ELSEVIER Computer Networks 148 (2019) 241-261 Contents lists available at ScienceDirect Computer Networks journal homepage: www.elsevier.com/locate/comnet Computer Networks Internet of Things applications: A systematic review Parvaneh Asgharia, Amir Masoud Rahmania,*, Hamid Haj Seyyed Javadi b Department of Computer Engineering, Science and Research Branch, Islamic Azad University, Tehran, Iran b Department of Mathematics and Computer Science, Shahed University, Tehran, Iran Check for updates ARTICLE INFO Article history: Received 7 July 2018 Revised 23 November 2018 Accepted 7 December 2018 Available online 10 December 2018 Keywords: Application-based services Internet of things Systematic literature review Smart objects Quality of service ABSTRACT Internet of Things (IoT) is considered as an ecosystem that contains smart objects equipped with sensors, networking and processing technologies integrating and working together to provide an environment in which smart services are taken to the end users. The IoT is leading numerous benefits into the human life through the environment wherein smart services are provided to utilize every activity anywhere and anytime. All these facilities and services are conveyed through the diverse applications which are per- formed in the IoT environment. The most important utilities that are achieved by the IoT applications are monitoring and consequently immediate decision making for efficient management. In this paper, we intend to survey in divers IoT application domains to comprehend the different approaches in IoT ap- plications which have been recently presented based on the Systematic Literature Review (SLR) method. The aim of this paper is to categorize analytically and statistically, and analyze the current research tech- niques on IoT applications approaches published from 2011 to 2018. A technical taxonomy is presented for the IoT applications approaches according to the content of current studies that are selected with SLR process in this study including health care, environmental monitoring, smart city, commercial, indus- trial and general aspects in IoT applications. IoT applications are compared with each other according to some technical features such as Quality of Service (QoS), proposed case study and evaluation environ- ments. The achievements and disadvantages of each study is discussed as well as presenting some hints for addressing their weaknesses and highlighting the future research challenges and open issues in IoT applications. © 2018 Elsevier B.V. All rights reserved. 1. Introduction In recent years, the Internet of Things (IoT) has penetrated per- vasively into the most aspects of human life everywhere such as cities, homes, universities, industrial factories, organizations, agri- culture environments, hospitals and health-care centers [1–4]. Nu- merous capabilities such as produce/consume data and online ser- vices, improve daily life and activities around the world through the IoT context [5]. The facilities and smart services are carried through the various applications which are performed in the IoT environment [6]. As users' desires grow, innovative applications are being provided for monitoring, managing and automating human activities [7,8]. Also, IoT applications apply cloud service comput- ing for achieving proper composite services via composition of ex- isting atomic services for service-based applications in the IoT con- text [9,10]. IoT scenarios are applied to applications with smart de- vices which users apply them in their daily activities in various fields. Also, IoT applications have some benefits for users to choose the best opportunity in any case, decision making, managing, and monitoring environmental cloud resources [11]. Despite the different application domains' motivations, all of them have a shared and common goal: provisioning smart services to increase the quality of human life [12,13]. The main concern of IoT applications is satisfying Quality of Service (QoS) metrics. User's requirements should be supported by smart services in IoT applications that cover the QoS metrics such as security, cost, ser- vice time, energy consumption, reliability and availability. There are some technical surveys and review papers that do not concentrate on the IoT applications systematically [14,15]. The main aim of this research is to survey in different IoT applications to comprehend the diversity of approaches in IoT applications which have been recently presented. The key approaches of IoT applica- tions that have been focused in selected studies consist of health- care, environmental monitoring, smart city, commercial, industrial and general approaches. We present a Systematic Literature Review (SLR) method and overview opportunities of the IoT applications. The main commitments of this study are highlighted as follows: * Corresponding author. E-mail address: rahmani@srbiau.ac.ir (A.M. Rahmani). https://doi.org/10.1016/j.comnet.2018.12.008 1389-1286/© 2018 Elsevier B.V. All rights reserved. 242 P. Asghari, A.M. Rahmani and H.H.S. Javadi/Computer Networks 148 (2019) 241-261 Table 1 Related studies in IoT applications. Reference Main topic Publication year Covered years Li et al. [15] Han et al. [16] Ray [17] IoT technology 2015 IoT smart objects 2015 IoT architectures 2016 Bello and Zeadally [12] Talavera et al. [5] IoT services 2017 IoT in environmental and agro-industrial fields 2017 2004-2017 2006-2016 • • . Designing a technical taxonomy to classify the various IoT ap- plications. Presenting a discussion of the key challenges for the applica- tions approaches in IoT environment. Highlighting the future research challenges and open issues in IoT applications. The organization of this review is considered as follows: The related works are presented in Section 2. Section 3 provides a research selection technique and motivation regarding the SLR process. Section 4, outlines the applications approaches in IoT sys- tematically and categorizes them. Also, in this section a technical taxonomy and comparison of the approaches for selected papers is presented. In Section 5, a discussion is provided on the IoT appli- cations that have not been analyzed comprehensively up to now. Finally, Section 6 demonstrates the conclusion along with the pa- per restrictions. 2. Related work This section presents a momentary explanation of the related work studies in IoT applications. Bello and Zeadally [12] discussed the existing networking stan- dards in the IoT environment to clarify how they can afford to satisfy the QoS necessities of objects to make the smarter IoT ecosystem. Furthermore, an analysis on diverse applications and the risk of lack of cross-domain integration in the IoT environment was presented to realize to fulfill the interoperability and QoS re- quirements such as availability, reliability, scalability and security for providing the IoT services. The strong point of this study is to present a classification of various recent standards in the net- work layer and application layer in diverse areas including build- ing, transportation, smart city, business and grid systems. The main deficiencies of this study are the absence of presenting any sta- tistical information about discussed standards applied in different stated domains and similarly not giving any statistic chart for the risk analysis of the lack of interoperability between IoT objects and transport protocols to illustrate briefly the judgments. Talavera et al. [5] surveyed on IoT applications in environmen- tal and industrial agriculture. In this review paper four area in- cluding prediction, monitoring, control and logistics, are consid- ered. Two important subjects are mentioned and answered in this study. The first one is about the fundamental technological efforts in IoT-based applications for agro industrial and environmental is- sues, and the second one is about the infrastructures and tech- nologies used in the mentioned solutions. It is derived that the maximum of the papers were concentrated on monitoring (62%), and then control (25%), logistics (7%), and prediction (6%). Fur- thermore, according to the second question mentioned in this survey, it is realized that the most technologies and Infrastruc- tures applied in the IoT agro industrial and environmental applica- tions, categorized in seven sets including visualization approaches, storage approaches, edge computing technologies, communication techniques, power sources, actuators and sensing variables. The mentioned open issues in this review contain these topics: robust standardization, improved power consumption, security, reusabil- ity of software and hardware components, decrease cost, appro- priate compatibility with existing infrastructures and scalability issues. The authors presented an architecture for the IoT agro in- dustrial and environmental applications. The offered model con- sists of four layer including application, service, communication and physical layer. The advantage of this study is presenting bene- ficial and comprehensive statistics about the studies and efforts in the agro industrial and environmental applications in the IoT con- text. The weakness of this paper is that the explanation of the re- lated works is not sufficient. Han et al. [16] presented a review on problem of service com- position of Internet Protocol (IP) smart IoT objects. The authors provided a full investigation based on some issues such as smart IoT objects systems, service modeling, target applications, target platforms and service composition approaches for the IPs in IoT The main weakness of this survey is that the evaluation factors in- cluding availability, response time, cost, and scalability as impor- tant quality factors were not analyzed. Li et al. [15] provided a survey in IoT major techniques. This re- view discussed architecture layers including perception layer, net- work layer, service layer and the interface layer. The advantage of this review is presenting comprehensive open issues and chal- lenges in the IoT, however, the compatibility of each approach in the IoT applications was not considered. A survey on IoT was presented by Ray [17]. This study discusses about the topics such as Service Oriented Architecture (SOA), WSN, health-care systems and social computing. The main defect of this study is not providing any analysis on evaluation parameters such as availability, energy consumption, cost, response time and relia- bility as quality factors in this area. Table 1 represents a summary of the related review studies on the IoT applications issues with respect to the systematic literature review and survey studies. The core subject, publication year and covered years are listed for each study in this table. According to the existing review papers, the existing deficien- cies propose that we provide a comprehensive literature review to address these weaknesses as follows: • • The present studies do not provide any analytical assessment and taxonomy for application approaches in IoT. • Some studies do not evaluate the important assessment factors on applications in IoT. The structure of the presented studies does not have the sys- tematic arrangement and the paper selection method is not clear. 3. Research selection method This section provides a review based on SLR method as a research study assessment for classifying the IoT applications [18-20]. Considering the alternatives and other synonyms of the key es- sential components, the subsequent exploration string was defined [21-26]: • • ("Application" OR "Software" OR "Application-based" OR "Appli- cation layer" OR "App") AND ("IoT") OR ("Internet of Things") 18 17 16 15 14 13 12 11 10 9 8 7 6543210 P. Asghari, A.M. Rahmani and H.H.S. Javadi/Computer Networks 148 (2019) 241-261 IN 2011 2012 2013 2014 2015 2016 2017 2018 Elsevier III Springer IEEE ACM Other publishers Search in electronic databases IEEE, ACM, ScienceDirect, Research Springer, T&F, Wiley, identification Google scholar Inderscience, Sage Fig. 1. Distribution of research papers by publisher. Applying exclusion Survey and review papers. Not indexed in ISI. Non English context. Observing 185 published papers in IoT applications Research refinement Not peer-reviewed procedure. Applying inclusion Published between 2011 - 2018. Field of IoT application. Existing quality aspects of publication. Fig. 2. The selection criteria and evaluation chart of research studies. 243 Final selected studies 72 research studies in IoT application This SLR paper presents inclusive answers to the following An- alytical Questions (AQ) regarding the aims of this research [27]: • AQ1: Which domains are categorized in IoT applications? • • • AQ2: Which main contexts are considered for IoT applications? AQ3: What measurement environments are used for evaluating the IoT applications? AQ4: What are the evaluation factors usually applied in IoT ap- plications? •AQ5: What are the future researches directions and open per- spectives of IoT applications? Fig. 1 shows the distribution of the research studies completed by the leading scientific publishers regarding the article citations and review method, including Elsevier, IEEE, Springer, ACM, Wi- ley and Taylor & Francis. In this organization, scientific electronic databases including IEEE Xplorer and Science Direct are applied in the SLR method. After providing the analytical questions, the inclusion/exclusion criteria for the ultimate research selection was applied. Regarding the number of published papers, we just analyze the journal ar- ticles and conference papers indexed in WoS and ISI proceedings as the peer-reviewed papers for the applications approaches in IoT. Finally, 72 peer-reviewed papers were selected for analyzing and answering the mentioned analytical questions which are presented in detail in Section 4. Fig. 2 shows the selection principles and evaluation flowchart designed for the studies. The exclusion phase consist of ignoring non-peer-reviewed researches, short papers, white papers, book chapters, and low-quality studies (published in predatory journals) that did not present any scientific discussion and technical infor- mation. For mapping the final selected studies, the inclusion prin- ciples are considered as follows: • The studies available online between 2011 and 2018. The studies in IoT applications topics. The studies with a technical quality method in IoT applications. Studies using existing quality aspects. For mapping the final selected studies, the exclusion principles are considered as follows: • The studies presenting review and survey papers. • The studies not indexed in WoS and ISI. ⚫ The studies not written in English language. ⚫ The studies not peer-reviewed process. 244 P. Asghari, A.M. Rahmani and H.H.S. Javadi/Computer Networks 148 (2019) 241-261 Health-care Environmental IoT applications Smart city Commercial Industrial General aspects Smart Smart Wearables animal farming Smart home Smart grid Shopping systems Personal monitoring Smart agriculture Smart building Scheduling systems Wild vegetation monitoring Domestic waste treatment monitoring Urban computing Security and emergencies Traffic monitoring Regional climate change monitoring Mobile crowd sensing Retail 4. Organization of the IoT applications Vehicle monitoring Fig. 3. The taxonomy of IoT applications. This section presents a technical review of the selected IoT ap- plications for the existing studies according to the applied SLR pro- cess. Fig. 3 shows a comprehensive taxonomy on the IoT appli- cations that include health-care, environmental, smart city, com- mercial, industrial and general aspects [28,29]. Since, in each type of IoT applications, some problems can potentially emerge that should be focused to find effective solutions to make the IoT appli- cations more efficient and applicable in the real IoT environments, we review the papers that try to address some issues to support the IoT applications in a particular domain. For example, in smart city applications, the key subjects such as semantic-aware mo- bile crowd-sensing, vehicular monitoring, location finding, context- aware or QoS-aware service composition, scalable IoT platforms, managing scalable heterogeneous data streams and many other is- sues arise in various aspects of smart city IoT applications. There- fore, the presented taxonomy in this paper is based on different types of IoT applications in which particular subjunctives were dis- cussed and addressed in the selected research papers. Regarding the challenges and concerns in different categories of IoT applica- tions, firstly, we concentrate on type of the IoT applications, and then we try to study the main context focused in the selected pa- pers. Since, some concerns are general in the IoT applications, we introduce a category in our taxonomy named “general aspects", to classify the papers which presented an approach to cope with a specific challenge for supporting any type of IoT application. Of course, the general aspects in the proposed taxonomy are applied in all IoT application domains such as applied and systematical software, evaluation procedure or performance prediction of IoT applications [30]. In other words, the illustrated studies of the gen- eral aspects presented a novel conceptual approach for using in de- velopment any type of IoT application. The following subsections illustrate the different approaches in IoT applications. Also, the different studies will be compared in several sides such as the specified main context, case studies, ad- vantages, weaknesses and finally the particular outputs. 4.1. Health-care applications Kim and Kim [31] provided a guide for IoT health-care service providers based on the users' view point. In this study some im- portant offered attributes, have further impacts on users' confirma- tion of such services. An analysis was performed to evaluate the suggested factors such as trust and risk sensitivity to qualify the services. In this study by focusing on lifestyle disease, the results exposes that in South Korea people prefer trustworthy and safe health-care services. The advantage of this work is to give a simple and new guide for IoT health-care service producers and improve the reliability. The weakness of this study is that the extracted re- sults are based on the hypothetical service description instead of commercial service applied by the health-care user. Also the other deficiency is that the response time is not evaluated. Fafoutis et al. [32] proposed a platform to develop a monitor- ing system to detect and prevent chronic medical conditions in- cluding diabetes, obesity or depression in today life in a residen- tial environment. In this work, as a result of the related costs of recharging or replacing the batteries of the wearable devices, the issue of energy limitations is considered. In this paper, only the solutions which focus on battery-powered are taken into account. By deploying the asymmetry of the network resources, the energy efficiency of the wearable hardware is correspondingly improved. The proposed system is relied on Bluetooth low power and also is a portion of a residential platform which is equipped with video cameras on body and ecological sensors for distinguishing to real- ize immediate perceptive and making decision by means of ma- chine learning techniques. To evaluate the proposed framework, a prototyped wearable hardware and three prototyped receiver units were applied. The advantages of this paper are improving the energy feeding of the wearable devices and also enhancing the P. Asghari, A.M. Rahmani and H.H.S. Javadi/Computer Networks 148 (2019) 241-261 reliability of the proposed system. Furthermore, the precision of the RSSI and the transmission power is evaluated. The weakness of this work is that the response time is not assessed. Jimenez and Torres [33] developed a prototype as an IoT health- care monitoring system which could be extended in duration of use. In this work, the system was developed by means of low cost sensors and existing IoT platform which exist commonly in homes. The presented ad-hoc monitoring system checks the patient's bi- ological factors and the environmental data such as humidity and temperature that sensed by either present sensors or the new sen- sors, are added during the system execution time. In this paper, smart phones are considered as gateways for collecting the men- tioned information from sensors at patient's environment. The of- fered health-care monitoring system is used for collecting business necessities in actual circumstances to evaluate the performance and scalability of the presented prototype. The advantage of this work is using low cost sensors and low energy consumption de- vices that exist typically in homes. The weakness of this system is to satisfy just minimum service level according the collected busi- ness requirements. Ding et al. [34] proposed an IoT system for remote mobile medical monitoring. The authors suggested a model of the hu- man interaction and physiological parameters by using intelligent nodes. The offered system is implemented for developing the ef- ficient emergency alarm systems by saving the critical emergency information in hospital database. The advantages of this work are improving the rapidity and precision of the physiological factors measurement and also using low power consumption devices. The weakness of this work is that the cost was not considered. Savola et al. [35] offered a high-level adaptive mechanism for security management with respect to security metrics. For pre- senting this idea the security intentions of E-health IoT applica- tions particularly for health of aged people and the treatment of lasting diseases, are analyzed. Furthermore, in this paper, the re- quirements of adaptive security management and decision-making, which are necessary to set the security needs, and also applying the sufficient security controls in condition of changing security threats in such applications, are discussed. The presented mech- anism for adaptive security management considers several secu- rity factors such as security accuracy, effectiveness, efficiency, pri- vacy level and confidentiality. The main advantage of this study is improving the security intentions of E-health IoT applications. The weakness of this work is the absence of any detailed analysis of security metrics and adaptive decision-making algorithm in such E-health IoT applications. Baloch et al. [36] proposed a layered context-aware data com- bination approach for IoT health-care applications. The proposed approach consists of context attainment, condition structure and implication. In this paper, it is explained that Body Sensor Net- works (BSN) or Wireless Body Area Networks (WBAN) that are or- dinarily positioned to the patient's body to accumulate physiolog- ical data for IoT health-care applications. As the data are collected from multiple heterogeneous sources, a technique for combining these datasets is necessary which is known as "data fusion”. In this paper, a novel approach presented for demonstrating the col- lected data in a way that assistances taking on time precise de- cision. The advantage of this study is addressing some challenges such as sensor deficiency, limited coverage, irregularity and ambi- guity. The weakness of this paper is that not presenting any spe- cific algorithm to evaluate the suggested approach. Subrahmanyam et al. [37] proposed a least error adjusted IEEE 802.15.4 transceiver for health-care applications in IoT environ- ment. In this paper, an improved frequency offset evaluator sug- gested with better performance of error modification than the cur- rent assessors for IEEE 802.15.4. The advantage of this study is a considerable improving in bit error amount and packet error 245 frequency of the offered transceiver in contrast to the standard ar- chitecture. Also, another achievement of this paper is decreasing power consumption as a result of fewer number of retransmission in each packet for successful broadcast of the packet. Lin et al. [38] developed a novel paradigm for sensor tasking such as choosing to detect sensors in Cognitive Radio Sensor Net- works (CRSNs). Also, in this study, a conceptual attitude of the suggested model in contradiction of the Primary User Emulation Attack (PUEA) is proposed. The advantage of offered approach is improving the effectivity of attack detection. The main weakness of this paper is that not considering the reliability and operabil- ity of the scheme according to the influence of node movement on detection performance. Damis et al. [39] presented an analysis for three epidermal loop antennas in theoretical and experimental aspects for measuring bi- ological factors in order to apply in health-care IoT applications. In this study, Error-Vector Magnitude (EVM) examinations and BER factors are showed and evaluated, correspondingly, to validate the truth of GSM and BLE communications for the QL antennas. Experi- mental results show proper dependability between radiation forms and refection factor. Data communication assessments represented that the attained BER could be suitable for this antenna technology to operate in a 4- QAM wireless connection. Elappila et al. [40] presented a routing algorithm for WSN based on congestion and interference awareness of energy efficiency. The proposed technique was developed to operate in the networks with heavy traffic and interference on the connection among the nodes for the reason that several IoT objects deliver their data to a same target, which is a usual situation in the IoT monitoring appli- cations. The proposed algorithm applies a function for choosing the next party node which works with three parameters: (1) signal to interference and noise ratio of the connection (SNIR), (2) the sur- vivability parameter of the route from the next party node to the endpoint, and (3) the congestion degree at the next party node. The acquired results from simulation show improvement in net- work throughput, packet broadcasting ratio, energy consumption of the nodes and also decreasing the amount of the lost packets. Jebadurai and Dinesh Peter [41] proposed an architecture for IoT healthcare applications in order to process retinal images which captured by smartphone funduscopy. An algorithm based on super-resolution (SR) using kernel Support Vector Regression (SVR) method, was offered to increase the quality of the images. The out- comes display that the suggested algorithm executes more efficient in comparison to other present super resolution algorithms. Malik et al. [42] presented an analysis on performance of nar- row band IoT (NB-IoT) which provides heavy communications with low data rate for long lifetime batteries and simple sensors, in IoT health-care applications. The main aim of this paper is to analyze the latency and throughput of NBIoT in health-care monitoring ap- plications. 4.1.1. Analysis of the reviewed health-care applications Table 2 shows the classification of the above papers and the important aspects to evaluate the health-care approach in IoT ap- plications. The most important main contexts in the health-care approach are based on the user preferences for lifestyle disease management, residential maintenance-free long-term activity mon- itoring, home monitoring by remote mobile medical and metrics- driven adaptive security management. Table 3 presents a side by side evaluation for the above stud- ies applying the evaluation elements in health-care IoT applica- tions. The parameters consist of availability, response time, en- ergy consumption, cost, reliability, security and throughput. In the health-care approach, most research studies assessed their of- fered approach in the response time, cost and energy consumption properties./n1-s2.0-S1389128618305127-main.pdf Reading assignment: 1. What is IoT? Write down five applications of IoT. 2. How will the Internet of Things (IoT) evolve? This assignment will count as "Assignment 0." The due date is 29th January 2024.

Fig: 1