Internet of Everything and Machine Learning Applications: Issues and Challenges

Omid Ameri Sianaki* , Ashkan Yousefi ? , Azadeh Rajabian TabeshΦ, Mehregan Mahdavi§ * § Department of Science and Engineering, ? IEOR, ΦSchool of Marketing *

§{omid.amerisianaki, Mehregan.Mahdavi}@vu.edu.au, ? [email protected], Φ[email protected] *

§ Victoria University, ?UC Berkeley, ΦCurtin University

Abstract -

The Internet of Things (IoT) is recognized as one of the major key areas of future technology and is gaining vast attention from an extensive range of industries. The sensors and devices are generating massive amounts of high-dimensional and heterogeneous data that need to be stored and processed. Machine learning encompasses the widespread techniques of artificial intelligence that can deduce patterns and relationships from unstructured data. Big data analytics are advanced statistical and predictive analytic methods which are capable of manipulating data in a range of Exabytes and more. This paper presents a review of the applications of the Internet of Everything and the machine learning techniques in the fields of health, smart electrical grid, and supply chain management. A review of the literature has been conducted to demonstrate the future issues and challenges that researchers will face. Keywords—Internet of Everything; machine learning; eHealth; smart grid; supply chain management

I- INTRODUCTION

To date, the economic, architectural and fundamental changes to the traditional internet have revolutionised not only the social environment but also the business ecosystem. However, the transformation of a large set of unstructured and streamed data to knowledge is still a challenge ahead of researchers even though the cyber-physical infrastructure, sensor technologies and telecommunication systems have collected and transmitted the data. Machine learning encompasses the widespread techniques of artificial intelligence that can deduce relationships from unstructured data which can be provided by IoT without needing to define them in advance [1]. For example, in the health sector, the knowledge acquisition and actionable perceptions from the complex, multi-dimensional and heterogeneous data is a major challenge for healthcare. Several Internet and network-based paradigms are under the internet of everything (IoE) umbrella, such as Internet of Things (IoT), Internet of People (IoP), and Industrial Internet (II) [2]. IoE is a modern smart technology paradigm envisioned as a universal connection network of machines and smart devices capable of networking with each other, business processes, people, and social environment.

The IoE is recognized as one of the major key areas of future technology and is gaining vast attention from an extensive range of industries [3]. IoE sensors and devices are generating massive amounts of high-dimensional and heterogeneous data that need to be stored and processed. The big data analytics are advanced statistical and predictive analytic methods which are capable of manipulating data in a range of Exabytes and more. This paper will explore the challenges and issues of IoE and machine learning techniques in different business domains such as health management, smart grid and supply chain management. The remainder of the paper is structured as follows: in the next section, the application of IoT in health management is discussed. Section III explains the issues and challenges associated with the smart electrical grid. In Section IV, the effect of IoT in supply chain management.

II- HEALTHCARE

A) Issues and Challenges

The aging population is putting pressure on the healthcare budget and new solutions are required to address the scarcity of healthcare resources. One of the promising technologies that may solve this problem is the Internet of Things (IoT). The IoT role in healthcare can be considered as a facilitator for monitoring, diagnostic and even the possibility of tele-surgery via the Internet. The challenges regarding the implementation of smart and connected devices for healthcare in remote areas are discussed in [4, 5]. Considering the wide distribution of IoT devices in the healthcare sector, one of the main challenges will be related to the identification and security of the nodes. The security of the nodes is a major issue as the received information needs to be detected and then assigned to the correct node with the guarantee of no manipulation. The security of the nodes is particularly important as the system can be affected by malicious activities and the sensitive data can be manipulated or lost. One solution for this problem could be the use of trusted execution environments (TEE) [6,7]. The research related to the security of the IoT devices needs to address the following features: 1) A global ID method to locate the items effectively and efficiently 2) Identity management to be able to secure the full cycle of the encoding/encryption, authentication, and repository management.

The other issue which needs to be addressed, particularly in the health sector, is the management of the telecommunication technology. A selection of the most optimized combination for the telecommunication technologies could significantly reduce the downtime of the device and maximize the reliability of the patient monitoring systems [8]. The next challenge for the healthcare IoT is the development of a location tracking technology. The high penetration of the IoT devices in the healthcare industry demands real-time location tracking systems. An effective technology in this area could be a combination of global positioning systems (GPS) and the Internet of Things. The addition of local positioning tracking systems is necessary to improve the accuracy of the location tracking [9]. The other challenge is related to the continuous acquisition of data from patients, which could significantly increase the diagnostic accuracy as a result of the availability of a massive amount of data. Some of the traditional data acquisition methods include heart and blood oxygen saturation detections. Some of the recent developments enable the extensive use of instruments such as accelerometers, gyroscopes, and surface electrodes to record the data [10]. The challenge here is the management of the big data and the compatibility of the different data architectures. Extensive research is required for the design and integration of the applications [11]. The other challenge which needs to be considered is the cost of storage. Although the cost of storage is rapidly decreasing, the collection and storage of a tremendous amount of data is still a very costly exercise. There is a strong need for the development of intelligent algorithms that can distinguish the redundant data and remove it from unnecessary storage [12]. 

B) Research Trend

In this section, we referred to the Web of Science (WoS) database to discover the research in the health sector, considering IoT and machine learning topics since 2000. As shown in Fig.1, the research fields are shown in different colours and the size of the spheres shows the volume of papers published in each field. For example, keywords such as security, healthcare system, communication technology, IoT technology and RFID are in green colour, indicating that most of the research has focused on security. However, the second category comprises keywords such as heartrate, wearable devices, detection, health monitoring, real time, and cloud computing. These are categorized in another group indicated in red; among them, cloud computing and health monitoring have attracted the most research. In general, the trend of research in IoT and health among 1000 published papers between 2000 and 2017 is towards security, protocol, cloud computing, health monitoring, detection and wireless networks. We have conducted the same research on machine learning and health topics. The result of the correlation between the research subjects is shown in Fig. 2. The result shows that 2190 papers have been published between 2000 and 2017 and these can be divided into five categories. Among various machine learning methods, the support vector machine, random forest, and logistic regression are the main methods studied in the researches. Fig. 2 demonstrates that there is a significant relationship between artificial intelligence, big data, expert system, and ontology and health topics such as infection, genomic, intensive unit care, bioinformatics, and pathogens. However, the logistic regression technique has been applied mainly in research on health topics such as depression, stress, and anxiety. We used the VOSviewer software to analyze the data [13]. 

Fig1. The bibliometric networks in IoT and Health topics between 200 and 2017

Fig.2 The bibliometric networks in Machine Learning and Health topics

III. MACHINE LEARNING APPLICATIONS IN SMART GRID

Machine learning is an emerging tool in various fields and can be used extensively from research to commercial applications. In particular, the fast-growing applications of deep learning have increased the development of machine learning. Consequently, deep learning has become a hot research topic in research organizations and tech companies. In general, deep learning uses a multi-layer neural network model to extract high-level features into a combination of low-level abstractions to find the distributed data features, in order to solve complex machine learning problems. The smart grid in general and the Cleantech are some of the early adopters of the machine learning technologies. In the following, some of the practical applications of the deep learning and machine learnings are reviewed. Fault detection in solar photovoltaic (PV) arrays is a crucial task for increasing the reliability and safety of PV systems. Because of the PV systems’ nonlinear characteristics, a variety of faults may be difficult to detect by conventional protection devices. A faulty PV system could potentially lead to safety issues and in some cases can cause fires. Machine learning techniques can potentially solve this problem by detecting faults based on an analysis of voltage, current, irradiance, and temperature. A graphbased, semi-supervised learning model using only a few labelled training data that are normalized for better visualization can be deployed to detect a fault in the photovoltaic arrays [15]. The other application of machine learning in smart grids relates to the energy generated by the renewable plants. Uncertainty quantification is a key element input to maintain acceptable levels of reliability and profitability in power system operations. The Extreme Learning Machine (ELM) is a fast regression model, trained in varied ways to obtain both point and quantile forecasts of solar power generation [16]. The increasing penetration of solar generation systems in the main electricity grid demands a more accurate forecast of electricity generation. The sun and its intensity is the main element in the generation of electricity via PV systems. The smart grid needs to be able to predict the electricity generated by PV and adapt the system accordingly. One of the new methods used to predict the PV generation is based on various satellite images and a support vector machine (SVM) learning scheme that utilizes satellite images of atmospheric motion vectors (AMVs) [17]. The other applications of the machine learning in the smart electricity environment are related to the prediction of the customers’ electricity usage pattern from the retailer point of view. Machine learning can help by developing a reliable model of the aggregate behavior of the price-sensitive customers to reduce the uncertainty associated with the estimation of electricity consumption. The outcomes of the electricity consumption can lead to optimum purchasing decisions for the electricity retailers and eventually lower electricity cost for the customers [18]. The other applications of machine learning are related to micro-grids. The historical data which is collected by energy management systems can be used to predict the pattern of generation and consumption in the micro-grid to optimize its overall performance. The machine learning tool can significantly improve the operation of the micro-grids and eventually optimize the return on investment [19].

IV. APPLICATIONS IN SUPPLY CHAIN MANAGEMEMNT

Organizations cannot act in isolation and they depend much on the capabilities and resources embedded in their suppliers, customers and collaborators. The focus on supply chain management (SCM) began in 1980 when organizations realized the benefits of collaborative relationships via SCM within organizational boundaries [20]. The management of supply chains in a sustainable manner has become an increasing concern for organisations of all sizes and across a wide range of industries. This more reactive approach of responding to external pressure from governments, consumers and nongovernmental organizations (NGOs) and media can be complemented by the development and introduction of sustainable products [21]. In the current competitive environment, supply chain professionals are struggling to handle the huge amount of structured and unstructured data. They are considering new techniques to investigate how data are produced, captured, organized and analysed in order to provide industries with valuable insights. Big Data analytics is one of the best techniques which can help them to overcome their problem. IoT in the context of SCM is defined as “a network of physical objects that are digitally connected to sense, monitor and interact within a company and between the company and its supply chain enabling agility, visibility, tracking and information sharing to facilitate timely planning, control and coordination of the supply chain processes” [22]. Our search of the Web of Science database revealed that 308 papers have been published on the topics of “internet of things” and “supply chain” between 1980 and 2018. Fig. 3 shows various countries’ contributions to the research on these topics; Fig. 4 shows the trend of publications from 2008 to Feb 2018. 

Fig.3. Publication of research papers in IoT and SC topics by countries from 1980 to Feb 2018

Fig. 4. Publications related to IoT and SC topics since 2008

Overall, the acceptance of sustainability by an organisation can boost its competitiveness as it can enhance the firm’s image in the eyes of the customers and can increase the firm’s economic performance [23]. In the stream of supply chain management, sustainability is mainly discussed in terms of green supply chain management (GSCM) that is not a new concept since it emerged in 1989 [24]. However, literature on GSCM criteria is rare since the GSCM philosophy parallels or overlaps other eminent management agendas such as cleaner production.

The bibliometric networks’ study on GSCM and IoT topics shown in Fig. 5 reveals that 634 papers were categorised under GSCM and IoT topics and were published between 2000 and Feb 2018.

Environmental protection and sustainable development, green product, and green supplier selection are mainly interrelated, whereas GSCM implementation, green purchasing and economic performance are in the green practices and drivers category.

The challenges and issues of IoT in SCM have been discussed broadly in the literature [25-27]. Research has been conducted on each component of the chain of supply from supplier control and purchasing to customer relationship management and recycling. Technologies such as RFID tags and wireless sensor networks and cloud computing have been widely utilized to improve the efficiency of SCM.

The challenges of warehouse operation in the era of industry 4.0 have been discussed by [28]. The transparency and traceability of inventory accuracy by an ambient intelligence system that detects warehousing activities and operations are one of the main challenges in warehouse management. [28] proposes the IoTbased warehouse management system which utilizes RID tags to control and identify the raw materials and semi-finished components. In another research conducted by [29], a Q- learning algorithm as a reinforcement learning model was proposed for overcoming a major SCM challenge – the reduction of inventory costs through the coordination of ordering activities.

Fig.5. Publication of research papers in IoT and SC topics by countries from 1980 to Feb 2018 

V. CONCLUSIONS

This paper reviews the applications of IoE technologies to sense, transfer, and store data, and machine learning as a technique to manipulate data to give meaning to information and produce useful insights. These applications are discussed in three diverse fields: health, smart electrical grid and SCM. This research shows that in each business domain, different machine learning techniques are employed for specific problems. For example, in the health sector, the pattern recognition techniques are applied for health monitoring, or as deep learning techniques for health management. In regard to the smart grid, different methods have been employed for different sectors. For example, the SVM method can be used for forecasting power generation by solar energy and neural networks can be employed for consumption pattern recognition. Our study demonstrates that IoT technology can be employed for practices and processes of GSCM. Big data and cloud computing which are strongly associated with the IoE were not discussed as they are beyond the scope of the paper.

REFERENCES

[1] K. P. Murphy, Machine learning: a probabilistic perspective. MIT press, 2012.

[2] L. T. Yang, B. Di Martino, and Q. Zhang, "Internet of Everything," Mobile Information Systems, vol. 2017, 2017.

[3] I. Lee and K. Lee, "The Internet of Things (IoT): Applications, investments, and challenges for enterprises," Business Horizons, vol. 58, no. 4, pp. 431- 440, 2015.

[4] L. M. R. Tarouco et al., "Internet of Things in healthcare: Interoperatibility and security issues," in Communications (ICC), 2012 IEEE International Conference on, 2012, pp. 6121-6125: IEEE.

[5] M. Rohokale, N. R. Prasad, and R. Prasad, "A cooperative Internet of Things (IoT) for rural healthcare monitoring and control," in Wireless Communication, Vehicular Technology, Information Theory and Aerospace & Electronic Systems Technology (Wireless VITAE), 2011 2nd International Conference on, 2011, pp. 1-6: IEEE.

[6] Y. J. Fan, Y. H. Yin, L. Da Xu, Y. Zeng, and F. Wu, "IoT-based smart rehabilitation system," IEEE transactions on industrial informatics, vol. 10, no. 2, pp. 1568-1577, 2014.

[7] Y. H. Yin, Y. J. Fan, and L. D. Xu, "EMG and EPP-integrated human– machine interface between the paralyzed and rehabilitation exoskeleton," IEEE Transactions on Information Technology in Biomedicine, vol. 16, no. 4, pp. 542-549, 2012.

[8] Y. Yuehong, Y. Zeng, X. Chen, and Y. Fan, "The internet of things in

healthcare: An overview," Journal of Industrial Information Integration,

vol. 1, pp. 3-13, 2016.

[9] R. Peng , M.L. Sichitiu , Angle of arrival localization for wireless sensor networks, in: Proceedings of the 3rd Annual IEEE Communications Society on Sensor and Ad Hoc Communications and Networks, 2006. SECON’06. 2006, Vol. 1, IEEE, 2006, September, pp. 374–382.

[10] J. Bussmann , J. Tulen , E. van Herel , H. Stam , Quantification of physical activities by means of ambulatory accelerometry: a validation study, Psychophysiology 35 (5) (1998) 4 88–4 96 .

[11] T. Erl , Service-oriented architecture: concepts, technology, and design, Pearson Education India, 2005 .

[12] X. Zheng , P. Martin , K. Brohman , L. CLOUDQUAL Da Xu , A quality

model for cloud services, IEEE Trans. Ind. Inf. 10 (2) (2014) 1527–1536 .

[13] N. J. Van Eck and L. Waltman, "Software survey: VOSviewer, a computer

program for bibliometric mapping," Scientometrics, vol. 84, no. 2, pp. 523-

538, 2010.

[15] Y. Zhao, R. Ball, J. Mosesian, J.-F. de Palma, and B. Lehman, "Graphbased

semi-supervised learning for fault detection and classification in

solar photovoltaic arrays," IEEE Transactions on Power Electronics, vol.

30, no. 5, pp. 2848-2858, 2015.

[16] F. Golestaneh, P. Pinson, and H. B. Gooi, "Very short-term nonparametric

probabilistic forecasting of renewable energy generation—With application

to solar energy," IEEE Transactions on Power Systems, vol. 31, no. 5, pp.

3850-3863, 2016.

[17] H. S. Jang, K. Y. Bae, H.-S. Park, and D. K. Sung, "Solar power prediction

based on satellite images and support vector machine," IEEE Transactions

on Sustainable Energy, vol. 7, no. 3, pp. 1255-1263, 2016.

[18] K. Dehghanpour, H. Nehrir, J. Sheppard, and N. Kelly, "Agent-based

modeling of retail electrical energy markets with demand response," IEEE

Transactions on Smart Grid, 2016.

[19] L. Oneto, F. Laureri, M. Robba, F. Delfino, and D. Anguita, "Data-driven

photovoltaic power production nowcasting and forecasting for

polygeneration microgrids," IEEE Systems Journal, 2017.

[20] Giannakis, M. and S. R. Croom (2004). "Toward the Development of a Supply Chain Management Paradigm: A Conceptual Framework." Journal of Supply Chain Management 40(1): 27-37

[21] S. Seuring and S. Gold, "Sustainability management beyond corporate boundaries: from stakeholders to performance," Journal of Cleaner Production, vol. 56, pp. 1-6, 2013.

[22] M. Ben-Daya, E. Hassini, and Z. Bahroun, "Internet of things and supply chain management: a literature review," International Journal of Production Research, pp. 1-24, 2017.

[23] J. D. Linton, R. Klassen, and V. Jayaraman, "Sustainable supply chains: An introduction," Journal of operations management, vol. 25, no. 6, pp. 1075- 1082, 2007.

[24] M. Chien and L.-H. Shih, "An empirical study of the implementation of green supply chain management practices in the electrical and electronic industry and their relation to organizational performances," International Journal of Environmental Science and Technology:(IJEST), vol. 4, no. 3, p. 383, 2007.

[25] S. Tiwari, H. Wee, and Y. Daryanto, "Big data analytics in supply chain management between 2010 and 2016: Insights to industries," Computers & Industrial Engineering, vol. 115, pp. 319-330, 2018.

[26] M. Ben-Daya, E. Hassini, and Z. Bahroun, "Internet of things and supply chain management: a literature review," International Journal of Production Research, pp. 1-24, 2017.

[27] T. Nguyen, Z. Li, V. Spiegler, P. Ieromonachou, and Y. Lin, "Big data analytics in supply chain management: A state-of-the-art literature review," Computers & Operations Research, 2017.

[28] C. Lee, Y. Lv, K. Ng, W. Ho, and K. Choy, "Design and application of Internet of things-based warehouse management system for smart logistics," International Journal of Production Research, pp. 1-16, 2017.

[29] S. K. Chaharsooghi, J. Heydari, and S. H. Zegordi, "A reinforcement

learning model for supply chain ordering management: An application to

the beer game," Decision Support Systems, vol. 45, no. 4, pp. 949-959, 2008.708