This paper highlights a novel approach to address multiple networking-based VM allocation and migration objectives at the cloud data center. The proposed approach in this paper is structured into three distinct phases: firstly, we employ a Bi-Directional Long Short Term Memory (BiLSTM) model to predict Virtual Machines (VMs) instance’s prices. Subsequently, we formulate the problem of allocating VMs to Physical Machines (PMs) and switches in a network-aware cloud data center environment as a multi-objective optimization task, employing Linear Programming (LP) techniques. For optimal allocation of VMs, we leverage the Branch-and-Bound (BaB) technique. In the third phase, we implement a VM migration strategy sensitive to SLA requirements and energy consumption considerations. The results, conducted using the CloudSim simulator, demonstrate the efficacy of our approach, showcasing a substantial 35% reduction in energy consumption, a remarkable decrease in SLA violations, and a notable 18% increase in the cloud data center’s profit. Finally, the proposed multi-objective approach reduces energy consumption and SLA violation and makes the data center sustainable.

For the safety and reliability of the system, Remaining Useful Life (RUL) prediction is considered in many industries. The traditional machine learning techniques must provide more feature representation and adaptive feature extraction. Deep learning techniques like Long Short-Term Memory (LSTM) achieved an excellent performance for RUL prediction. However, the LSTM network mainly relies on the past few data, which may only capture some contextual information. This paper proposes a hybrid combination of Convolution Neural Network (CNN) and LSTM (CNN+LSTM) to solve this problem. The proposed hybrid model predicts how long a machine can operate without breaking down. In the proposed work, 1D horizontal and vertical signals of the mechanical bearing are first converted to 2D images using Continuous Wavelet Transform (CWT). These 2D images are applied to CNN for key feature extraction. Ultimately, these key features are applied to the LSTM deep neural network for predicting the RUL of a mechanical bearing. A PRONOSTIA data is utilized to demonstrate the performance of the proposed model and compare the proposed model with other state-of-the-art methods. Experimental results show that our proposed CNN+LSTM-based hybrid model achieved higher accuracy (98%) with better robustness than existing methods.

This book reviews research works in recent trends in blockchain, AI, and Digital Twin based IoT data analytics approaches for providing the privacy and security solutions for Fog-enabled IoT networks. Due to the large number of deployments of IoT devices, an IoT is the main source of data and a very high volume of sensing data is generated by IoT systems such as smart cities and smart grid applications. To provide a fast and efficient data analytics solution for Fog-enabled IoT systems is a fundamental research issue. For the deployment of the Fog-enabled-IoT system in different applications such as healthcare systems, smart cities and smart grid systems, security, and privacy of big IoT data and IoT networks are key issues. The current centralized IoT architecture is heavily restricted with various challenges such as single points of failure, data privacy, security, robustness, etc. This book emphasizes and facilitates a greater understanding of various security and privacy approaches using the advances in Digital Twin and Blockchain for data analysis using machine/deep learning, federated learning, edge computing and the countermeasures to overcome these vulnerabilities.

This book reviews research works in recent trends in blockchain, AI, and Digital Twin based IoT data analytics approaches for providing the privacy and security solutions for Fog-enabled IoT networks. Due to the large number of deployments of IoT devices, an IoT is the main source of data and a very high volume of sensing data is generated by IoT systems such as smart cities and smart grid applications. To provide a fast and efficient data analytics solution for Fog-enabled IoT systems is a fundamental research issue. For the deployment of the Fog-enabled-IoT system in different applications such as healthcare systems, smart cities and smart grid systems, security, and privacy of big IoT data and IoT networks are key issues. The current centralized IoT architecture is heavily restricted with various challenges such as single points of failure, data privacy, security, robustness, etc. This book emphasizes and facilitates a greater understanding of various security and privacy approaches using the advances in Digital Twin and Blockchain for data analysis using machine/deep learning, federated learning, edge computing and the countermeasures to overcome these vulnerabilities.

The smart healthcare system advancements have introduced the Internet of Things, enabling technologies to improve the quality of medical services. The main idea of these healthcare systems is to provide data security, interaction between entities, efficient data transfer, and sustainability. However, privacy concerning patient information is a fundamental problem in smart healthcare systems. Many authentications and critical management protocols exist in the literature for healthcare systems, but ensuring security still needs to be improved. Even if security is achieved, it still requires fast communication and computations. In this paper, we have introduced a new secure privacy-enhanced fast authentication key management scheme that effectively applies to lightweight resource-constrained devices in healthcare systems to overcome the issue. The proposed framework is applicable for quick authentication, efficient key management between the entities, and minimising computation and communication overheads. We verified our proposed framework with formal and informal verification using BAN logic, Scyther simulation, and the Drozer tool. The simulation and tool verification shows that the proposed system is free from well-known attacks, reducing communication and computation costs compared to the existing healthcare systems.

Reversible data hiding (RDH) based on prediction error expansion (PEE) needs a reliable predictor to forecast the pixel. The hidden information is inserted into the original cover image pixels using the Prediction Error (PE). To improve the accuracy of pixel predictions for cover images, there are a number of algorithms available in the literature. Based on the different gradient estimations, several academics have suggested prediction methods. More research on this gradient-based pixel prediction method is presented in this article. In order to improve exploration gradient estimates, we have looked at a number of local contexts surrounding the current pixel. It has been stated that experiments have been conducted to evaluate the effect of different neighborhood sizes on gradient estimation. Additionally, we investigate two methods for choosing paths according to gradient magnitudes. To incorporate the data into the initial pixels, a new embedding technique called Prediction Error Expansion has been suggested. In the context of reversible data concealment, experimental results point towards a better gradient based prediction employing an prediction embedding technique. © The Author(s), under exclusive licence to Springer Nature Singapore Pte Ltd. 2024.

This paper proposes a novel approach to enhancing education systems by integrating federated learning techniques with big data analytics. Traditional data analysis methods in educational settings often need help regarding data privacy, security, and scalability. Federated learning addresses these issues by enabling collaborative model training across distributed datasets without data centralization, thus preserving the privacy of sensitive information. By harnessing the vast amounts of educational data generated from various sources such as online learning platforms, student information systems, and academic applications, federated learning empowers educational institutions to derive valuable insights while respecting data privacy regulations. Leveraging the collective intelligence of decentralized data sources, federated learning algorithms facilitate the development of robust predictive models for student performance, personalized learning recommendations, and early intervention strategies. Moreover, federated learning enables continuous model improvement by aggregating local model updates from participating institutions, ensuring adaptability to evolving educational landscapes. This paper explores the technical foundations of federated learning, its application in education systems, and its potential benefits in improving learning outcomes and fostering data-driven decision-making in education. Through a comprehensive review of existing literature and case studies, this research aims to provide insights into the opportunities and challenges associated with implementing federated learning-based big data analytics in education systems, ultimately paving the way for a more efficient and personalized approach to education

Blockchain technology is in fact a public ledger that gathers data in a chain of blocks, which gradually improves security, trust, transparency, quality, decentralization, and immutability while operating businesses. In the present scenario of business, the organization is not only concentrating on improving the activities related to operational aspects, but it also needs to meet the expectations of various stakeholders. Corporate social responsibility (CSR) is such a concept which facilitates the organization to cater the information related to various social and environmental concerns arising out of the business operations. It is now the liability on the part of the organization to communicate these CSR-related concerns in such a way that they effectively meet the expectations of stakeholders. CSR communication has become an integral part of the organization’s marketing strategy not only through the rise of public awareness on environmental and social issues but also because there is a demand for the correct use of CSR communication. However, organizations face difficulties in their CSR activities and actions, and due to this challenging situation, there is a rampant need for a solution. Blockchain is one of the most rewarding technology because it stores and records information in such a way that it makes it practically impossible to change or cheat the system. In fact, blockchain provides the desire transparency, traceability, decentralization, and accountability that CSR communication lacks recently. Therefore, this study identifies those common difficulties of CSR communication based on a literature review and proposes implementing blockchain as a solution for these problems. Finally, the objective of this study is to investigate what are the common problems or difficulties in CSR communication, and furthermore, what are the usefulness and benefits of blockchain, and could these benefits really overcome the identified difficulties

Daily digital payments in Financial Technology (FinTech) are growing exponentially. A huge demand is for developing secure, lightweight cryptography protocols for wearable IoT-based devices. The devices hold the consumer information and transit functions in a secure environment to provide authentication and confidentiality using contactless Near-Field Communication (NFC) or Bluetooth technologies. On the other hand, Security breaches have been observed in various dimensions, especially in wearable payment technologies. In this paper, we developed a threat model in the proposed framework and how to mitigate these attacks. This study accepts the three-authentication factor, as biometrics is one of the user’s most vital authentication mechanisms. The scheme uses an “Elliptic Curve Integrated Encryption Scheme (ECIES)”, “Elliptic Curve Digital Signature Algorithm (ECDSA)” and “Advanced Encryption Standard (AES)” to encrypt the messages between the entities to ensure higher security. The security analysis of the proposed scheme is demonstrated through the Real-or-Random oracle model (RoR) and Scyther’s widely accepted model-checking tools. Finally, we present a comparative summary based on security features, communication cost, and computation overhead of existing methods, specifying that the proposed framework is secure and efficient for all kinds of remote and proximity payments, such as mini, macro, and micro-payments, using wearable devices.

Remotely-Controlled Aerial Vehicles (RCAV), popularly known as drones, have gained wide popularity in several applications from military to civilian due to the usage of sensors, actuators, and processors with wireless connectivity for data collection and processing. However, data security and authentication are still challenging as sensitive data is collected and shared in RCAV using an open channel, i.e., the Internet. The existing security of Internet-of-Drones (IoD) communication mainly depends on the hardness of discrete logarithms and factorization problems. However, due to Shor's algorithm, both authorized and secure transmissions are challenging in the presence of highly scalable quantum computers. To mitigate the aforementioned issues and challenges, in this article, we propose an authorized and secure communication scheme for IoD based on Ring Learning With Error (RLWE) problem on lattices, which have the potential to sustain low computation and quantum attacks. The proposed scheme supports mutual authentication and has an efficient session establishment. The evaluation results show that the proposed scheme has superior performance as compared to the existing state-of-the-art solutions on benchmark data sets using various evaluation metrics.

We explore the usage of char-RNN which is special type of recurrent neural network (RNN) in generating music pieces and propose an approach to do so. First, we train a model using existing music data. The generating model mimics the music patterns in such a way that we humans enjoy. The generated model does not replicate the training data but understands and creates patterns to generate new music. We generate honest quality music which should be good and melodious to hear. By tuning, the generated music can be beneficial for composers, film makers, artists in their tasks, and it can also be sold by companies or individuals. In our paper, we focus more on char ABC-notation because it is reliable to represent music using just sequence of characters. We use bidirectional long short-term memory (LSTM) which takes input as music sequences and observer that the proposed model has more accuracy compared with other models.

Corn is one of India’s most popular food grains; due to crop disease, there is a significant loss in the productivity of corn. It not only leads to a significant adversarial impact on the Indian economy, but it is also a danger to food availability. In this paper, we propose a corn crop disease detection using three different state-of-the-art deep neural network frameworks known as ResNext101, ResNext50, and Inception V3. The corn dataset of 4187 images contains four categories of images such as healthy, cercospora, common rust, and northern leaf blight. The experimental results show that ResNext101, ResNext50, and Inception V3 attained 91.59%, 88.43%, and 78.5% average accuracy. Further, to check the performance of all the models, we calculated different performance metrics such as f1-score, recall, precision, and support values.

Automatic recognition of violence and nonviolence activities in the crowd management system is a broad area of interest in today’s scenario. In this paper, we propose a hybrid combination of the Convolution Neural Networks (CNNs), and Long Short-Term Memory (LSTM) model to recognize violence/nonviolence activities in a crowded area. In the proposed approach a stream of video is applied to a pretrained Darknet-19 network, then a CNN with LSTM network is used to extract spatial and temporal features from the video. In the end, these spatial features are applied to a fully connected layer to identify the violence/nonviolence condition. The experimental results show that 98.1% accuracy was achieved in the case of video, and 97.8% accuracy was achieved in the case of the image frame by our proposed violence/nonviolence detection model.

Due to the rapid utilization of cloud services, the energy consumption of cloud data centres is increasing dramatically. These cloud services are provided by Virtual Machines (VMs) through the cloud data center. Therefore, energy-aware VMs allocation and migration are essential tasks in the cloud environment. This paper proposes a Branch-and-Price based energy-efficient VMs allocation algorithm and a Multi-Dimensional Virtual Machine Migration (MDVMM) algorithm at the cloud data center. The Branch-and-Price based VMs allocation algorithm reduces energy consumption and wastage of resources by selecting the optimal number of energy-efficient PMs at the cloud data center. The proposed MDVMM algorithm saves energy consumption and avoids the Service Level Agreement (SLA) violation by performing an optimal number of VMs migrations. The experimental results demonstrate that our proposed Branch-and-Price based VMs allocation with VMs migration algorithms saves more than 31% energy consumption and improves 21.7% average resource utilization over existing state-of-the-art techniques with a 95% confidence interval. The performance of the proposed approaches outperforms in terms of SLA violation, VMs migration, and Energy SLA Violation (ESV) combined metrics over existing state-of-the-art VMs allocation and migration algorithms.

Software failure prediction is the process of building models that software interpreters can use to detect faulty constructs early in the software development life cycle. Faults are the main source of time consumption and cost less over the life cycle of applications. Early failure prediction increases device consistency and reliability and decreases the expense of software development. However, machine learning techniques are also valuable in detecting software bugs. There are various machine learning techniques for finding bugs, ambiguities, and faulty software. In this paper, we direct an exploratory review to assess the performance of popular techniques including logistical regression, decision tree, random forest algorithm, SVM algorithms, and DNN. Our experiment is performed on various types of datasets (jedit, Tomcat, Tomcat-1, Xalan, Xerces, and prop-6). The experimental results show that DNN produces a better accuracy among all techniques used above.

Heart disease is one of the foremost health problems nowadays, and deadliest human disease around the world. It is the main reason for the enormous range of deaths in the world over the previous few decades. Therefore, there is a need to diagnose it in an exceeding specific time to avoid abandoned dangers. In this paper, we propose a hybrid approach to heart disease prediction by using a given range of feature vectors. Furthermore, a comparison of several classifiers for the prediction of heart disease cases with a minimum number of feature vectors are carried out. We proposed two different optimization algorithms like genetic algorithm (GA), and particle swarm optimization (PSO) for feature selection, and convolution neural network (CNN) for classification. The hybrid of GA and CNN is known as genetic neural network (GCNN), and hybrid of PSO and CNN now as particle neural network (PCNN). The experimental results show that accuracy values obtained by PCNN is approximately 82% and GCNN is 75.51%.

-

Widespread transmission of digital image in social media has come up with security, confidentiality and authentication issues. Ensuring copyright protection of digital images shared through social media has become inevitable. To address these issues, a robust and secure digital image watermarking scheme using Redundant discrete wavelet transform (RDWT) - Singular value decomposition (SVD) hybrid transform is proposed in this paper. In the proposed scheme, digital image is divided into 4 × 4 non-overlapping blocks, and low information blocks are selected for embedding to ensure higher imperceptibility. For watermark embedding 1-level RDWT is applied on the selected blocks followed by SVD decomposition to make the proposed scheme highly robust against common attacks. One watermark bit is embedded in each left and right singular SVD matrices by adjusting the coefficients. This makes the proposed scheme free from false positive error and achieve high embedding capacity. Before embedding, watermark encryption is done by using a pseudo random key. The pseudo random key is generated adaptively from the cover image by using discrete wavelet transform saliency map, block mean approach and cosine functions. High imperceptibility and robustness is indispensable for the digital images shared through social media. But, these watermarking characteristics are in trade-off. In the proposed scheme, the trade-off is balanced by using optimized scaling factor (embedding strength). Scaling factor is optimized by using the proposed JAYA-Firefly (Ja-Fi) optimization. Experimental results demonstrate that the proposed scheme provides high imperceptibility, robustness, embedding capacity and security. Furthermore, performance comparison with the recent state-of-the-art schemes affirms that the proposed scheme has superior performance.

Credit scoring plays a vital role for financial institutions to estimate the risk associated with a credit applicant applied for credit product. It is estimated based on applicants' credentials and directly affects to viability of issuing institutions. However, there may be a large number of irrelevant features in the credit scoring dataset. Due to irrelevant features, the credit scoring models may lead to poorer classification performances and higher complexity. So, by removing redundant and irrelevant features may overcome the problem with large number of features. In this work, we emphasized on the role of feature selection to enhance the predictive performance of credit scoring model. Towards to feature selection, Binary BAT optimization technique is utilized with a novel fitness function. Further, proposed approach aggregated with 'Radial Basis Function Neural Network (RBFN)', 'Support Vector Machine (SVM)' and 'Random Forest (RF)' for classification. Proposed approach is validated on four bench-marked credit scoring datasets obtained from UCI repository. Further, the comprehensive investigational results analysis are directed to show the comparative performance of the classification tasks with features selected by various approaches and other state-of-the-art approaches for credit scoring.