Industrial Internet of Things (IIoT). Группа авторов

which before were not available in real time and are now needed [8–11].

Thus, the architectures of industrial automation systems, which have adherence to Industry 4.0, manage to integrate different devices in favor of industrial evolution, with more and more sensors, cameras, and systems that will be monitoring the entire industrial production process, evaluating and supervising the performance of equipment, and providing, in addition to the already known layers of operational control and the entire control framework, the IoT and IIoT layer, where it will converge all this data into a Big Data, delivering operational control possibilities (Figure 1.2), with decision-making in prognoses and with the possibility of autonomous actions [10–12].

      Optimizing the production process of the industry is the main reason for the application of IoT in the production line of the factories, since the IoT technology and its IIoT aspect allows the equipment that makes up the industrial yard of a company today that can be connected in a network. With the data collected and stored in the cloud, it allows the decision-makers of the companies to have quick and easy access to all the information of the company and its collaborators; in other words, this makes all the industrial machinery work automatically through of highly programmable intelligent sensors [13, 14].

Figure 1.2 Big data illustration.

      Wherefore, this chapter is motivated and has the purpose to originate an updated overview of IoT and IIoT, addressing its evolution and branch of application potential in the industry, approaching its relationship with current technologies and synthesizing the potential of technology with a concise bibliographic background.

1.2 Relationship Between Artificial Intelligence and IoT

      The emergence of solutions and tools with AI (Artificial Intelligence) technology means solutions, tools, and software that have integrated resources that automate the process of making algorithmic decisions. The technology to be used can be anything from independent databases employing Machine Learning to pre-built models that can be employed to a diversity of data sets to solve paradigms related to image recognition and text analysis. Applied in the industry, it can help a business achieve a faster time to evaluate, reduce costs, increase productivity, and improve the relationship with stakeholders and customers [15, 16].

      Machine Learning is only part of AI, that is, it is an AI application in which it accesses a large volume of data and learns from it automatically, without human intervention. This is what happens in the case of recommendations on video streaming platforms and facial recognition in photos on social media pages. AI is a broader concept that, in addition to Machine Learning, includes technologies such as natural language processing, neural networks, inference algorithms, and deep learning, in order to achieve reasoning and performance similar to that of human beings [15, 16].

      An AI system is not only sufficient and capable of storing, analyze, and manipulating data, but also of acquiring, representing, and manipulating information and knowledge. Including the characteristic to infer or even deduce new knowledge, new relationships between data-generating information about facts and concepts, from existing information and knowledge and to use methods and procedures of representation, statistical analysis, and manipulation to solve complex questions that are often incognito and non-quantitative in nature [17].

The increase in mass data collection over the years, related to IoT devices, has boosted AI, given that the volume of information produced by people has been growing exponentially. But allied with Big Data technology to understand this massive set of data, which serves as a basis for learning the most diverse software, such as Machine Learning. This data revolution favored the AI scenario, i.e., with more information available, more intelligent, and automated ways to process, analyze, and use the data [18, 19].

      Big data is the term employed to refer to the enormous amount of data that is produced and stored daily, evaluating that from this abundance of information, there are intelligent systems created to organize, analyze, and interpret (that is, process) the data, which are generated by multiple sources [19, 20], still pondering on predictive analysis as the ability to identify the probability of future results based on data, statistical algorithms, and machine learning techniques. From Big Data, it is possible to do this type of analysis, identifying trends, predicting behaviors, and helping to better understand current and future needs and, finally, to qualify decision-making in machines, equipment, and software, taking technology to a new level. AI is impacting society with machine learning systems, neural networks, voice recognition, predictive analysis, and natural language processing (NLP) and continuously remodeling new aspects of human life [19, 20].

      Forecasting and adaptation are possible through algorithms that discover programmed data patterns, the solutions learn and apply their knowledge for future predictions. If a sequence of bits exists, then the AI recognizes the sequence and predicts its continuity. This is also able to correct spelling errors or predict what a user will type or even estimate time and traffic on certain routes in transit (autonomous vehicles based on AI) [17].

      Decision-making through data analysis, learning, and obtaining new insights is able to predict or conjecture a more detailed and faster decision than a human being. But it helps to increase human intelligence and people’s productivity. Through continuous learning, AI can be considered a machine capable of learning from standards [21].

      Also related to its characteristics in the ability to build analytical models from algorithms, learning to perform tasks through countless rounds of trial and error. In the same sense, NLP provides machines and computing devices the capability to “read” and even “understand” human language [22].

      1.2.1 AI Concept

Another characteristic of the basic types of AI is purely reactive, since it acts after the perception of the problem, exemplifying an AI software that identifies the chess pieces on the board and their movement, but has no memory of past movements, ignoring everything before the current movement, that is, it only reacts to the position of the pieces on the board. In the legal field, lawyers focus on more complex aspects of law practice, given the use of text analysis, Jurimetrics, text review, data mining, contract analysis, computational argumentation, and other possible AI-derived features [17, 23], still pondering the characteristics of AI-related to its capacity for intelligent perception, such as visual perception, speech perception, auditory perception, and processing and learning of perceptual information. Reflecting on autonomous cars and virtual assistants, there is not only a programmed answer to specific questions but answers that are more personalized [23–25].

      Through AI solutions, it is possible to eliminate boring tasks that may be necessary, but with machine learning, it performs basic tasks, considered human-computer interaction technologies, or even related to the more robust use found in conversational interfaces that use machine learning to understand and meet customer needs [23–25].

      Even through AI solutions, it is possible to concentrate diffuse problems where data inform all levels of the operation of a modern company, i.e., it has a lot of material to interpret, so it is necessary to consume this amount of information at scale. Since the extent of the data available today has gone beyond what humans are capable of synthesizing, making it a perfect job for machine learning. Through the data, the information is extracted from various sources of public and private data, still comparing them and making changes when necessary [25].

      Through AI solutions, it is possible to distribute data, given that modern cybersecurity leads to the need to compare terabytes of internal data with a quantity of external data. With machine learning, it can automate the process of detecting attacks as cybersecurity problems change and increase, vital for dealing with distributed data problems, assessing that humans are unable to involve their actions around a distribution so wide of information. AI solves