Challenges in Integration of Heterogeneous Internet of Things

Internet of Things (


Motivation
IoT is the expansion of current Internet services to provide connectivity to each object of this world. IoT has become the most prominent technology across the globe. It is an emerging technology that is under development process where everyone is trying to interpret it according to their needs. e implementation and interpretation of IoT face some serious challenges like security, virtualization, and heterogeneity. Heterogeneity itself is a multifaceted challenge hindering the large-scale implementation of IoT vision. It is due to these challenges that so far only limited implementations of IoT systems have become a reality. is motivated us to perform a systematic literature review to identify those IoT heterogeneity challenges and their solutions. Another contribution of this study is conducting a depth analysis of those challenges using the chi-square test based on digital libraries and timeframe.

Introduction
In today's technological world, IoT is considered an important advancement among the trending technologies. e term IoT can be simply de ned as the devices that can be connected with sources of the Internet [1]. In past years, these devices have been constantly growing. Reference [2] reported that around 500 billion devices will be connected to the Internet by 2030. In the physical as well as in the virtual world, these IoT devices will further interconnect with other devices in a large number, which will give new birth to the forms of interaction. is will enable us to connect all objects of our surroundings in every corner of the world in a single period. ese objects can be sensors, smartphones, automobiles, industrial robots, refrigerators, thermostats, tablets, etc. e IoT is widespread in both academia and industry. It is producing business opportunities in multiple fields of industrial markets, in both public and private sectors in a very broad range. e industrial revolution of IoT will have billions of heterogeneous devices on the Internet of ings in the near future.
At the other extreme, the IoT vision of a large-scale implementation faces serious challenges across many dimensions. One of the main obstructions in IoT is its inclination toward heterogeneity, and the heterogeneous nature might be in form of protocols, device data format, communication capabilities of the devices, technologies, hardware, etc. [3,4]. It is due to these types of challenges that so far only limited implementations of IoT systems have become a reality. For IoT to evolve toward its vision of global implementation, these obstructions need to be reduced on different levels. To activate and provide the service, devices must be connected to the Internet. e identification of heterogeneity-based challenges that exist at different levels is needed, and the current solutions adopted and/or implemented by different studies for handling heterogeneity in IoT systems need to be highlighted. e objective of this study is to conduct a systematic literature review to identify those IoT heterogeneity challenges and find out the solutions implemented by different studies to handle IoT heterogeneity challenges. e significance of this study is that it will provide the identification and analysis of heterogeneity challenges in IoT systems and provide a summary of different studies that implemented various solutions to handle the heterogeneity of IoT systems. Another significant contribution is that it will provide a future direction to researchers to make a better stand-alone architecture aiming to tackle the heterogeneity challenges at different levels of IoT systems. As a result, IoT systems can be utilized and implemented in a wide range of industrial fields. e organization of this paper is as follows: Section 3 presents a literature review related to IoT history, challenges, and heterogeneity concerns. Section 4 explains the research methodology used to achieve the objectives of this research. Section 5 provides the result and discussion of IoT heterogeneity challenges and the solutions to those challenges found in this study. Section 6 concludes this review and suggests future work.

Literature Review
Because of the broad and complex nature of IoT, it has not yet got a single unique definition that is acceptable to the whole global community of users. Many researchers, practitioners, academicians, developers, and corporate people have defined IoT in their terms, but the credit must be attributed to Kevin Ashton, an expert on digital innovation who for the first time used and defined it. According to [5], IoT could be nicely defined as follows: a very comprehensive and accessible network of intelligent devices which can act and react in accord with situations; self-organize; share information, data, and resources; and be subject to change in the environment.
IoT is growing and maturing day by day. It is the latest, most fine, and excellent concept in information technology. It is a new paradigmatic shift in information technological advancement. e expression "Internet of ings," concisely shortened to IoT, is comprised of two words, "Internet" and " ings." Internet uses a standard set of Internet protocols (TCP/IP) to connect and serve a large number of users around the globe [6,7]. It is a global system of interconnected computer networks. Internet is interconnected networks that includes a large number of local to regional commercial networks that may be private, governmentowned, public, or academic networks. ese networks are connected through a wide range of electronic, wireless, and optical network technologies [8]. e Internet is generally defined as a global network that connects millions of computers. About 190 states of the world are linked through the Internet that constantly shares data, opinions, and news [9]. According to [10], there is an estimated 5,080,388,142 Internet users around the world. is large sum of users indicates that about 40% of the world's total population uses the Internet. e word " ings" in the "Internet of ings" can be any object or person identifiable in the real world. Daily necessities include electronic devices that we come across as well as daily used advanced technological items like equipment and gadgets. In the near future, other certain everyday objects are expected to connect with the Internet, which will lead to a period of extreme expansion of the Internet known as the Internet of ings. IoTsystem is based on devices to sense, actuate, control, and monitor activities [11]. IoT devices that are connected to other devices and applications can exchange data with each other, they may receive information from some other IoT devices. To process data, they may send the data either to centralized serves locally or it may send data for processing to cloud. e National Intelligence Council (NIC) of the United Nations (UN) has considered IoT as one of the six "Disruptive Civil Technologies" [12]. In this respect, we can list many fields that are already benefiting from the services of different architectural forms of IoT like transportation, e-governance, smart city, smart health, life support, education, retail, logistics, agriculture, automation, manufacturing of industrial products, and management of businesses.
Ericsson [13] and Evans [14] have conducted surveys and estimated that the use of the Internet will further increase, grow, and be boosted tenfold in the coming days. According to their estimation, about 50 billion devices would have connected by 2020. is expected number of new Internet devices shall be supposedly called constricted devices [15].
ese devices are small in size, are enclosed in nature, and have a low cost. ey are specifically designed for the purpose of executing specific tasks like monitoring the physical environment. IoT devices are limited in terms of 2 Scientific Programming communication capabilities, processing power, and energy consumption due to their low cost. at is why these constrained devices are very heterogeneous in terms of their essential communication protocols, device data formats, and technologies.
Due to the heterogeneity of IoT, devices on the market nowadays have diversity in communication protocols, methods of network connectivity, and resulting models of application. It is not feasible to support such kinds of diversities in IoT, because developers typically lack the proper resources required to have a grip on the specifics of the constrained devices and network [16]. e main objective is to decrease, hide, or eradicate such a broad range of diversity of the technologies, applications models, and protocols from the users of IoT [17]. Because of the heterogeneous nature of IoT, it is one of the newly emerging research areas which has robust potential to bring a paradigm shift in the understanding of fundamental computer science principles and standards of our future living [18]. e demand for constrained IoT devices is expected to increase; this problem is expected to get worse in the future. erefore, there will be a need to improve the integration of a large number of constrained devices in IoT. In this study, we have performed SLR aiming to identify the heterogeneity challenges and provide a summary of the solutions adopted for those challenges.

Research Methodology
A systematic literature review (SLR) methodology is adopted to identify the heterogeneity challenges in IoT systems, hindering a global IoT vision, and to find solutions to the identified heterogeneity challenges.

Research Questions.
e first step of a systematic literature review is to define research questions. e research questions of this study are mentioned in Table 1.

Search String.
e second phase of a systematic literature review is to find relevant studies on the research topic. We identified digital libraries in which primary search was carried out: IEEE Xplore, SpringerLink, Google Scholar, ScienceDirect, and ACM. We then defined a set of keywords related to our research topic: "Internet of ings," "IoT," "heterogeneity," "heterogeneous," and "challenges." Finally, search strings were defined and used to collect published articles related to the research topic. Search strings are provided in Table 2.

Study Selection.
e research selection process is to perform search in digital libraries based on the tollgate approach considering the search strings. Figure 1 shows a selection of articles using the tollgate approach.
In snowballing process, we selected 9 papers from journals (IEEE TMC, TPDS, JSAC, ToN, TWC) and conferences (SIGCOMM, MobiCom, MobiSys, INFO-COM), the content of which is analyzed and discussed in Section.
Initially, 3854 papers were selected by applying search protocol to the selected digital libraries. A selection process has been applied based on keywords, titles, duplication removal, abstracts, and full text of selected papers. We excluded the papers of the following types: (1) Studies published in sources other than conferences, journals, patents, and technical reports (2) Research papers not published in the English language (3) Studies published before 2010 (4) Studies that are not related to the defined search strings To evaluate the quality of the included research papers, we assessed the following aspects: (1) e study provides information about any challenge related to IoT heterogeneity (2) e study represents a clear solution to the identified challenge of heterogeneity (3) e published study is from a stable and recognized publication source Figure 2 shows a summary of paper selection based on digital libraries, and Figure 3 shows the paper selection on a yearly basis.

Analysis and Discussion
In this section, we have discussed the results relevant to our RQs. For the analysis, we used the linear chi-square test of association. For categorical values of predictor and outcome variables, the chi-square test is counted as more significant than other statistical tests. To answer RQ1, identified challenges/critical issues through the SLR are presented in Table 3. We have found a significant difference in heterogeneity challenges.
We set an occurrence percentage threshold of 30%. Accordingly, "Heterogeneity of devices," "heterogeneity in formats of data," and "heterogeneity in communication" are the most critical identified challenges. Table 4 shows the analysis of the identified challenges based on digital libraries. We have Google Scholar, IEEE Xplore, SpringerLink, ScienceDirect, and ACM as digital libraries. From the analysis, we have found the following:

Comparison of Challenges Based on Digital Libraries.
(1) Heterogeneity in communication is critical in Google Scholar and SpringerLink (2) Heterogeneity of devices is critical in Google Scholar, IEEE Xplore, and SpringerLink Scientific Programming No.

Research questions RQ1
What are the challenges in heterogeneous IoT in the literature? RQ2 What are the solutions to these challenges in heterogeneous IoT in the literature? ("Heterogeneous Internet of ings") AND ("issues OR challenges") IoT heterogeneity IEEE Explore "Heterogeneous" AND ("IoT" OR "Internet of ings") AND "challenges" ScienceDirect ("IoT" OR "Internet of ings") AND ("heterogenous") AND ("challenges") ACM ("IoT") AND ("heterogenous") AND ("challenges") SpringerLink ("IoT") AND ("heterogeneity") AND ("challenges") (3) Heterogeneity in formats of data is critical in Sci-enceDirect, SpringerLink, and IEEE Xplore (4) Interoperability issue is critical in ACM and SpringerLink Table 5 shows the analysis of the identified challenges based on timeframe. We have divided the duration into two timeframes: Timeframe I from 2011 to 2016 and Timeframe II from 2017 to 2021. From the analysis, we have found the following:

Comparison of Challenges Based on Timeframe.
Heterogeneity in communication is critical in Timeframe I from 2011 to 2016 as shown in Table 5.
(1) Heterogeneity of devices is critical in Timeframe I and Timeframe II (2) Heterogeneity in formats of data is critical in Timeframe I and Timeframe II

Proposed Solutions.
To answer RQ2, solutions to the identified challenges are presented in Table 6. We have found a total of 81 solutions for those challenges, with at least 5 unique solutions for each challenge.
Fragmentation in connectivity, protocols     PEIoT, a data streaming system with enhanced privacy (i) Providing users with the ability to subscribe to/unsubscribe from data (ii) Enabling data controllers to invoke various privacy-enhancing technologies [70] 2019 Model Two modules: (i) Secure data processing system to maintain IoT data security and integrity (ii) Drone-based data analytic system using edge computing and ondevice computing [71] 2016 Architecture A novel architecture consisting of a distributed interface having enodes-inexpensive, simple, and embedded nodes Communication security [72] 2019 Protocol Multigroup key management protocol to ensure upstream and downstream secrecy, recovery from collision attacks, and network security [73] 2019 Mechanism A physical layer security mechanism to validate the single source for heterogeneous IoT [74] 2020 Protocol Using proxy re-signature, a privacy-preserving authentication protocol for heterogeneous system [75]

Conclusion and Future Work
is research is a systematic literature review that reviews the literature in the domain of IoT heterogeneity. e review has been implemented using a systematic methodology to select different studies addressing the challenges faced by heterogeneous IoT. To conduct this study, a total of 81 research papers that are published in different digital libraries from 2011 to 2021 were selected. For analysis purposes, we divided this duration into two timeframes. One is from 2011 to 2016, and the other is from 2017 to 2021. In this SLR, we have identified 14 different heterogeneity challenges that need to be addressed to implement IoT on a large scale. Challenges with occurrence percentage of more than 30% are defined as the most critical ones. In this study, we analyze the occurrence of those challenges based on digital libraries as well as timeframe. In this analysis, we found out that some challenges were more critical in the earlier timeframe than in the recent timeframe.
We also found out that some challenges are still critical in both timeframes. After identifying those challenges, we found at least 5 solutions for each identified challenge. e summary of those solutions is given in Table 6. In future work, we want to categorize the challenges and prioritize the solutions by using a multi-criteria decision-making problem.
Data Availability e data collected during the data collection phase will be provided by the corresponding author upon request.

Conflicts of Interest
e authors declare that there are no conflicts of interest.