Machine Learning in IoT: A Report on Problems, Solutions, and Research

This report provides an overview of machine learning in the Internet of Things (IoT), focusing on identifying the importance of IoT devices in machine learning and evaluating the challenges associated with their integration. The literature review critically analyzes previous research, highlighting key aspects of both machine learning and IoT. The study reveals that IoT devices facilitate machine learning by enabling the analysis of large datasets. Security is identified as a significant concern due to potential data breaches. The report discusses problems such as configuration conflicts, data mining issues, and connectivity problems, and it also explores potential solutions and future research directions. It concludes by outlining the advantages and disadvantages of using machine learning in IoT, emphasizing the need for robust security policies and developer focus on security during IoT device development. Desklib provides access to similar solved assignments and resources for students.

IOT 0
Machine Learning in IoT

IOT 1
Abstract
IoT refers to the internet of things which is more effective computer technology
that provides a way of controlling and monitoring computer devices in the
organizations. Mainly, it connects physical devices with things and sensors by using
internet connection and exchange data between two or more networks. Machine
learning is a part of AI technology which delivers advanced systems that has the ability
to learn automatically and manage services and activities of an organization. The major
focus of this report is to identify the importance of IoT devices in machine learning and
evaluate challenges linked with machine learning in IoT. The literature review critically
reviewed the results and findings of previous papers and also analysed the key aspects
of machine learning and IoT. This study shows that IoT devices help machine learning
for analysing and evaluating a large number of data sets in an effective way. It is
identified that security is a common problem for IoT devices due to which companies
may suffer from data breach related issues. Therefore it is recommended that
developers should focus on security while developing IoT devices in machine learning
and companies should design effective security policies.
Table of Content

IOT 2
s
Introduction....................................................................................................................................................... 3
Milestone 1: Literature review................................................................................................................... 3
Milestone 2: Problems in machine learning in IoT............................................................................. 7
Configuration conflict................................................................................................................................ 8
Data mining problem................................................................................................................................. 8
A security breach or attack...................................................................................................................... 8
Connectivity issues..................................................................................................................................... 9
Milestone 3: Research issues in machine learning in IoT................................................................9
Milestone 4: solutions in machine learning in IoT........................................................................... 10
Milestone 5: future research..................................................................................................................... 10
Milestone 6: Advantages and disadvantages..................................................................................... 10
Advantages.................................................................................................................................................. 10
Disadvantages....................................................................................................................................... of 11
Milestone 7: Conclusion.............................................................................................................................. 11
References........................................................................................................................................................ 12

IOT 3
Introduction
Internet of thing is an advanced technology which is used for controlling and
connecting physical devices with computer networks. The use of IoT devices is
increasing quickly and many companies are using this approach for improving their
business. Machine learning is advanced innovation where the fundamental of the
internet of thing is used to communicate with machines (Cui, Kim, & Rosing, 2017). The
aim of this research is to explore the key idea of IoT and examine their importance in
the machine learning process. A literature review will be added in this research in order
to analyse the key findings and results of other journal papers. This study is divided into
several milestones for example literature review, a problem linked with machine
learning in IoT, research issues, solutions and future research.
Milestone 1: a Literature review
According to Endler, et al., (2017) machine learning is defined as the application
of AI technology which delivers systems for performing works and activities in an
automatic manner and it is an advanced technology which is now used in many business
industries. This journal paper highlighted the key aspects of both machine learning and
internet of thing and reviewed the significance of IoT technology in the area of machine
learning. In the past generation business industries used the human-based system for
controlling and managing their works which takes more time due to which recently
most the companies changed their IT systems and adopted IoT based technologies.
IoT plays a major role in the machine learning technology where it helps
machines and devices for effectively exchanging data or signals from one network to
another. After analysing this paper it has found that mainly, machine learning focus on
the enhancement of computer systems which can access data and utilize it learn from
themselves but IoT provide a way for controlling such kinds of computer devices very
easily. A recent study conducted by Kishore Ramakrishnan, Preuveneers, & Berbers,
(2014) identified that machine learning is a subsection of a computer system which is a
kind of information technology that helps machines and devices to learn without
explicit programming or source code.

IOT 4
Figure: Machine learning in IoT
(Source: Kumar, & Gandhi, 2018)
It includes internet of thing for controlling and managing the machines and other
physical devices from any location and it is a very appropriate technology for improving
the performance of businesses and services provided by the organizations. In this
journal paper, the author discussed the major key parts of the machine learning and
identified how the internet of thing is used in the development of machine learning
process. In the last few years there are numbers of authors provided their views on this
topic and evaluated the key concepts of machine learning and IoT technologies. This
literature review provides in-depth analysis about the research topic and Kumar, &
Gandhi, (2018) analysed that machine learning can be used for controlling the computer
devices and other mechanical parts in the industries. from this paper, it has found that
in order to analyse a large number of data sets gathered to form the internet of things
machine learning is more sufficient which effectively perform such kinds of activities.
Mainly, the internet of a thing contains various kinds of devices and sensors
which are plugged into the internet connections and able to exchange signals and
information between two or more devices. As compared with the other papers
Lavassani, Forsström, Jennehag, & Zhang, (2018) provided in-depth analysis about the
machine learning and internet of things and also conducted a literature survey for
improving the quality of research. This paper critically reviewed the challenges and

IOT 5
problems faced by companies while using IoT and machine learning technologies.
However, it is analysed that the security of data is the biggest concern associated with
IoT devices due to which consumers and organizations can suffer from the data breach
related issues. The authors also reviewed the advantages and drawbacks of the machine
learning which can help the other readers for better understanding of the research
topic. As per the findings of Li, Ota, & Dong, (2018) it is examined that the use of IoT and
machine learning is increasing very fast and it is expected that in the year 2022 around
50 billion systems and devices will be connected with these technologies. It is identified
that analysing and managing a large amount of data is a common problem for every
organization for which machine learning provided a platform to analyse and short these
data and facts.
Figure: Data about machine learning in IoT
(Source: Mahdavinejad, et al., 2018)
According to Mahdavinejad, et al., (2018) machine learning is an effective
technology in which developers uses the concept of IoT for improving the performance
of devices in an appropriate manner. Such kind of technology requires numbers of
sensors and networks for developing a learning approach in the machines and these
devices can be handled with a single technology that is the internet of things (Tang, Sun,
Liu, & Gaudiot, 2017). After analysing this paper it has found that the authors provided
theoretical facts about the internet of things which creates a gap in the research but this
literature section critically review the results of other papers. Meidan, et al., (2017)

IOT 6
reviewed that the largest benefit which machine learning brings to the internet of things
is the automation of analysis of the colossal amount of produced and exchanged
information. Analysing the data of consumers through the manual process takes more
times and also increases complexity for the companies and in the past generation, every
business industries used the same process. But in this modern generation, machine
learning changed the way of interaction and reduces the complexity of the data analysis
process.
However, the machine learning along with the internet of things can be used in
the case of predictive analysis where the machine can work more sufficiently as
compare to the individuals. A journal paper published by Pandey, (2017) evaluated that
a machine learning technology is not just to detect the abnormal nature but also to
support consumers and companies to analysis and develop long term trends. This needs
the biggest job of selection, sorting, detecting and processing a vast quantity of obtained
data for producing the comprehensive predictions. Internet of thing based devices also
helps the machine learning to critically monitor the overall performance of the business
industries.
Patil, & Thorat, (2016) identified that another problem linked with machine
learning is that it require human guidance and feedback process for controlling
automated devices and networks. In any case, the present AI innovation can't manage
without human direction and criticism, and this may not change for a long time.
Persistent rectifications and supervision is the thing that makes these frameworks
especially compelling in information investigation, particularly with regards to the
measure of information produced by IoT. Adding human experience and instinct to
oneself learning frameworks is as yet essential to keeping them destined for success. As
indicated by Shafique, et al., (2018) customers and associations may never get a 100
present precise and reasonable model freely worked by a machine. Be that as it may,
cooperating and giving direction to these machines is the best way to get to this
objective as close as could be expected under the circumstances. IoT has truly detonated
in the course of recent years, showing its potential in applications going from wearable
and robotized autos to shrewd homes and brilliant urban areas, making an effect all
over the place.

IOT 7
As indicated by an on-going examination by Gartner, there are around 16 billion
gadgets associated with the IoT now and this is relied upon to ascend to 25 billion by
2020. All such associated gadgets create a downpour of data that should be observed
and investigated, so they gain ceaselessly from the accessible arrangements of
information and improve themselves with no manual mediation. Shanthamallu, Spanias,
Tepedelenlioglu, & Stanley, (2017) identified that the fundamental reason behind
machine learning is to mechanize the advancement of various investigative models to
empower calculations to ceaselessly learn with the assistance of accessible information.
Google's self-driving vehicle is one of the best examples that utilize diverse machine
learning procedures with IoT in order to produce a self-ruling vehicle.
It joins the propelled highlights of various present-day vehicles which include
discourse acknowledgment, path help, versatile journey control, leaving associates and
guides. It is identified that IoT without a doubt utilizes the Internet in an astonishing
way, yet there are still a few difficulties, especially if it's executed without machine
learning. People have never played around with such huge numbers of gadgets
producing such tremendous measures of expanded information, ever previously.
Without using the internet of things in the machine learning can produce several
challenges and issues, for example, device diversity, device management, integration of
data from various sources, reduce the performance of systems, impact on the flexibility
of devices and so on.
After analysing literature review it has found that there are numbers of
requirements for this project, for example, appropriate data, data analysing methods,
evaluate security-related issues, and challenges linked with the machine learning in IoT.
Therefore, for this research project security issues, challenges in machine learning in
IoT and data analysis methods will be selected from the requirement. These will be
discussed in this report.
Milestone 2: Problems in machine learning in IoT
There are various problems and issues linked with the machine learning used
with the internet of things which are described below:

IOT 8
Configuration conflict
It is a common challenge faced by consumers while using the internet of things in
machine learning. The configuration is the first stage of machine learning where
developers configure computer networks, sensors and other electronic devices with the
machines and produce a new device (Ventura, et al., 2014). In the case of IoT issue of
configuration occur when developers or consumers do not follow proper instructions
and guidelines that also impact on the performance of machine learning.
Figure: Issues with machine learning in IoT
(Source: Zhang, et al., 2017)
Data mining problem
It is identified that the internet of things collect and evaluate large numbers of
data sets and machine learning uses these data sets for analysing purpose. So, in such
kind of process extracting the useful information from these data and signals is a major
challenge faced by the internet of things devices and servers. While using such kind of
technology in machine learning can produce a problem for future prediction and mining
data in the company (Zhang, et al., 2017).
Security breach or attack
It is a very serious problem occurred in the internet of things which directly
impact on the performance of computer devices and systems used in machine learning.

IOT 9
Mainly, both machine learning and IoT use internet connection for exchanging
information which is also linked with the data breach and security risk that impacts the
user’s private data. It is observed that hackers attack the sensors and devices used in
IoT while implementing machine learning after that they block their servers and collect
the personal data and signals of companies or industries.
Figure: Security issues in IoT
(Source: Endler, et al., 2017)
Connectivity issues
It is analysed the IoT interconnected devices or system with the other computer
networks for exchanging data but due to lack of proper connections the consumer may
suffer from connectivity issues. Machine learning uses the concept of IoT and also
connects their servers with the IoT devices in which hackers transfer the traffic signals
to the main servers which increase connectivity related issues in the system. However,
machine learning and Iot both use internet connection for exchanging information due
to which consumers may suffer from the connection problems because it requires
proper configuration and more reliability in the system.
Milestone 3: Research issues in machine learning in IoT
Machine learning in IoT is a very common topic for the investigation and there
are numbers of researchers provided their opinions on this topic. From recent papers, it
is identified that machine learning is a broad area of research and it is very complex for

IOT 10
the authors to select the appropriate points and topics. However, collecting data and
analysing these data both are very common research issues faced by the authors while
conducting research in the machines learning and IoT. In which gathering the effective
and reliable data is a serious problem because there are numbers of facts available on
the internet but selecting the appropriate data is not easy which require more time and
efforts. Moreover, there are many other research issues occurs in the machine learning
in IoT such as supporting arguments, integration of collected data, obtaining data from
multiple sources, analysing collected data sets, identifying challenges linked with the
IoT, and so on.
Milestone 4: solutions in machine learning in IoT
For solving problems linked with the machine learning and IoT companies
should focus on the security and data integrity while developing an automatic device.
Moreover, collecting data and analysing data these issues can be solved by adopting the
appropriate research methodologies in the investigation. These methods include
research design, research approach, primary and secondary data collection method,
descriptive statistical analysis method and SPSS tools. For the security of data,
companies can develop security policies and provide a proper connection to the
machine learning in IoT.
Milestone 5: future research
In future research, the authors will reduce the drawbacks of this investigation by
adopting effective research methods. Moreover, the authors will analyse the security
risks linked with machine learning in IoT and provide an in-depth analysis of their
advantages and disadvantages.
Milestone 6: Advantages and disadvantages
Advantages
 More effective
 Can be used for controlling devices and services
 Can be used in the companies for improving their performance
 Large reliability and flexibility

IOT 11
 Provide data analysis process
 Reduce human efforts and errors
Disadvantages
 More costly
 Increase security risks and threats
 Require proper configuration
 Designing and implementing these devices are more complex
Milestone 7: Conclusion
From the above analysis, it is concluded that IoT plays a crucial role in machine
learning where it controls and manages collected data in an effective way. This report
identified the importance of the internet of things in machine learning and critically
reviewed the opinions of other authors. It is identified that security is a very serious
concern in IoT and machine learning due to which business industries can suffer from
data breach related problems. Therefore, it is suggested that while developing IoT based
devices developers should focus on the security and privacy and companies should
develop effective security policies in their workplace.