance metrics obtained through this model has outperformed the other models.

Acute lymphoblastic leukemia (ALL) is a cancerous condition which affects bone marrow and blood. It is a fast developing illness that, if not identified and treated as soon as possible, could be fatal. ALL is often identified by hematologists through observing the blood and bone marrow smears under a microscope. In order to diagnose and classify leukemia, sophisticated cytochemical tests are employed. However, such processes are resource-intensive, time-consuming, and reliant on the expertise of the doctors doing them. In order to diagnose leukemia, image processing techniques are used to examine microscopic smear images for signs of cancerous cells. These methods are simple, quick, cheap, and not influenced by the views of specialists. In this paper, a computer-aided automated diagnostic method is proposed to classify ALL and healthy cells based on Random Forest classifier with most significant features. For this model, the public dataset ALL-IDB 2 has been utilized. The proposed approach provided an accuracy of 99.73% to classify the cells (ALL and healthy). Also, it shows an improvement in accuracy of 6.16%, 16.4%, and 10.43% in comparison to the approaches, i.e., morphological + color feature with SVM, Hausdorff dimension + shape feature with SVM, and GLCM + Morphological with SVM, respectively.

Federated Learning (FL) permits the cooperative training of a joint model for several medical facilities while maintaining the decentralization of the data owing to privacy considerations. However, Federated optimizations often struggle with the heterogeneity of data dissemination among medical facilities. Nowadays, the domains of medical image classification, compression, and privacy are particularly difficult for diagnosing disease. The transmission of these medical images through the internet for diagnostic reasons must be protected against cyberattacks. In this proposed method, a Federated Learning approach with a Convolutional Neural Network (FedCNN) and Federated Averaging (FedAVG) is employed for classification problems. This technique adjusts the contribution of each data sample to the local goal during optimization based on knowledge of the client’s label distribution, thereby minimizing the instability caused by data heterogeneity. The model utilizes a hybrid approach to ensure consistency in time-series data. The datasets, namely, COVIDx-19 X-ray and malaria that are freely accessible are the subject of our in-depth investigations. The experimental results have been analyzed by evaluation metrics, namely, accuracy (78.79 and 98.92), precision (73.72 and 95.73), and recall (71.91 and 93.91) for proper validation. The findings demonstrate that FedCNN achieves better convergence performance than the main FL optimization methods under comparison.

Acute lymphoblastic leukemia (ALL) is a cancerous condition that affects the bone marrow and blood. It is a fast developing illness that, if not identified and treated as soon as possible, could be fatal. ALL is often identified by looking at blood and bone marrow smears under a microscope. Leukemia can be detected and classified using detailed cytochemical tests. However, these procedures are expensive, time-consuming, and dependent on the knowledge and skills of the specialists involved. Using image processing techniques that examine microscopic blood smear images to search for the leukemic cells, leukemia can be detected. These methods are simple, quick, cost-effective, and unaffected by the judgments of experts. The suggested study describes a computer-aided diagnosis method that uses deep convolutional neural networks (CNNs) that have already been trained to compare leukemia images to normal images. The public dataset ALL-IDB 2 was used for the proposed research. The study uses the pre-trained model DenseNet-201 for performing the classification. With the DenseNet201 pre-trained networks employed in the study for the ALL_IDB2 dataset, a classification accuracy of 94.6% is achieved. In all of the classifications carried out, optimization strategies such as cross-validation, fine-tuning, and real-time augmentation are also compared. Also use Pre-trained series models like ResNet-50, VGG-19, Inceptionv3, MobileNet-v2, Xceptionv3, and VGG-16 for performing the comparison. The experimental result gives an improvement in accuracy (17.76, 10.6, and 13.2%) in comparison to the other approaches namely, residual neural network, customized combined CNN, and conVNet neural network, respectively.

The rise of chronic diseases has become a major public health challenge globally. Early prediction and prevention of these diseases can help reduce their prevalence and improve patient outcomes. The proposed disease prediction system, which is based on predictive modeling, may anticipate the user’s illness by using the user’s symptoms as input. The framework evaluates the symptoms taken as input by the user and generates the likelihood of developing the disease. The disease prediction framework based on machine learning (ML) techniques can help in a more accurate diagnosis than conventional methods. In the current manuscript, we have designed a disease prediction methodology using multiple ML techniques. The proposed framework also has the potential to enhance disease surveillance and support public health interventions, including disease management and resource allocation. The accuracy of our approach is shown over the benchmark data sets, which consist of more than 230 diseases. The suggested diagnostic algorithm outputs the disease name that a person might be experiencing based on the symptoms taken into consideration. The proposed framework provides a scalable and effective solution for public health decision-makers to manage chronic diseases and improve patient outcomes.

Early detection of diabetes plays a crucial role in improving health outcomes and can help people avoid harmful diabetes complications. Machine learning algorithms are being used to diagnose a disease in its early stages. This study proposes an optimized weighted ensemble model that can predict the risk of type 2 diabetes mellitus. A diabetes dataset of 403 patients from the Department of Medicine at the University of Virginia given by Dr. John Schorling has been used. We assessed ridge regressor, LASSO, feedforward artificial neural networks, and linear regression prediction performance. These models were then combined to create an optimized weighted ensemble model for prediction. We evaluated our prediction models using standard performance metrics: coefficient of determination (R2 score), root mean square error (RMSE), mean square error (MSE), and mean absolute error (MAE). The results showed that the proposed optimized weighted ensemble model outperformed individual models, achieving the highest 0.81 (R2 score) and lowest 0.98 (MSE).

Image captioning is the method of generating textual descriptions for an image using deep neural networks. Its objective is to produce accurate results to specify the hidden features and to satisfy its wide applications. There are various Convolutional Neural Network (CNN)-based encoder architectures available in the literature for image caption generation and there is a need to empirically evaluate the best-performing architecture on multiple and diverse datasets to check their efficacy and generalization capability. To address this, we performed the experiments using the Flickr30K dataset containing 31,783 images along with the commonly used Flickr8K dataset consisting of 8091 images. In this study, we aim to discover the best-suited CNN architecture models for caption generation. The study evaluated various encoder architectures, including VGG16, VGG19, InceptionV3, and InceptionResNetV2, for extracting image features and Long Short-Term Memory (LSTM) as a decoder for generating accurate captions. The models are analyzed based on accuracy variation and value loss metrics on both datasets. The results depict that all models perform better on the larger dataset, i.e., Flickr30K achieves better accuracy with minimum loss values and VGG19 shows the best results among all.

Thanks to constantly advancing technology, the world is changing rapidly. One such idea that has contributed to the reality of automation is the Internet of Things (IoT). IoT links various non-living objects to the internet and enables them to communicate with their local network to automate processes and simplify people's lives. The IoT's potential needs to be completely realised despite the enormous efforts of standards, organisations, coalitions, businesses, academics and others. A number of problems remain. The enabling technology, applications and business models, as well as the social and environmental repercussions, should all be taken into consideration while analysing these difficulties. This article's emphasis is on unresolved issues and challenges from a technological perspective. The main objective is to provide a thorough evaluation of IoT in terms of energy and security, as well as unresolved problems and obstacles that need further study. We provide some perspectives on some new concepts to help future research.

Cloud computing (CC) has transformed the way businesses store, process, and access data, and spot pricing has emerged as a key feature of this technology. Spot pricing enables users to bid on unused computing resources at lower prices, providing cost-effective access to computing resources and helping businesses optimize their operational infrastructure expenses. The proposed research is focused on case studies and existing research based on the impact of spot pricing in CC, including its effect on workload distribution and resource allocation. It also emphasizes the significance of spot pricing in CC and its potential to benefit businesses of all sizes. The paper concludes with open research directions, including the benefits and challenges of spot pricing, the various factors to consider when using spot pricing, and the need for future research in this area highlighting its importance and potential impact on businesses.

The Non-Line-of-Sight (NLOS) communication in millimeter wave (mmWave) experiences high path loss in the urban region due to reflection, blockage etc. To design an efficient 5G system, the channel should be modelled such that data rate and capacity are high. This paper presents the close-in (CI) free space reference distance model, CI model whose path loss exponent is frequency weighted (CIF) and the alpha-beta-gamma (ABG) model for 26 GHz. The use of these models in 3rd Generation Partnership Project (3GPP) and Fifth Generation Wireless System design has drawn the attention of researchers to investigate more. As 26 GHz 5G band is commercially used for 5G communication in India, we have analyzed the path loss and capacity in this frequency taking different distances for Urban Macrocell (UMa), Urban Microcell (UMi) and input office scenario considering NLOS communication. The results show that path loss varies with variation of cell size.

In recent years, there is a surfeit of digital currencies, virtual currencies, and cryptocurrencies. These currencies serve as alternatives to fiat currencies in the form of physical currencies or deposits in banks. Some of the common characteristics that differentiate these currencies are how they gain and maintain their value, the anonymity of transactions, and considerations of security, data integrity, and transaction performance in a distributed computing scenario. Digital Currencies are generally issued by Central Banks raising privacy concerns. Cryptocurrencies work outside such formal mechanisms raising concerns about volatility and the possibility of loss. In this paper, we discuss a new variant called δ-Currency which attempts to navigate these concerns and arrive at an alternative mechanism that addresses privacy and financial security considerations in a novel manner. The δ-Currency is currently at a conceptual stage. We make use of Model Checking to validate the architecture, design principles, and implementation approach of δ-Currency. Model Checking focuses on fundamental building blocks of computation and is generally used to validate complex systems operating in an environment that enables a high degree of concurrency. In this paper, we make use of Model Checking to arrive at a common vocabulary, abstractions, and framework to understand seemingly disparate systems which nevertheless achieve the same objective.

In recent years, blockchains have attracted a great deal of interest from researchers, engineers, and institutions. Additionally, the implementation of blockchains has begun to revitalisea large number of applications, such as e-finance, smart homes, smart health, social security, logistics, and so on. As supply chains become increasingly global, management and control become more complicated. Blockchain technology is establishing promise in tackling several supply chain management (SCM) concerns as a distributed digital ledger platform that ensures security, traceability, and transparency. Government, societal, and consumer pressures have encouraged us to consider how blockchain can help us meet our sustainability goals and improve supply chain sustainability. True blockchain-led supply chain and corporate transformation is still in its early phases. This chapter examines the possible application of blockchain technology (BT) and smart contracts to SCM. In order to study the operations of blockchain based supply chain management a queueing based model is developed. It assesses the measures of the model, such as the mean transactions per unit time, the mean transactions in the unconfirmed transaction pool, the mean time required for transaction confirmation, and the average waiting time of transactions in the unconfirmed transaction pool. Additionally, it evaluates the characteristics of the supplychain (SC) system.

BiFrost is a revolutionary solution for secure online communication and data storage. It addresses security concerns such as eavesdropping, man-in-the-middle attacks, and censorship by offering decentralization, immutability, and data security through the use of blockchain technology. User-submitted data are added directly to the blockchain, creating a global copy in each node, and only authorized users can access the data using private keys. The system is fully decentralized, meaning there is no central authority, making it immune to censorship and government oppression. The technology used in BiFrost includes IPFS, smart contracts, Ethereum, INFURA, Solidity, and SPF. BiFrost provides a secure and decentralized solution for online communication and data storage, eliminating the need for a trusted intermediary and ensuring the privacy and security of data.

Adoption of cloud computing based computing solutions is growing day by day in nearly all sectors of society. Technological advancements such as load balancing help service providers uphold the quality of service, and thereby retain the confidence of service consumers. Collaborations among cloud environments can be established for resolving load balancing and fault tolerance issues up to a certain extent. However, the gigantic increase in consumption of cloud services makes cloud resource management difficult in intercloud environments too. Load balancing helps resolve workload balancing problems for collaborated cloud platforms. We present an enhanced and priority-oriented mechanism for sharing resources for workload balancing in collaborated cloud environments. In order to provide enhanced load balancing solution, the suggested resource sharing mechanism works on the priority values of participating instances. Employment of the suggested load balancing mechanism avoids starvation by means of lowering waiting time. The suggested technique has been implemented on a physical cloud testbed built using OpenStack cloud computing setup on CentOS Linux operating system. The experimental results reveal lesser waiting time of the highly loaded cloud instances.

The development in software, hardware and communication technologies has made the broadcasting of sensory data collected from various devices very easy and simple. Interconnected devices through Internet technology form the Internet of Things (IoT). Applying intelligent methods for the analysis of this big data is the key which develops smart IoT applications. The world today has become increasingly dependent on digitized data which raises various security concerns and the need for advanced and reliable security technologies to deal with the increasing number of cyber-attacks. The work depicted in this paper makes use of machine learning techniques to detect cyber-attacks using the UNSW-NB15 data set and the KDD CUP 1999 dataset. Decision Tree, k- means clustering, multi-layer perception (MLP), Naive Byes and Random Forest classifier are the algorithms used in this work in order to find higher level information about the data.

The twenty-first century has witnessed widespread adoption of Artificial Intelligence (AI), Machine Learning (ML) and Deep Learning (DL). These techniques have provided reliable solutions in various areas, including statistics, information theory, and mathematics. Given the prevalence of ML techniques, there exist various adversaries which question the robustness of ML models. Adversaries aim to manipulate models to their advantage, reducing their performance and accuracy. Data poisoning attack is one such adversary in which the attacker manipulates models by introducing specially crafted poisoned data into the training dataset. This paper presents the performance analysis of different machine learning models with and without the influence of data poisoning attack to predict the probability of diabetes and its effect on accuracy and precision. It has been observed that the SVM (RBF) classifier performs best on clean data, while the KNN classifier is highly affected by data poisoning, with a lesser impact on the SVM (Linear kernel) classifier.

From breaching the service and allowing humans to surpass it upon correct verification, nowadays, it is achievable by bots and machines using machine learning, making the service vulnerable. It can pose several challenges like hacking, crashing, exploiting to name a few. In the present work, a novel framework namely BypassCaptcha model to test the security strength of web services via captchas is proposed. This model examines the vulnerability of the security via automation and captcha decoding using deep learning models particularly, i.e. Convolutional Neural Network, Recurrent Neural Network, and Connectionist Temporal Classification Loss. It involves full automation processing i.e. from opening the service, entering the credentials, getting a captcha from the service, and inputting the right decoded captcha. This complete process is dynamic. The credentials are provided via file, read during runtime, and are inputted at their required place. For training the model, a dataset combinely having four different types of captcha i.e. arc, dotted, rotated, and noisy is used. The automation process is working satisfactorily on specific services but it is still not a convenient way for a large number of services at a time. The proposed model is having a Val_Loss of 97% in the clear captcha case.

Watermarking is the method in which an image or a text is secretly embedded in an image taken as the original image. It is helpful in recognizing the original possessor of a particular content. It generally works on different objects like pictures, audio, video etc. Reversible and Robust Watermarking is a technique in which the original image as well as the watermarked image can be retrieved successfully even if the original or the image formed after the embedding phase is attacked by noise or other factors. This work is presenting an algorithm or a technique based on Robust and Reversible Watermarking. The algorithm presented in this work is divided into three phases: namely, watermark generation phase; the phase in which the generated watermark is inserted into the original image; and the phase of extraction of watermark and the original image separately or lossless. The watermark is obtained and embedded to the original image by using weber’s differential excitation descriptor, reference from law of weber, and interpolation linearly. Paillier Cryptosystem is used to encrypt the embedded image. To make the embedded image more secure, Discrete Wave Transform is applied. The original image is not necessary for the extraction process. Linear interpolation in inverse method is used to retrieve the value of the watermark of each block consisting of pixels. To determine the accuracy and robustness of the original image and the watermark that is extracted from the embedded image, different operations, or factors like PSNR, BER and Surviving bit rate are calculated. Random tampered zones could also be found using this algorithm.

Penetration testing and vulnerability assessment are critical components of modern information security strategies. The Metasploit Framework is one of the most widely used pen-testing tools, offering a range of capabilities for detecting and exploiting vulnerabilities in systems and applications. This paper presents a comparative analysis of the Metasploit Framework with other popular pen-testing tools, highlighting its strengths and weaknesses. The study also evaluates the effectiveness and efficiency of the Metasploit Framework through a series of experiments and simulations, using various criteria such as accuracy, speed, and ease of use. The results show that the Metasploit Framework offers a powerful and flexible toolset for pen-testing and vulnerability assessment, with several unique features and advantages over other tools. However, the study also identifies some limitations and areas for improvement, such as the need for better documentation and support for advanced techniques. The findings have important implications for information security professionals and organizations, providing insights into the strengths and weaknesses of the Metasploit Framework and its role in modern security strategies.

The security aspect of cloud computing is much more challenging for researchers. Preventing the attack requires knowledge about the type of attack, its origin, and how vulnerabilities and tools are used for the attack. The cloud security methodology protects customer data, information, and applications from attackers. Due to digitalization, the volume of data is increased. The protection of data is very challenging for cloud service providers. This work proposes a new cloud intrusion detection system security infrastructure based on Honeynet, Honeywell, Honeypot, and Honeytoken Strategies. The proposed strategy effectively identifies intrusion detection, attack behavior, and attack scenario. The testing of data was carried out in OpenStack environments.

Manufacturing, energy, finance, education, transportation, smart home, and medicine employ IoT technology. IoT solutions can efficiently manage hospital patients and mobile assets to provide high-quality medical services. The Internet of Medical Things (IoMT) integrates IoT with medical equipment to increase patient comfort, cost-effective medical solutions, hospital treatment speed, and personalized healthcare. This work uses Constrained Application Protocol to secure remote patient health data (CoAP). Nevertheless, the CoAP DTLS layer lacks key control, session establishment, and multicast message exchange. Hence, IoMT communication requires an efficient protocol for safe CoAP session formation. Consequently, to address key management and multicast security issues in CoAP, we presented an efficient and secure communication method to establish a secure session key between IoMT devices and distant servers utilizing lightweight, energy efficient, and Secure CoAP Elliptic Curve Cryptography (E2SCEC2). E2SCEC2 can use a smaller key size than Rivest-Shamir-Adleman (RSA) due to its tiny key size. To determine if these algorithms are compatible in limited contexts, the paper examines key creation, signature generation, and verification of E2SCEC2 and RSA algorithms, energy consumption, and radio duty cycle.

For a company's product life cycle to be faster, more efficient, and more effective, supply chain management (SCM) is critical. Nevertheless, present SCM systems must be improved to ensure genuine products and transactions are private and secure. As a result, this study proposes a safe SCM system depending upon blockchain and the Internet of Things to ensure that the products are genuine. All SCM stakeholders can initiate an encrypted, secure transaction for their goods or services due to the Quick Response (QR) scanner that works with the Internet of Things (IoT) and the blockchain-integrated distributed system. Finally, a genuine product from the manufacturer will be sent to the customer. To make authentication quicker and easier for scattered IoT devices, a lightweight asymmetric key encryption mechanism known as Diffie-Hellman key exchange and Hyperledger Fabric-based blockchain technology with on-chain smart contracts are deployed. The service provider registers each SCM stakeholder and provides them with their own public and private keys, which will be used to authenticate participants and IoT devices. Examining security and scalability shows that the recommended solution is more trustworthy and secure than existing methods.

Increase in wireless devices made mobile communication pervasive. Global Mobile Networks (GLOMONET) provision the roaming service to accomplish this, where mobile users must experience secure and seamless roaming services over multiple foreign agents. The main objective of network providers is to have mutually authenticated, secured, and lightweight service to guard mobile user’s data and privacy. Many interesting roaming authentication protocols have been proposed to achieve the security and privacy of mobile users in traditional communication networks. But they all suffer from one or another known security attack with the fact that current mobile networks are prone to attacks. Blockchain technology offers its advantages to establish a secure connection and authentication by safeguarding mobile user information and privacy with its immutable nature. The study shows that limited work has been done in space protocol design for GLOMONET using blockchain technology and the main goal of the protocol is to maintain security for transactional data and privacy of the mobility users along with anonymity property. In this article, soulbound tokens are used to issue credentials between the mobile user and Home Agent (HA) by serving as a secure and decentralized form of digital identity. The idea behind using soulbound tokens for issuing credentials is to create a tamper-proof and easily verifiable system that reduces the reliance on centralized authorities for identity verification. In addition, the smart contracts for user authentication have been implemented through solidity programming and the security strength of the proposed protocol is verified through a formal verification tool called AVISPA (Automated Validation of Internet Security Protocols and Applications).