Machine Learning, Blockchain Technologies and Big Data Analytics for IoTs: Methods, technologies and applications
2: Machine Intelligence Research Labs (MIR Labs), Australia
3: School of Software, University of Technology Sydney, Australia
4: Vilnius Gediminas Technical University, Lithuania
Internet of Things (IoTs) are now being integrated at a large scale in fast-developing applications such as healthcare, transportation, education, finance, insurance and retail. The next generation of automated applications will command machines to do tasks better and more efficiently. Both industry and academic researchers are looking at transforming applications using machine learning and deep learning to build better models and by taking advantage of the decentralized nature of Blockchain. But the advent of these new technologies also brings very high expectations to industries, organisations and users. The decrease of computing costs, the improvement of data integrity in Blockchain, and the verification of transactions using Machine Learning are becoming essential goals.
This edited book covers the challenges, opportunities, innovations, new concepts and emerging trends related to the use of machine learning, Blockchain and Big Data analytics for IoTs. The book is aimed at a broad audience of ICTs, data science, machine learning and cybersecurity researchers interested in the integration of these disruptive technologies and their applications for IoTs.
Inspec keywords: blockchains; decision making; Internet of Things; learning (artificial intelligence); health care
Other keywords: medical information systems; health care; Internet of Things; decision making; data analysis; blockchains; peer-to-peer computing; learning; data privacy; artificial intelligence
Subjects: Machine learning (artificial intelligence); General electrical engineering topics; Data security; Education and training; Mobile, ubiquitous and pervasive computing; Computer communications; Distributed databases; General and management topics
- Book DOI: 10.1049/PBSE016E
- Chapter DOI: 10.1049/PBSE016E
- ISBN: 9781839533396
- e-ISBN: 9781839533402
- Page count: 679
- Format: PDF
-
Front Matter
- + Show details - Hide details
-
p.
(1)
-
1 Introduction to machine learning, blockchain technologies, and Big Data analytics for IoTs: concepts, open issues, and critical challenges
- + Show details - Hide details
-
p.
1
–16
(16)
Today a plethora of applications are trying to shift toward using the Internet of Things (IoT) or Internet Connected Things. IoTs can be used in many applications such as Healthcare, Agriculture, Transportation, and Logistics. These devices can help healthcare professionals take care of patients in real time. However, there are several issues around smart devices such as security, privacy, and trust. Blockchain is the most secured technology today. It is a novel concept to provide a decentralized and distributed structure and is based on a distributed ledger concept. This chapter discusses IoT and its uses in smart AI-based applications with the integration of blockchain.
-
2 Image enhancement on low-light and dark images for object detection using Artificial Intelligence for field practitioners
- + Show details - Hide details
-
p.
17
–41
(25)
In recent times, there is a lot of demand for Artificial Intelligence solutions based on computer vision in various fields. Many solutions like object detection, fault detection, environment description, and scene prediction are helping to solve many real-life problems. But these solutions are dependent on vision-based computations. Generally, all these computations are designed in such a way that environment in each frame is visible and computation performed with data captured from the frame is in visible condition. But in the case of dark condition, the photons count that a camera capture decreases drastically, and the environment may not be visible. In this scenario, the system will fail to compute all the tasks that are dependent on visibility of environment. With the increase in Artificial Intelligence solutions using vision data, it is important to process low-light/dark images and draw intelligence from them. The information and subsequent intelligence available during low-light scenarios can be extracted wisely by our proposed deep learning architecture. The algorithm will process the raw data taken from the camera sensor and provides you the enhanced JPEG images. These enhanced images will be used to train the object detection using TensorFlow lite to detect the objects in the frames. The entire solution will be ported into the mobile devices for capturing the raw data to enhance images and detect the objects on the enhanced images. The proposed chapter will also explain how this solution will be used in the field assistance where user can be able to see the objects in the scene clearly with the enhanced images and detect the objects for machine repair and maintenance of various tasks.
-
3 Cache memory architecture for the convergence of machine learning, Internet of Things (IoT), and blockchain technologies
- + Show details - Hide details
-
p.
43
–73
(31)
This chapter describes the need for cache memory architecture for the convergence of machine learning (ML), the Internet of Things (IoTs), and blockchain technologies with a brief introduction of cache memory and its types. Furthermore, this chapter implements cache memory design for single-bit architecture (CMDSBA). Single-bit architecture comprises six transistors static random access memory cell (SRAMC), a CWD, and latch sense amplifier (LSA) such as voltage latch type sense amplifier, and current latch type sense amplifier that has been implemented and compared on different values of resistance. Results depicted that cache memory design for single-bit six transistor SRAMC (STSRAMC) voltage LSA architecture consumes 14.32 µW of power. Apart from it, to optimize the consumption of power, power reduction sleep transistor technique, power reduction sleepy stack technique, and power reduction dual sleep technique are applied over different blocks of cache memory designed for single-bit architecture and the conclusion arises that single-bit STSRAMC with sleep transistor technique CLSA with sleep stack technique in architecture consumes 9.38 µW of power with 40 number of transistors.
-
4 Machine learning algorithms for Big Data analytics including deep learning
- + Show details - Hide details
-
p.
75
–98
(24)
Owing to recent development in technology, major changes have been noticed in human being's life. Today's lives of human being are becoming more convenient (i.e., in terms of living standard). In current real-world applications, we have shifted our attention from wired devices to wireless devices. As a result, we moved into the era of smart technology, where a lot of Internet devices are connected together in a distributed and decentralized manner. Such Internet-connected devices (ICDs) or Internet of Things (IoTs) engender tremendous data (i.e., via communicating other smart devices). With the tremendous increase in the amount of data, there is a higher requirement to process this huge amount of data (generated through billions of ICDs) using efficient machine learning (ML) algorithms.
In the past decade, we refer data mining algorithms to make some decision from collected data-sets. But, due to increasing data on a large scale, data mining fail to handle this data. So, as substitute of data mining algorithms and to refine this information in an efficient manner, we require tradition analytics algorithms, i.e., ML or data mining algorithms. In current scenario, some of the ML algorithms (available to analysis this data) are supervised (used with labeled data), unsupervised (used with unlabelled data) and semi-supervised (work as reward-based learning). Supervised learning algorithms are like linear regression, classification and k-nearest neighbor (KNN), etc. Whereas, unsupervised learning algorithms are clustering, k-means, etc. In general, ML focuses on building the systems that learn and hence improves with the knowledge and experience. Being the heart of artificial intelligence (AI) and data science, ML is gaining popularity day by day. Several algorithms have already been developed (in the past decade) for processing of data, although this field focuses on developing new learning algorithm for big data computability with minimum complexity (i.e., in terms of time and space). ML algorithms are not only applicable to computer science field but also extend to medical, psychological, marketing, manufacturing, automobile, etc.
On another side, Big Data including deep learning are the two primary and highly demandable fields of data science. A subset of ML, computer vision or AI, deep learning is used here. The large (or massive) amount of data related to a specific domain which forms Big Data (in form of 5 V's like velocity, volume, value, variety, and veracity) contains valuable information related to various fields like marketing, automobile, finance, cyber security, medical, fraud detection, etc. Such real-world applications are creating a lot of information every day. The valuable (i.e., needful or meaningful) information are required to be processed (or retrieved) from analysis of this unstructured/ large amount of data for further processing of the data for future use (or for prediction). Big organizations have to accord with the tremendous volume of data for prediction, classification, decision making, etc. The use of ML algorithms for big data analytics, which extracts the high-level semantics from the valuable (meaningful) information form the data. It uses hierarchical process for efficient processing and retrieving the complex abstraction from the data.
Hence, this chapter discusses several algorithms of ML, to analysis of Big Data. Also, the subset AI like ML algorithms, deep learning algorithms are being discussed here (i.e., to analysis this Big Data for efficient prediction). Later, this chapter focuses on benefits of ML, deep learning algorithms in analyzing tremendous volume of data (i.e., in unsupervised or unstructured form) for numerous complex problems like information retrieval, medical diagnosis, cognitive science, indexing using semantic analysis, data tagging, speech recognition, natural language processing, etc. Also, weakness, raised issues, and challenges (during analysis big data) using (in) ML or deep learning have been discussed in detail. In other words, research gaps in using ML, deep learning algorithms for big data will also be discussed (covering future research aspects/trends). Finally, this chapter discusses the significance of the smart era, computational intelligence, and AI in depth.
-
5 Machine learning-based blockchain technologies for data storage: challenges, and opportunities
- + Show details - Hide details
-
p.
99
–121
(23)
Technical problems have dominated machine learning (ML) research. ML is being increasingly commonly employed in healthcare, and it is assisting patients and physicians in a variety of ways. ML is a pattern recognition technology that may be used on medical pictures. After that, the ML algorithm system determines the optimum combination of these image attributes for categorizing the picture or generating a metric for the specified image area. The goals were to promote fundamental and applied research in the application of ML methods to medical problem solving and research, to provide a forum for reporting significant results, to determine whether ML methods are capable of underpinning research and development on intelligent systems for medical applications, and to identify areas that needed more research. Several research agenda suggestions were presented, covering both technical and human-centered issues.
-
6 Clustering crowdsourced healthcare data from drones using Big Data analytics
- + Show details - Hide details
-
p.
123
–144
(22)
Nowadays, a vast amount of data is considered for statistical analytical tools. Big Data analytics have generated an exponentially increasing variety of medical data from IoT devices such as drones. Today, the epidemiological data collection pattern in disease monitoring is gathering with drones. This includes organized, semi-structured and unstructured data, leading to drone data trawling. This chapter provides big data to increase healthcare quality by applying effective machine learning (ML) strategies for segregation and drones-compilation data. The process intends to resolve the drone's use of trawled data and provide real-time analyses of the data. The ML algorithm implemented the Apache Spark core for smoother segregated streaming from various crowd sources. Three drones were configured for the experiment. The principle will increase the accuracy of health care forecasts based on the investigation findings. This result showed that relative to hierarchical clustering and density-dependent clustering, the K-means cluster has the highest smoothness rate of real-time segregating data.
-
7 Authentication and authorization in cloud computing using blockchain
- + Show details - Hide details
-
p.
145
–163
(19)
Cloud computing is a technology that creates a virtual environment by configuring various computing devices. It provides software, platform as well as infrastructure as a service. Most companies have shifting or have included online operations mode along with offline mode of operation to expand business. Organizations are using cloud services to save on infrastructure and other associative costs. Most cloud services are chargeable on the basic of usage. Authentication is an important aspect of cloud computing. Various authentication algorithms which are presently in use are having some limitations. This chapter reviews various existing authentication and authorization mechanisms used for the cloud environment and will discuss how the blockchain technology with its features like transparency, immutability, traceability can be used to provide authentication in cloud environment.
-
8 Fundamentals of machine learning and blockchain technologies for applications in cybersecurity
- + Show details - Hide details
-
p.
165
–186
(22)
Artificial Intelligence (AI) as a potentially transformative and resource focused on cybersecurity aims to analyze trends in all aspects related to advances in digital threats trying to predict what lies ahead and to anticipate as much as possible these cyber threats by providing a resource through machine learning (ML) that can help companies and public agencies to protect themselves more efficiently, in an ever faster response to cyber-attacks. In addition, the AI can interpret security events to act automatically or even perform the proper screening for security specialists, already adding logs and information that will help in the analysis. AI along with other cybersecurity technologies can be used to assist in discovering the network, identifying vulnerabilities, or generating intrusion scenarios for training cyber defense teams. Another example is learning about digital security done much more quickly and effectively using ML, demonstrating that the use of AI in this process will allow implementing the best security processes without the need to mobilize large operational teams. In this scenario, improving processes and having an increasingly sophisticated base for responding to attacks can help many organizations to protect data effectively. Therefore, this chapter aims to provide an updated research background of cyber-security-related digital privacy and threats to personal data in the context of security in AI and ML, addressing fundamental concepts, showing its relationship with disruptive technologies, with a concise bibliographic background, synthesizing the potential of technology.
-
9 Real-world applications of generative adversarial networks and their role in blockchain technology
- + Show details - Hide details
-
p.
187
–209
(23)
This new term called "Blockchain" has come into the limelight over this decade, significantly impacting almost all major industries' development and efficiency. Blockchain is nothing but just a database. This database stores data blocks that are encrypted and chained together to form a single point of truth (SPOT) for data. Here the digital assets are not transferred or duplicated but are distributed. This creates a rigid record of assets. The asset is localised, giving public access in real-time and transparency. Document's decency is preserved via a transparent record of changes, which builds trust in the asset. A technology of this calibre when blended with another revolutionary technology called Generative Adversarial Networks (GANs) gives an outcome of utmost efficiency. GAN is the hottest topic of recent research. Since 2014, GANs have been widely studied for many years, and there are a large number of algorithms proposed by many researchers. However, very few complete studies explain the link between GAN variability and how it has evolved. In this chapter, we attempt to provide updates on the various uses of GANs in running applications. In this chapter, we discussed the way GANs are used in various real-time applications. We had reviewed various models for estimating generative models used in healthcare, Internet of Things (IoT), credit card fraud detection, and some other applications, in which GANs are used. In addition, GANs are integrated with other machine learning (ML) algorithms for specific applications, such as less supervised Learning, learning transfer, enhanced learning. This chapter compares the similarities and differences in these GAN methods. Second, theoretical issues related to GANs are being investigated. Third, the general use of GANs in ML, image processing, the medical field, and data science is also mentioned. Here we will be mainly focussing on real-time use of GAN such as self-driving, credit card fraud, and field medical. Also, there have been quite a few applications of GANs in blockchain, which we will be discussing in this chapter.
-
10 Internet of Things (IoTs)-enabled security using artificial intelligence and blockchain technologies
- + Show details - Hide details
-
p.
211
–234
(24)
Blockchain is a private key cryptography protocol of network-to-network chain. Blockchain technology is used in many sectors since it improves the value-exchange business to an advanced level. Starting from bitcoin to healthcare, it plays a substantial role in the application. In this real-world, blockchain applications transform society to a higher level by protecting business deals, quick processes, error free, safety, and efficiency. The following are some of the protruding applications of blockchain. Blockchain healthcare is for the secure sharing of medical data. Blockchain music is for royalties tracking, transparency, and ownership rights. Blockchain asset management is a real estate processing platform used for crossborder payments. The blockchain ledger simplifies the whole process, reduces error by encrypting the records along with no intermediaries. The Internet of Things (IoT) is a peer-to-peer computing device. It transfers data without human-computer interaction. IoT with blockchain brings an improvement in the lifestyle through smart cities project. It helps to combat security breaches, and it can control devices when away from home, alert when cookies are ready, etc.
Blockchain plays an important role in supply chain and logistics monitoring. It uses sensors through that it provides location and condition transparency around the world. The supply chain is one of the application areas among blockchain. It has tremendous scope since it is an integral part of the Indian economy. Also, it is the best use-case for blockchain. Other use-cases are a charity, trading, certificate verification, digital identity, and copyright protection. Many mobile-based and online-based applications are available for specific processing benefits, such as Burstiq, Mediachain, Propy, Opskins, Chainalysis, Filament, Hypr, Xage Security, Civic, Evernym, Ocular, Dhl, Maersk, Madhive, Voatz, and Delaware.
Blockchain application demand is fascinating and advancing since it is a new technology. Every year, it increases the market growth. Blockchain technology and crypto economics are the heart of many growing industries. Blockchain applications are the solution to improve data integrity to the highest standards. Their technology has the potential to revolutionize big data with better security and data quality.
-
11 Blockchain network with artificial intelligence - DeFi affair management
- + Show details - Hide details
-
p.
235
–255
(21)
To create a completely decentralized exchange platform using blockchain technology that allows companies and investors to be able to tokenize real-world assets and provide them with financial services like yield farming, lending, and borrowing. Tailored commendations, announcements, and propositions can be done with decisions by artificial intelligence. Artificial intelligence plays a key role in analysis and prediction added to excellent decision-making skills on par with human intelligence. Blockchain technology proves its efficiency in high interpretability. Both blockchain and artificial intelligence compensate each other in its own feature. This chapter provides a detail description about useful topics like online education, finance, etc.
-
12 Vulnerabilities of smart contracts and solutions
- + Show details - Hide details
-
p.
257
–276
(20)
In recent decade, blockchain concept was proposed in cryptocurrency, completion of transition without any third party. Today, this concept is being used in all possible applications like e-healthcare, transportation, communication, etc. In general, blockchain technology uses the concept of smart contract to transfer any money to end user. Smart contracts provide efficient performance of credible transactions (trackable and irreversible functionality) without involving third parties. These contracts are self-executing contracts with the terms of the agreement between buyer and seller. Note that smart contracts have distributed, decentralized blockchain network. Some applications of smart contacts are the following: land records, cryptocurrencies like Bitcoin, Ethereum, etc. With this much popularity of smart contract, today we are facing many vulnerabilities on such contract, for example, today's potential vulnerabilities are nearly 30% on smart contracts (examined by national university of Singapore). It is a critical issue and legitimate (valid) cause for having concern in the blockchain technology and we should take care about such critical issue because people's money or information are being shared through such smart contracts. In general, security vulnerabilities in software can be fixed by patching, but patching security vulnerabilities of decentralized applications on the Ethereum blockchain is not so straightforward. Due to the immutable nature of smart contracts, it is very difficult to upgrade the already deployed contracts. But, having the availability of smart contract as public, chances of vulnerability are higher on stored smart contracts. Hence, this chapter provides several mitigated critical smart contract vulnerabilities (in the past decade), possible technique to detect vulnerabilities on smart contract, and possible suggestions including opportunity (and research gaps) for future toward protecting smart contract code or agreement.
-
13 Data analytics for socio-economic factors affecting crime rates
- + Show details - Hide details
-
p.
277
–297
(21)
Crime has always been a major issue in the society. The places having high crime rate affect social harmony, tourism, and several factors related to economy. This study tries to understand the socio-economic factors that affect the crime rate. In all countries, the blunder complication happens due to the fact that the crime occurrence signifies that there is a social phenomenon which reflects all the evidence that lead crime in the same way with minor changes in those areas. Thus, the proposed work demonstrates the analytics over a standard dataset of crime covering all the factors. And also due to the wider availability of the event/incidence of crime rate reported in the crime dataset of India. This work targeted to consider the socio-economic factors affecting crime rates in India. The post-independence data of 13 most populous states according to the 2011 survey has been collected. This study compares the socio-economic and criminal data of different states using the data visualization method and tries to understand factors that are highly responsible for increased rate of reporting of crime in these states. The factors such as Gender Ratio, Human Development Index (HDI), Anti-Corruption Efforts, Literacy Rate, Per Capita Income, Unemployment Rate, and Poverty Index were taken into account. This work also attempts to apply regression model on the dataset to predict the crime rate. The result of this study is that these factors affect the reporting of crime in India and, by working on these factors, will increase the beneficiaries of government policies.
-
14 Deployment of automated teller machinery for e-polling
- + Show details - Hide details
-
p.
299
–321
(23)
Voting plays a significant role in electing a suitable person by the public to rule the country. In this modern age of digitalization, we still use electronic voting machine (EVM), which is not entirely automated and has got many limitations. This machine replaces the traditional way of collecting votes through boxes consisting of voting papers. This method is called as the paper ballot. This system has slowly changed or altered into a new technical and mechanical system that uses EVMs to avoid misconceptions. Even though this voting machine is fast and accurate, this system needs more work-force and also it is not much more reliable. Most of the eligible voters cannot cast their vote due to immigration for a job, education, etc. Also, the government spends a lot of money arranging the vote booths and maintaining the decorum of the election.
This chapter proposes a framework that changes the automated teller machine (ATM) framework to function as EVM. One of the fundamental advantages of utilizing ATMs as EVM is encouraging the political decision process for everybody and letting immigrants vote for their constituency. One of the main benefits of using ATMs as EVM is to facilitate the election process for everyone. Also, using ATMs as EVM will save time for all voters and eliminate the waste of your time in the long queue for the regular election process. Any voter needs just to hit for the closest ATM in his\her area and afterward utilizing electronic virtual card (EVC) embeddings it into the ATM picks the choice political decision.
-
15 Machine learning-based blockchain technology for protection and privacy against intrusion attacks in intelligent transportation systems
- + Show details - Hide details
-
p.
323
–349
(27)
Intelligent transportation system (ITS) is a rapidly growing field of technology that combines network connectivity, modern sensors, control system, and data processing technologies to improve our daily life. With the growing popularity of ITS, concerns about its security have garnered considerable attention. SQL injection, denial of service, and ransomware assaults are all prevalent types of threats in an ITS. In automation/transportation systems, privacy and trust are also significant challenges. Today, everyone requires a vehicle to move around. Together with this, data security is crucial in a computerization system since the vehicle's user data is transferred to the extra operator via the web through wireless devices and routes such as radio channels, optical fiber, and so on. Certainly, every device is linked to the internet and one another, constituting the Internet of Things (IoT). Even as the network transitions to wireless devices, numerous risks to autonomous cars/vehicles have developed into a serious issue for service providers and car owners. The bulk of these attacks is detectable and preventable using a variety of intrusion detection methods. The blockchain and, more broadly, peer-to-peer techniques may be critical in the development of decentralized and data-intensive applications that run on billions of devices while maintaining user privacy. To solve these problems, this chapter proposes a machine learning (ML)-based blockchain intrusion detection for protection and privacy against intrusion attacks in ITS. The blockchain was used for aiding information exchange in ITS. As a result of the immutable and decentralized nature of blockchain-enabled ITS systems, a variety of desirable qualities such as security, decentralization, transparency, automation, and immutability are expected to exist. The experimental analysis was performed on the UNSWB-NB15 dataset. The results obtained reached an accuracy that is more than 99%. The AUC, recall, Mathew Correlation Coefficient (MCC), and training time metrics were also used to evaluate the performance of the models.
-
16 Blockchain-enabled Internet of Things (IoTs) platforms for vehicle sensing and transportation monitoring
- + Show details - Hide details
-
p.
351
–373
(23)
Technical evolution of Internet of Things (IoTs) shifts its power, storage, and computational capabilities away from centralized cloud platforms to a decentralized IoT edge. Traditional security policies may not be effective against modification attacks, eavesdropping, distributed denial-of-service (DDoS) because of large attack surface area in IoT where devices cannot have their own ID and cannot keep their privacy and resiliency to the cyber-attacks. Blockchain can support secure, transparent, immutable data and computation-intensive applications such as services for factory automation, assisted living and automotive driving, transportation monitoring and vehicular networks. In addition, for facilitating blockchain applications at low-power mobile IoT systems, mobile edge computing (MEC) can be a convenient alternative for solving consensus protocols for mobile IoT users (offloading to MEC providers).
In this chapter, blockchain-enabled IoT platforms and solutions for vehicle sensing and transportation monitoring from various aspects are investigated. After providing a more general background, we consider vehicular networks and blockchain applications. In a typical vehicular network, a large number of vehicles need ultra-reliable, low-latency communications and secure, transparent, immutable data sharing to avoid multiple-vehicle collisions. Therefore, we consider blockchain applications in vehicular networks for more scalable, transparent, and secure Internet of Vehicles (IoV). In addition, we consider cybersecurity issues in Internet of Drones (IoD) and unmanned traffic monitoring (UTM) systems and then provide the existing blockchain-based solutions/platforms for these problems.
With 5G of mobile broadband systems, blockchain needs to deal with scalability problems due to a very large number of users in IoV. Artificial intelligence (AI) techniques can be applied with blockchain in connected vehicles in this manner. Implementing AI techniques can provide more scalable, transparent, and secure blockchain applications in vehicular networks. AI techniques also help blockchain achieve privacy and personalization for the users in IoV at the same time. In the sequel, we consider more lightweight protocols for more scalable blockchains. Moreover, we consider the solutions of game-theoretic models to jointly maximize the profit of the MEC service provider and the individual utilities of the miners. Then, we can consider resource-provision problems under different pricing schemes offered by MEC service providers. Finally, we conclude the chapter.
-
17 Blockchain-enabled Internet of Things (IoTs) platforms for the healthcare sector
- + Show details - Hide details
-
p.
375
–398
(24)
In the last decade, world has changed gradually from mechanical to smart digital world. Defense, supply chain, industries, finance, healthcare, education, automation, and more applications are automated by the use of sensors. The world is built like a small village with inclusion of smart devices. Also, the latest developments in information technology and blockchain technology have enhanced market of the electronic healthcare sector. In healthcare, Internet of Things (IoT) provides real-time sensory data from patients to process and analyze, which requires either centralized or decentralized computation, processing or storage. For various sectors including finance, defense, healthcare, and many more, the security concerns during transmission or processing needs are very important. Security concerns include single points of failure, distrust, data tampering, and privacy evasion of IoT platforms. For healthcare systems in particular, maintaining the privacy of data is most important. Hence, security problems require resolution without relying on trusted third parties. Blockchain provides security by exchanging digital tokens and transparency of data among health sector members in a peer-to-peer network, avoiding the requisite of trusting centralized third-party entity. In this chapter, we will survey solutions to security related issues of healthcare data in IoT platforms. Future context of IoT platforms analysis for sensor data scenarios consider different healthcare applications with Blockchain. Finally, we discuss future scope and challenges in the health sector.
-
18 An integrated dimensionality reduction model for classifying IoT-enabled smart healthcare genomic data
- + Show details - Hide details
-
p.
399
–417
(19)
Machine Learning (ML) is now a powerful factor in everyday life, and in most fields that we desire to improve. ML is a field for creating systems that can learn from data, whether labeled or unlabeled, or from the ambient. ML is employed in various of disciplines, but it is incredibly useful in the healthcare sector since it leads to improved decision-making and prediction approaches. Since ML in healthcare services is scientific research, we need to save, obtain, and properly utilize data, knowledge, and provide expertise to the issues that face the healthcare industry, as well as learning for proper decision-making. Owing to most of these innovations there is indeed a big development in health care sectors over the decades. Healthcare analysis of data has become one of the greatest favorable research fields. Healthcare includes data from diverse kinds with medical data, functional genomic data, in addition sensor data, obtained through involvement of various wearable and wireless sensor devices. Manually processing this relevant data is really challenging. ML has developed as a crucial data analysis technique. Health professionals employ these tools and techniques to analyze healthcare data to identify hazards as well as provide effective diagnosis and management. In this study, a dimensionality reduction model is suggested for classifying IoT enabled healthcare data analysis, using a COVID-19 dataset with partial least square (PLS), linear discriminant analysis (LDA) on support vector machine (SVM) and Kth nearest neighbor (KNN) classifiers. This study uses COVID-19 dataset, ML approach such as PLS, LDA, SVM, and KNN on a MATLAB tool for the analysis.
The result obtained shows that PLS-LDA-SVM outperformed PLS-LDA-KNN with 90% accuracy. The review of this study has proven that this study is efficient for decision making by practitioners to adopt for efficient analysis of healthcare data analysis.
-
19 Blockchain-based learning automated analytics platform in telemedicine
- + Show details - Hide details
-
p.
419
–446
(28)
Telemedicine - a field having immense potential to improve the landscape of healthcare domain with its innovative and interactive concept of e-cure. Having created its niche in the healthcare ecosystem, it has now become an integral part of it after the genesis of COVID-19. The world has witnessed the rapid growth of telemedicine according to their literacy and technological maturity but the rate of adopting varies between countries. Telemedicine is becoming the boom in India's healthcare industry where 65% of the population resides in rural and remote areas. It delivered essential healthcare at a critical time where traditional medicine is perceived to be at risk as a result of the nature of the infectious disease like COVID-19. This communicable infection has shifted the paradigm of telemedicine to necessity with providing solutions and forming a cushion at the demand-supply shock and also has a potential to take care of the skewed doctor and patient allocation which exist in India, where for every 10 lakh patients, there are six doctors. With its increasing demand in this new normal and across all ages, genres and geographies, had made the organizations and many new age start-ups not only strive toward improved services in this area but also for the storage and security of its digital assets through a variety of technology tools and keeping this in mind the aim of this chapter is to design a framework to deploy Blockchain technology to protect the most important element of telemedicine that is data. Deployment of blockchain in telemedicine, still in its infancy, will be a boon for the confidentiality and data security while automated analytics can be instrumental not only to improving the service quality, but also to make it quick and personalized [33]. This chapter is an attempt to bring to light all these vital aspects of telemedicine domain in healthcare.
-
20 A sensor-based architecture for telemedical and environmental air pollution monitoring system using 5G and blockchain
- + Show details - Hide details
-
p.
447
–473
(27)
Within the recent advancements in technology, there is a tremendous growth in digital healthcare technologies for refining and transmuting healthcare which takes the account of plummeting human miscalculations, enlightening clinical consequences, facilitating care coordination, and humanizing practice efficiencies, with the help of integrated approaches. To create a more effective and safer health care environment in the places where there is no medical health care facility, Internet of Medical Things (IoMT) can provide a Tele medical room with a 5G infrastructure. The Tele medical room contains a smart chair facility, wherein the data of patient denotes heart rate of the patient, electrocardiography, blood pressure of the patient, and body temperature of the patients that can be monitored remotely and the assistance can be provided by the doctors through remote conferencing. The 5G-enabled communication environment is created to withstand higher data transmission in the remote correspondence interface.
During the recent decade, we have seen an incredible improvement concerning miniaturized electronic devices, in order to facilitate their integration in handheld equipment. This invention demonstrates the estimation of blood glucose level by designing a substrate-based flexible antenna. The change in the concentration of raw material provides moderate dielectric constant that is attained in the flexible antenna. Synthesized nickel particle-based flexible nickel aluminate (NiAl2O4) is utilized as a substrate material for microwave applications. The substrate was made with 42% of nickel, with a dielectric constant of 4.979 and a thickness of 1 mm. Design and performance analyses were investigated based on the reflection coefficient in normal and bent conditions. Dielectric properties of human blood over a broad frequency range were measured with and without adding the glucose content for different types of blood groups. These results were manipulated as a data set required for machine learning approaches.
The optimized Ensemble algorithm which provides feature reduction, hyper parameter optimization, along with asymmetrical cost assignment can be used to segregate the normal and abnormal heart beat signal. The receiver unit comprises a headphone jack which can be inbuilt with a Wi-Fi modem in order to clearly hear the pulmonary, cardiac, or digestive noises being monitored. The Internet of Things (IoT) sensor information is created from different heterogeneous gadgets, correspondence conventions, and information organizes that are gigantic in nature. This gigantic measure of information is not incorporated and examined physically. This is a huge issue for IoT application designers to make the mix of IoT sensor information. Semantic explanation of IoT information is the establishment of IoT semantics. Bunching is one way of settling the combination and investigation of IoT sensor information. Semantics and learning approaches are the keys to resolve the issue of sensor information combination and investigation in IoT.
-
21 Blockchain-enabled Internet of Things (IoT) platforms for financial services
- + Show details - Hide details
-
p.
475
–497
(23)
Blockchain technology is considered a breakthrough since the advent of Internet. This technology can be used financially for the Internet of Things (IoT)-based commerce platform building and marketing platform is a new business model based on the management of private transactions on IoT devices technologies that can manage a shared network using current blockchain solutions, most of which are digital currencies. The emerging popularity of digital currency makes them acceptable as one of the payment sources. The use of IoT-based blockchain technology such as social procurement, e-commerce blockchain applications and use cases describes various aspects of finance, including payments, security, supply chains, automation of work with smart contracts, and ethical practices for financial transaction transparency. With the rapid development of cryptography and the proliferation of computers systems, blockchain technology is widely expected to transform many industries with better clarity, higher security, and lower purchase costs. This chapter shall provide IoT-based blockchain technology use in ecommerce applications such as social shopping. The chapter shall provide the number of impacts of learning bodies, platform operators, and developers of blockchain technology in the finance field. The chapter discusses the use of IoT and blockchain technology in the finance industry. Blockchain applications and use cases are discussed for various aspects of finance like payment, security, supply chain, work automation with smart contract, ethical practices for transparency in financial transactions.
-
22 Blockchain and machine learning: an approach for predicting the commodity prices
- + Show details - Hide details
-
p.
499
–516
(18)
In today's environment, visualizing data is getting attraction from various organizations and sectors. Visualization of the commodities data on individual countries are correlated with GDP and the development of their individuals. Machine learning (ML) deals with the data to perform the Mathematical and Statistical operations to find exciting patterns present in the data. ML models rely on different algorithms and techniques used to make predictions on the collected datasets that are used for prediction and try to provide stakeholders an overview of growth and development. Moreover, prediction of commodity prices will significantly determine the economic status of the Country. This chapter presents a forecast of commodity prices, especially crude oil, gold, and wheat, by ML techniques. We have built a neural ML model to predict the respective commodity prices. Results presented in this work are helpful for the students, researchers, etc., working in the area of ML and their applications to understand the overall working process, steps in determining or building effective ML models for any datasets in real-time.
-
23 Knowledge extraction from abnormal stock returns: evidence from Indian stock market
- + Show details - Hide details
-
p.
517
–535
(19)
In perceptive, the stock market seems to be enigmatic, with days in which the stock values burgeon unparallel to the initial stocks and days it shrivels to plummet like a falling knife nobody wants to catch. In contrast, companies put their best foot forward to boost investors' morale and keep their company aloft in the ever dynamically changing market by declaring dividends. In its vital sense, data analytics aims to scrutinize and model this dynamic scenario to assist with future prediction and overall decision making. This chapter thus fixates upon one of the applications of big data in finance. This chapter mainly examines abnormality in stock returns pre and post-dividend declaration using data analytics. Our sample includes dividend-paying companies featuring in Nifty Mid-Cap, for the financial year 2018–2019. Statistical validation of actual and predicted returns obtained by regression coefficients through a series of regression runs can help one converge to an inference of abnormality on the dividend declaration. The study reveals an abnormality between actual and predicted stock returns on the first-day post-declaration of dividend. The study provides valuable knowledge extraction to understand the behavior of stock market participants over abnormality in returns using big data.
-
24 Impact of influence analysis of social media fake news - a machine learning perspective
- + Show details - Hide details
-
p.
537
–561
(25)
Nowadays, fake news and fake user profiles spread across online communities, such as social media. This fake news is becoming more popular and influencing people's opinions, emotions, and behaviors due to information diffusion among social media users. Social media influence analysis becomes more prevalent in the activities of billions of users on a day-to-day basis. Sometimes this fake news is spread by posts using some fake user profile. This work predicts fake news in social media and fake profiles using machine learning algorithms. We calculate the influence study of this fake news by predicting the online user behavior analysis. The experimental study evaluates the effects of such news propagation on online users and uses simple prediction techniques. We analyze the influence of fake news on public psychological analysis by using machine learning techniques. The proposed system analyses various possible real and fake news indicators and implementation that correctly represent the classifications. The outcomes show that the influence score seems to be more efficient in determining actual essential factors.
-
25 Application of machine learning techniques based on real-time images for site specific insect pest and disease management of crops
- + Show details - Hide details
-
p.
563
–590
(28)
This research work is concerned with the innovative methodology of the deep learning (DL) (convolutional neural network) techniques, where we implemented a novel approach to accurately recognize the mung bean (Vigna radiata) pests and diseases. Our methodology significantly increases the mean Average Precision (mAP) to 77.16% (approx.); in comparison with the existing models, the final result improved by 9.8% approximately. In this chapter, we sophisticatedly instill our designed model to the Android-based mobile platforms which are easy to operate for the farmers and experts. Our advanced application smoothly identifies real-time pests and diseases and deliver necessary measurements on pest and disease control. So, we have assembled a healthy and disease and pest infected mung bean crop imaging database with eleven typical mung bean varieties (WBU 109 (Sulata), IPU 02-43, NUL 7, LU 391, KUG 479, VBG 04-008, TU 40, LBG 787, VBN 8, SONALI, and PANNA). The acquired database was trained by five consecutive models, finally, single shot detector (SSD) with inception optimized the model. Furthermore, the data augmentation (DA) method with a dropout layer was employed for achieving high mAP value. We developed an Android application where we performed the experiment and the designed model application appropriately illustrate and recognized images after trial and testing. In our database, there are different types of images collected like a plenty of images are diseases and pests affected and the rest images are healthy samples. In a few images, we can properly identify the indistinct objects, which specifies when the pests or the disease affected portion is blurred in that particular image, our application accurately identified that portion from that image. Few images show the diverse postures of disease, and in many images, we have displayed that the variety of colors of each sample. So, our methodology indicated that our model efficiently operates for different shapes and different colors of the several classes of the image data set. The two main elements in multi-object identification are feature extraction and occlusion processing, which have been included in our work. The experimental result shows that our methodology is very useful both for pest and disease control and performs superior to the many more advanced or existing models. This application is simple and requires low cost and is improved in terms of ecological compliance, turnaround time, and correctness by color and texture compared with earlier works. This makes the proposed method suitable for pest and disease monitoring tasks through drones and the Internet of Things (IoT). In this work, we have also given attention towards the commercial perspective, because the pulse is a high-priced commodity and the cultivation of pulses are restricted due to excessive insect pest and disease impedances, particularly in mung bean in India. Our approach based on machine learning interventions provides immense benefit to the farmers to tackle the disease and insect pests and to increase the productivity of mung bean in India.
-
26 A prioritized potential framework for combined computing technologies: IoT, Machine Learning, and blockchain
- + Show details - Hide details
-
p.
591
–613
(23)
In today's scenario, every organization is dependent on the use of computers for automating business and work. The data generated by the organization is huge and has great significance in analyzing the processes. It can be rightfully called as the lifeblood of any organization. The world of computing requires three technologies: Internet of Things (IoT), machine learning (ML), and blockchain. The confluence of these technologies is inevitable in the coming future due to the benefits which are involved in future applications. There is barely any activity today which is cannot apply the use of these three technologies e.g., healthcare, automation, education, etc. IoT can be defined as interconnection of various autonomous devices which are capable of communicating with each other. IoT requires an intermittent Internet connection and address for every device. The user of these devices can remotely monitor and manage by retrieving the information on a handheld device similar to the cellphone. This way, the devices are connected 24/7 and continuously generating data. Challenges of IoT involve the following: security, connectivity problems, and huge data. The devices can be made capable of taking intelligent decisions by incorporating Artificial Intelligence (AI) and ML technology. ML is a sub-branch of AI. It has got huge potential to detect the patterns and anomalies in the data which is generated by the wireless sensor nodes in IoT. The advanced decision-making process of ML has already influenced our daily routines, for example: banking, healthcare, gaming, transportation, and space exploration. Challenges faced by ML are the following: security, centralized architecture, and resource limitations. To cover the security aspect of these two technologies, blockchain technology is the perfect answer. It is a decentralized peer-to-peer network which stores the records and transactions in blocks which cannot be altered. This technology secures the communication by eliminating the need for any trusted third party. The blocks are stored in such a manner which makes it impossible to hack or tamper the data by taking control of device or capturing the records. This chapter will present a comprehensive and quantitative analysis of the existing research and how these technologies can be a transformative impact for access to information by the users. The convergence of blockchain, ML, and IoT will provide scalable, secure high-level intellectual functioning that will be the new paradigm of digital information. This book chapter presents futuristic potential of convergence of three technologies and elaborate discussion of the past researches.
-
27 Conclusion to this book
- + Show details - Hide details
-
p.
615
–623
(9)
As discussed in previous chapters, we can find out that the integration of these emerging technologies definitely will change the way of living, working in the next decade. But together these technologies get serious challenges like leaking of personal information, habits by IoT devices, require high energy for machine learning (ML) to refine big data, require more storage in cloud computing and many data center to store this data, security of this data by blockchain and providing authentication process to these integrated (ML + blockchain + Internet of Things (IoT)) systems, and many more. Hence, this section discusses few faced critical challenges in the integration of these technologies as:
-
Back Matter
- + Show details - Hide details
-
p.
(1)