See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/329408769

A Study of the Advances in IoT Security

Conference Paper · September 2018

DOI: 10.1145/3284557.3284560

CITATIONS READS

4 1,623

2 authors:

Andrew Dean Michael Opoku Agyeman

The University of Northampton The University of Northampton
2 PUBLICATIONS   4 CITATIONS    85 PUBLICATIONS   287 CITATIONS   

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

A survey of recent contributions of high performance NoC architectures View project

Quasi-Arithmetic Scoring Theory for Edumetrics View project

All content following this page was uploaded by Michael Opoku Agyeman on 26 April 2019.

The user has requested enhancement of the downloaded file.

 A Study of the Advances in IoT Security
Andrew Dean and Michael Opoku Agyeman
Department of Computing, University of Northampton, Northampton, United Kingdom

Abstract— The Internet-of-things (IoT) holds a lot of benefits

to our lives by removing menial tasks and improving efficiency of
everyday objects. You are trusting your personal data and device
control to the manufactures and you may not be aware of how
much risk your putting your privacy at by sending your data
over the internet. The internet-of-things may not be as secure as
you think when the devices used are constrained by a lot of
variables which attackers can exploit to gain access to your data /
device and anything they connected to and as the internet-of-
things is all about connecting devices together one weak point can
be all it takes to gain full access. In this paper we have a look at Fig. 2. Smart Fridge IoT
the current advances in IoT security and the most efficient
methods to protect IoT devices.
The internet-of-things is very vast which covers a range of
Keywords— Hardware, IoT, Security, Efficiantcy domains, which have specific architectures based on their
needs. IoT Technology has a lot of real world applications in a
I. INTRODUCTION wide range of fields such industry automation which can
streamline the manufacturing process and optimize efficiency.
The internet-of-things is how physical devices
interconnect together and is made up of billions of devices
which use wireless technologies to communicate. The IoT II. IOT ARCHUTECTURE
world is growing rapidly with an estimated 8.4 billion
connected devices in 2017 with is an increase of 31% and is A. Layers
expected to grow by another 33% by 2018. The IoT architecture can be split into three basic layers,
however they can change based on the use case’s as some
Category 2016 2017 2018 2020 industry solutions may require further layers.
Consumer 3,963.0 5,244.3 7,036.3 12,863.0
Business: 1,102.0 1,501.0 2,132.6 4,381.4
Cross-
Industry
Business: 1,316.6 1,635.4 2,027.7 3,171.0
Vertical-
Specific
Total 6,381.8 8,380.6 11,196.6 20,415.4
Fig. 1. IoT Devices (Millions of Units) source : Gartner (January 2017)

IoT works by receiving data from there surrounding

in the physical world using sensors sensors, processors and
communication hardware and then act on the data. These
devices are often called “smart” devices and can talk to other Fig. 3. IoT Layer Diagram
devices using commination standards. And then act of the data
received. Fig. 2. Shows an example on how a smart fridge B. Preception
works on the IoT platform using a sensor (Perception layer) to
The perception layer is made up of physical devices such
detect the current temperate which can then be sent using
as your sensors and actuators which interact with other devices
WIFI (Network layer) to a mobile device and displayed on a
and the physical world to both send and receive data to other
app (application layer) which could be used to change the
devices using wireless technology. [1] This layers objective is
desired temperature to send back to the fridge to act on.
to collect all the information from its sensors and actuators.
Which can be sent to the network layer.
C. Network processing power and will come with a lower cost making
The network layer handles the data been sent between them more efficient then using a device with high
smart devices as well as network devices and servers and can specifications.
be also used to transmit and process perception layer data into
a readable format for the receiving device. Some say this is B. Communication
where the Internet-of-things happens as it bridges the cyber IOT to work needs a way to send to send the data between
and physical world which allows them to interact with each devices to both gather the needed data and receive instructions
other, this layer uses a range of technologies (Routers, based on the sent data. Depending on the deployment of the
switches, cloud computing) which processes and directs the device it could use a range of commutation technologies and
data to the specific application layer where the data can be there are both short and long-range standards.
read. many communication technologies are uses to transmit
the data which can depend on the deployment of the
Name Frequenc Range Examples
perception layer, but this includes more than just WIFI as
other wireless technologies can have advantages such as y
Bluetooth for collecting a consolidating data from short ranges RFID 13.56 10cm - Road tolls,
or RFID. [1] MHz 200m Building Access,
Inventory
D. Application
EnOcean 315 MHz, 30 - Wireless switches,
The application layer which can also be called the 868 MHz, 300m sensors and
business layer delivers specific services to the users, and
receives the data from the sensors/actuators from the
902 MHz controls
perception layer after being translated into a readable format NFC 13.56 < 0.2 Smart
by the network layer. The application layer can then use this MHz m Wallets/Cards,
data to provide services or perform operations based on the Action Tags,
data received. This layer can analysis and store the data Access Control
received to create predications or see trends which could be Bluetooth 2.4GHz 1- Hands-free
invaluable for a company to see the current and future states of 100m headsets, key
their products/devices. [1] dongles, fitness
III. ENABLING TECHNOLOGIES trackers
WIFI 2.4 GHz, 100m Routers, Tablets,
A. Hardware Platforms 3.6 GHz + etc
There is an abundance of IoT hardware platforms and Weightle 470– Up to Smart meters,
development kits available, Fig. 3 shows a few of the most ss 790MHz 10km traffic sensors,
popular and recent released and you can see the specifications industrial
vary quiet a lot because there designed for handling different monitoring
solutions. GSM 850 - n/a Cell phones, M2M,
Processor Clock System Flash voltag
speed Memory memory e 900MHz smart meter, asset
Arduino Atheros 16Mhz 2.5KB 32KB 5V tracking
Yun AR9331
Raspberry Broadcom 1.2GH 1GB SD card 5V Fig. 5. IoT Commication standards
Pi 3 B+ BCM2837 Z (-32gb)
ESP8266 Tensilica 80MH 64KB 94KB + 3.3V Fig. 5. Shows some of the most used communication
L106 32-bit Z 16MB standards as well as some deployed examples. All come with
microcontrol external pros and cons such as RFID which is ideal for very close
ler
Beaglebon AM335x 1GHZ 512MB 4GB 5V
proximity commutation but lacks any security so is vulnerable
e Black ARM® to data hijacking, but the attacker would need to be very close
Cortex-A8 to do so which makes it impossible to do long range attacks.
Intel 22 nm 500M 1GB 4GB 3.3 –
Edison Intel® SoC HZ 4.5V C. Cloud solutions
Netduino Cortex-M4 168M 164 + kb 1408KB 3.3 – Cloud solutions are very important to the internet-of-things
3 wifi HZ + 2GB 5V
SD as it allows ubiquitous access to a shared pool of resources, all
UP Celeron 2.4GH 2GB 32GB 5V the devices in the perception layer can using the network layer
Squared N3350 Z send the information to be analysed and accessed by the
Fig. 4. IOT Hardware Platforms application layer.

The devices with lower specifications would be ideal Provider model capture visualiz Analytics
ation
for solutions with smaller tasks which would require less
 Amazon web IaaS Yes Yes Yes B. Processing
services Processing on a device is also another constraint to
Google cloud IaaS Yes Yes Yes implementing better security as the process will need to
Windows azure IaaS Yes Yes Yes perform they designed task as well as the security on top of
Rackspace IaaS Yes Yes No which could involve extra gates/transistors and additional
modules to do so.
Open Cloud
Engine yard PaaS Yes No No C. Design
Red hat Open PaaS No No No The design of a device is also a constraint as the size of the
shift device could also be a factor which implements extra modules
Heroku PaaS No Yes Yes / transistors will influence the devices designs size and
Salesforce SaaS No No Yes complexity. These implementations could include cost and
efficacy which could render the solution to be unviable for
Microsoft office SaaS Yes No Yes
deployment. By running simulations, the design can be tested
365 and optimized before building which will reduce the cost and
Google apps SaaS Yes No Yes make the device more efficient.
Zendesk SaaS No No Yes
V. IOT DEVICE EFFICIANTCY
Fig. 6. Cloud Providers
Making a deice more efficient has the benefit of making the
There are some different types of cloud computing which device not need as much power or resources to do the same
suit different solutions Fig. 6. Shows some of the most used job which will as a result decrease the cost.
providers as well as their features.
A. Code compression [3]
1) IaaS Code compression can be used improve a devices
Infrastructure as a Service is a model is where an performance and power consumption when used with
organization/business will rent out specific services needed for encryption and integrity checking to secure processor memory
there solution and is usual a “pay as you go” basis. This means transactions can reduce the memory footprint as well as
you will only pay for what you use unlike other services which providing more information per memory access.
your rent all their services for a fixed price even if you don’t
VI. IOT DEVICE SECURITY
use some.
With this increase in IoT devices we can see many
2) PaaS advantages for both consumers and businesses which
Platform as a service is designed to streamline the streamline a lot of processes, but they can also come with
development process by shifting the system management to some disadvantages. One of these disadvantages is security &
the provider and offering pre-configured components for privacy and having our personal data (banking information /
businesses/organization to use such as Databases/ application location / activity) being transmitted between devices comes
servers / programming languages. with the risk of losing a lot of our privacy. The IoT opens the
doors to many malicious hacks who wish to exploit IoT device
3) SaaS weaknesses to access our personal information to be used for
Software as a service is a cloud service which offers their own gain.
software on demand which are hosted and managed by the A. Authentication
provider and is normally a subscription and can improve
1) Noise Insertion [7]
communication and team collaboration.
Noise cancelation aim to protect the raw data when its
inside the computing unit to avoid an attacker from using side-
IV. IOT DEVICE CONSTRAINTS channel attacks to retrieve the data.

A. Power Consumption The way this method works is by inserting noise using a
Devices are built for a purpose and will be designed based key for sensitive data, although this method isn’t as secure as
on that purpose. And the more a device must do such as encryption it has the benefit of being very lightweight in
store/collection information will add to the power comparison. By selecting key locations where the data noise is
consumption. Adding extra security to a device will require canceled out making it readable you can keep the data secure
more power then the original design, implementation methods in the device and eliminate unnecessary overhead.
such as encryption will increase the power needed to complete
the same operation.
Fig. 7. Noice canselling scheme

2) Logic Locking [9]

Logic locking is a relatively new technique which includes Fig. 9. Security auditing module
adding extra gates to the design for locking “Key gates” which
would change the output and affectively lock the gates correct Using this module other time can create profiles
functionality. representing the devices behavior which can be analysed to
improve efficiency and detect threats quickly which could
prevent loss/compromised data.

2) Attack Detection Unit [8]

The purpose of the attack prevention unit is to detect when
a device is under attack and alert the device of this attack, so it
can prevent device damage or a compromised device.

Fig. 8. Logic Locking example

This technique increases the security level over other less

secure methods such as IC Camouflaging which is the method
of introducing dummy contacts, so an attacker could extract an
incorrect net list. In Fig. 7. some XOR gates are used as key
gates which need k1 and k2 to be 0 for the process to
continued otherwise the output would hide the original output. Fig. 10. Attack Detection unit Implimentation
When comparing logic locking to an older method (OC Cell)
there was a dramatic decrease in the delay without This is done by monitoring the electrical signals from
compromising the security. the communication controller which will detect abnormalities
in the physical characteristics of the bus communication
between the CM and main device. This is not part of the main
B. Detection & prevention
device it won’t interfere with the devices processing power
1) Security Auditing Module [4] and can be implemented on many devices with a
The purpose of the security auditing module proposed as communication controller.
apart of a security architecture is to monitor both internal and
external operations to evaluate the devices stability which will 3) Random Canaries Repository [2]
prevent device damage, alarm the network of any fatal issues RSR aims to protect against Stack smash attacks which
with the device and detect security threats. exploit vulnerabilities of the buffer overflow to hijack control
of applications. RCR is an enhancement of Stack Smash
protector by producing a repository of random values of
canary which are used when the application detects an attack
the RCR approach increase the difficulty of the attack. To
implement this method no special hardware is needed and can Secpage is both a hardware and software lightweight
prevent SSA against canary stacks with negligible overhead. architecture which aims to protect the devices memory by
providing an isolated memory environment which protects
sensitive code and data. This security method provides
compromised systems software with a secure isolated and
trusted environment to avoid data and code access buy
unauthorized users but also provides availability to easy
access to pages which don’t need to be secure which reduces
the overhead of the architecture implementation.

Fig. 11. RCR Approach

Fig. 11 shows the steps of the RCR approach, and in the

RCR Processes section the safe code is prepared and the area
of random canaries stack at the memory is utilized. Using the
time, date and application ID random values are generated an
stored in the RCR which can then be copied three times and
stored in heap memory. Making the RCR read-only with only
Fig. 13. Building a SecPage aware rogram
“const” variables will prevent attackers form being able to
access or alter and as the RCR is a global variable it doesn’t
need to be referenced in the stack. VII. COCLUSION
C. Isolation Security is an important issue with IoT as we can see,
1) Secure sensing [6] and with IoT expending to a lot of industries and services such
as industry and healthcare security will keep being more
A sensor framework has been suggested based on
important to protect sensitive data and devices against
hardware isolation which will protect the sensors on a device
damage. The perception layer is obviously the most vulnerable
from a compromised application. This is done using the
as it is deployed in the real world and normal embedded into a
hardware isolation feature on ARM processors, and by
deploying a sensor IP into the isolated area which is protected product which could be taken apart and interfered with and as
from compromised applications. it will normally have a connection other device using the
network layer which could be exploited to either steal or send
harmful data. With devices normally being constrained by
normally cost, design and efficiency it means we can’t add the
best security without overhead side affect which would make
the deployment either unviable or too expensive we need to
invest in more lightweight security which can’t help
preventing attacks. IOT has many benefits but is growing at a
rapid pace and some devices are stuck with outdated security
leaving them vulnerable to attack and are constrained by the
devices processing & power consumption limitations.

REFERENCES
[1] J. Lin, W. Yu, N. Zhang, X. Yang, H. Zhang and W. Zhao, "A Survey
Fig. 12. Secure sensing overview on Internet of Things: Architecture, Enabling Technologies, Security
and Privacy, and Applications," in IEEE Internet of Things Journal, vol.
By deploying shared memory in the normal world 4, no. 5, pp. 1125-1142, Oct. 2017.
which the secure world can access we are able to share the doi: 10.1109/JIOT.2017.2683200
sensors data when needed. The application can then request to [2] D. A. H. Shehab and O. A. Batarfi, "RCR for preventing stack smashing
attacks bypass stack canaries," 2017 Computing Conference, London,
the secure monitor in the secure world which can then switch United Kingdom, 2017, pp. 795-800.
the application mode from normal world to secure world doi: 10.1109/SAI.2017.8252186
which can verify the request has been authorized. The senor [3] E. W. Netto, R. Vaslin, G. Gogniat and J. P. Diguet, "A Code
data once the request has been verified and checked for any Compression Method to Cope with Security Hardware Overheads,"
potential threats can be written to the shared memory which Computer Architecture and High Performance Computing, 2007. SBAC-
can be read by the normal world. PAD 2007. 19th International Symposium on, Rio Grande do Sul, 2007,
pp. 185-192.
doi: 10.1109/SBAC-PAD.2007.40
2) Secpage [5]
[4] F. Ye and Y. Qian, "A Security Architecture for Networked Internet of Conference on Dependable and Secure Computing, Taipei, 2017, pp.
Things Devices," GLOBECOM 2017 - 2017 IEEE Global 146-152.
Communications Conference, Singapore, 2017, pp. 1-6. doi: 10.1109/DESEC.2017.8073840
doi:10.1109/GLOCOM.2017.8254021 [8] R. Jinnai, A. Inomata, I. Arai and K. Fujikawa, "Proposal of hardware
[5] K. Liang, Y. Feng, J. Wei and W. Guo, "SecPage - A Lightweight device model for IoT endpoint security and its implementation," 2017
Memory Protection Architecture," 2016 IEEE IEEE International Conference on Pervasive Computing and
Trustcom/BigDataSE/ISPA, Tianjin, 2016, pp. 1917-1922. Communications Workshops (PerCom Workshops), Kona, HI, 2017, pp.
doi: 10.1109/TrustCom.2016.0293 91-93.
[6] M. Ye, N. Hu and S. Wei, "Lightweight secure sensing using hardware doi:10.1109/PERCOMW.2017.7917533
isolation," 2016 IEEE SENSORS, Orlando, FL, 2016, pp. 1-3. [9] T. Thangam, G. Gayathri and T. Madhubala, "A novel logic locking
doi: 10.1109/ICSENS.2016.7808904 technique for hardware security," 2017 IEEE International Conference
on Electrical, Instrumentation and Communication Engineering
[7] Y. W. Lee and N. A. Touba, "Computing with obfuscated data in
(ICEICE), Karur, Tamilnadu, India, 2017, pp. 1-7.
arbitrary logic circuits via noise insertion and cancellation," 2017 IEEE
doi: 10.1109/ICEICE.2017.8192439

View publication stats