Enhanced Protection of Ecommerce Users’ Personal Data and
Privacy using the Trusted Third Party Model
Mukuka Kangwa
1a
, Charles S Lubobya
1
and Jackson Phiri
2b
1
Department of Electrical Engineering, University of Zambia, Lusaka, Zambia
2
Department of Computer Science, University of Zambia, Lusaka, Zambia
Keywords: Personally Identifiable Information, Data Privacy, Electronic Services, Data Protection, Random Electronic
Identity, Hardware and Software.
Abstract: The rapid adoption of electronic delivery of services by various electronic service providers such as
ecommerce and e-governance services leaves the users of these services with no option but to adapt if they
are to continue accessing their desired services. To access these services, very often one has to reveal some
of their personal data in order to get registered on the platforms made available courtesy of the service
providers. One person is likely to surrender their personal identifying data to several service providers hence
making their aggregated data susceptible to leakage online. Despite several solutions already in use data
leakage is still prevalent. Our research proposes and tests a method that aggregates personal identifying data
and seeks to enhance its protection from leakage using a novel approach formulated from software and
hardware. This paper outlines the design and explains in detail how the approach is expected to protect data.
It further gives details of the results that were obtained from experiments conducted on the constructed key
component of the proposed solution.
1 INTRODUCTION
The information Age has seen an unprecedented rise
in the delivery of various services using electronic
means. Most of the providers of electronic services
such as ecommerce and e-governance require a
person to submit some elements of their Personally
Identifiable Information (PII) before they could grant
that potential service consumer access to their
electronic services (Kangwa, Lubobya, & Phiri,
2021). This has resulted into huge amounts of
aggregated PII being collected across number of
service providers and thereby making it vulnerable to
intentional or inadvertent leakage (Patent No. US
2019 / 0333054 A1, 2019). Leaked PII puts the owner
of the information at high risk; users can have their
Bank account broken into, their privacy compromised
and even pose physical threat to the victim. Despite a
number of solutions having been formulated and
implemented in order to address this challenge, the
problem persists. With the advent of the Covid-19
a
https://orcid.org/0000-0002-4568-7497
b
https://orcid.org/0000-0002-4430-1580
Pandemic across the globe, more service providers
have opted to use electronic means to deliver their
services to their clients to minimize physical contact
hence putting more PII at risk of being leaked. In fact
a number of incidents have already occurred where
data privacy has been breached (Hauer, 2015).
There are several methods and approaches such as
cryptography that are being used to provide
confidentiality and privacy to data (Pawar & Harkut,
2018). This paper proposes a simplified, yet effective
method that can be employed to protect personal data.
It builds on what other scholars have formulated to
come up with a more effective approach.
2 RELATED WORK
A number of scholars have proposed varying
approaches to help protect personal data while
allowing the owners access to online services. Frank
and Michael proposed and patented a solution to help
116
Kangwa, M., Lubobya, C. and Phiri, J.
Enhanced Protection of Ecommerce Users’ Personal Data and Privacy using the Trusted Third Party Model.
DOI: 10.5220/0010576201160126
In Proceedings of the 18th International Conference on e-Business (ICE-B 2021), pages 116-126
ISBN: 978-989-758-527-2
Copyright
c
2021 by SCITEPRESS Science and Technology Publications, Lda. All rights reserved
protect personal identifying data. In their proposal
they have a Trusted Party that provides static
Identities (ID) to users. In addition, they proposed the
use of Block chain technology to protect personal
data for the users. In the scenario shown in Figure 1
below, the user obtains an ID from the digital ID
provider and submits it to the service provider as
proof of identification. The service provider verifies
with the ID provider if the user is indeed genuine and
the response the ID provider returns determines
whether or not a service will be rendered to the user.
Frank and Michael are proposing the use of an offline
escrow to keep the identifying data to be accessed via
legally approved means. They further submitted that
pseudonymization rather than anonymization be used
to make it possible to identify a user when need arise
(Patent No. US 2019 / 0333054 A1, 2019).
Figure 1: Proposed Approach by Frank and Michael (Patent
No. US 2019 / 0333054 A1, 2019).
The implied use of static electronic IDs in the
proposed approach above might not be sufficient to
help maintain the privacy of the users as a static ID
can be profiled easily and hence compromise the
privacy of the owner while surfing online (Kangwa,
Lubobya, & Phiri, 2021). Furthermore, the use of
Blockchain technology might not be very feasible as
the technology is currently resource intensive (Bao et
al., 2020). To provide a global solution would
consume a huge amount of resources for the proof of
works to be used to protect data from being leaked or
modified or even deleted. The Blockchain technology
will require some modifications to make it friendlier
to the proposed approach. Lighter Blockchain-like
implementation on a local scale might be more
feasible than a peer-to-peer global approach that is
currently wide spread.
Another key challenge with Blockchain
technology is its lack of scalability due to its design.
The solution seeks to solve a global problem by
providing access that is global hence scalability is
very key to accommodate everyone who desires to
use the proposed solution. Moreover, Blockchain
faces some privacy and security challenges hence
might provide a solution to one problem and
introduce more challenges (Bao et al., 2020) . In
addition, the time lag that the technology inherently
experiences due to its design poses a huge challenge.
When one node generates a transaction, a number of
other nodes need to confirm and reach consensus
before a transaction can be considered as complete.
This results in delayed completion of transactions (II-
Agure, Belsam, & Yun-ke, 2019). Further, its
complexity makes the cost of building and
maintaining it prohibitive. Cheaper ways of
developing the technology need to be sought if it is to
be widely adopted. Perhaps build as a service where
costs can be shared (Zhang, Alkubati, Bao, & Yu,
2021). This paper proposes the use of a Trusted Third
Party, herein called KYC Agency, which would be a
government appointed entity tasked with the
registration of its citizens for the purpose of issuing
National IDs. It further proposes a solution to be
employed to effectively prevent PII leakage by the
appointed TTP and is a buildup of our earlier work in
the paper “Prevention of Personally Identifiable
Information Leakage in E-commerce via Offline Data
Minimisation and Pseudonymization” (Kangwa et al.,
2021).
3 TRUSTED AND UNTRUSTED
THIRD PART MODELS
The leakage of PII has been going on for some time
now. Ecommerce platforms normally hold client
information online so that they can identify them
before offering any service. A lot of user data is being
held in the cloud by various service providers hence
making that data susceptible to leakage (Ye, Dong,
Shen, Cao, & Zhao, 2019). For example, e-bay, an
Ecommerce platform, suffered a Distributed Denial
of Service (DDoS) attack where their databases were
scanned and client data was exposed (Innab &
Alamri, 2018). This attack was possible because the
databases were accessible online.
Fanghan et’al proposed a model that would
control access to data using a system whose security
was to be enhanced using encryption. The system was
aimed at replacing a Trusted Party model as they
believed that the Trusted Third Party (TTP) might not
be so reliable to control access to user data on behalf
of the user. They proposed a model that would give
the data owner the power to control who accesses
Enhanced Protection of Ecommerce Users’ Personal Data and Privacy using the Trusted Third Party Model
117
their data. Figure 2 below gives a summary of the
model:
Figure 2: Model without Trusted Third Parties (Ye et al.,
2019).
According the model, a cloud server would be
required to keep some of the data. The user encrypts
their documents before sharing and decides who can
be given access by sharing their decryption keys with
the people who need access (Ye et al., 2019). The fact
that some data is kept in the cloud makes that data
vulnerable to leakage. Even if it is encrypted, if the
key shared finds itself in wrong hands, then the
encrypted date, if accessed can be deciphered. In
addition, it means several can be granted access to the
data thereby increasing points of potential data
breach. Moreover, despite being encrypted, data is
most likely to be in plain text when being processed,
for example in response to a request for data, hence
making it vulnerable to leakage (Zhan, Fan, Cai, Gao,
& Zhuang, 2018).
Aarthy et’al proposed a Trusted Third Party
Model that was aimed at resolving the issues of trust
by users of cloud services. Users had huge concerns
over security and trust of cloud service providers
hence were hesitant with surrendering their data to the
service providers. Their proposed model Sought to
address this challenge by providing a Trusted Third
Party that would monitor and assess the cloud service
providers and provide assurance to users (Aarthy,
Aarthi, Farhath, Lakshana, & Lavanya, 2017). Figure
3 below gives a summary of how their proposed
model was expected to work.
Figure 3: Trusted Third Party Model (Aarthy et al., 2017).
The idea was to have a party that both the service
providers and potential users of the services can
trust. The data about the quality of service the cloud
service providers are offering is aggregated together
and held by a central and neutral party. The Trusted
Third Party provides that service. The approach was
expected to improve trust and eventually the
adoption of cloud services. This model can be
applied in other areas such as the provision of Know
Your Customer (KYC) services though in this case
the objective is to protect the PII of potential
electronic service users kept in a central place.
Protecting data in one place is much easier than
protecting data distributed across various online
platforms.
Locher et’al, proposed the use of a distributed
ledger to replace the use of a Trusted Third Party
(TTP) (Locher, Obermeier, & Pignolet, 2018). As
proposed by other authors earlier considered, the
distributed ledger owes its security around the
approval of any transaction to its requirements for
consensus before any transaction can be confirmed.
However, this approach results in delayed
transaction completion as well as huge resources
being required to make the technology operational
(Bao et al., 2020). Locher et’al acknowledged
that despite the distributed approach of using
block chain technology, users still needed to
trust each over (Locher et al., 2018). The aspect
of trust is what the Trusted Third Party model
aims to address hence the use of blockchain
technology might have a limited application to
certain use cases where trust might not be an
immense requirement.
The Trusted Third Party Model was also tabled
by Jamshiya et’al, to provide trust amongst Internet
of Things (IoT) devices where security and trust
establishment is a challenge. For these devices to
connect to each other and start sharing data, trust
needs to first be established. To achieve trust, a third
party that can be trusted by both parties needs to be
in place. The TTP then generates a key that is
distributed to all parties that need connecting to each
other. Each IoT device is first connected to the TTP
and assigned an id to be identified by. Then the TTP
can now be used as a Trustee to tell other devices
that desire to connect whom they can trust and
connect with. This, off course, is done via the
encryption of exchanged keys. Elliptic Curve
Cryptography (ECC) is employed because it
provides strong cryptography with a smaller key
ICE-B 2021 - 18th International Conference on e-Business
118
length (Jamshiya & Menon, 2018). Figure 4 below
shows their proposed model:
Figure 4: Trusted Third Party for IoT Devices (Jamshiya &
Menon, 2018).
The major element in the model was the
inclusion of a party that establishes trust in advance
with different parties (in this case IoT devices) that
might potentially connect to each other in future.
More scholars were considered regarding the
protection of PII and privacy of users while online.
Peter et’al recognizes the General Data Protection
Regulation (GDPR) as one way of addressing the
prevalent consumer privacy challenges. The
European Union proposed the GDPR as a way of
protecting the privacy of individuals by promoting
pseudonymization of PII in addition to already
existing data security techniques (Štarchoň &
Pikulík, 2019). The additional measure indicates
that the existing techniques applied by various
service providers are not sufficient hence the data
leakages and privacy violations that are experienced
from time to time. Peter et’al defined
pseudonymization as transforming data in such a
way that the resulting data cannot be associated with
the original owner without additional data. That is,
if one was to stumble upon pseudonymized data,
they should not be able to identify the owner without
requiring additional information to fill in the blanks.
The authors propose that Pseudonymization
techniques be applied by various data processors
such as mobile operators to protect user privacy.
Techniques such as scrambling or obfuscation,
blurring, masking, tokenization and encryption were
proposed (Štarchoň & Pikulík, 2019).
Pseudonymization requires that, if legally
demanded, the transformed data must be traceable
to the real owner via the data processors. The
challenge with having data scattered across various
service providers is that, the probability of that
information being leaked remains high as any of the
service providers holding the data might be
compromised.
Sergio et’al are of the view that the advancement
in technology has resulted in the need for more
effective techniques and solutions to provide
security and privacy to personal and other sensitive
information. They contend that current solutions
might not be sufficient to meet the required levels of
privacy and security demanded by regulations such
as the European GDPR (Ribeiro & Nakamura,
2019). The team proposed the use of methods
such as pseudonymization and anonymization.
Pseudonymization was preferred to anonymization
as they intended to use their solution for the
protection of Health data for children.
Pseudonymization provides a possibility of
identifying the actual individual using
additional information when need arise.
Anonymization, on the other hand, alters data in
such a way that it can no longer be traced back to the
actual owner.
The team further proposed combining
pseudonymization with other security techniques
such as hashing of pseudo IDs, encryption of
pseudonymized data and so on (Ribeiro &
Nakamura, 2019). It must be noted that as long as
data is kept online, the possibility of being leaked
remains. There is need to ensure that only
pseudonymized data is made online while raw
identifying data is kept unreachable from the
internet. Furthermore, necessary internal controls
must also be put in place to ensure data is not leaked
by internal parties.
4 SOLUTION DESIGN
As long as information is available online, the
possibility of someone accessing that information
without authorization remains despite the various
mechanisms used to protect that information. Our
considered view is that the best way to protect
information is to make it unavailable to the online
Enhanced Protection of Ecommerce Users’ Personal Data and Privacy using the Trusted Third Party Model
119
hackers hence the proposed design in Figure 5
below:
Figure 5: Know Your Customer Agency Operation.
The approach proposes the use of a Trusted Third
Party, herein called KYC Agency, which would be a
government appointed entity tasked with the
registration of its citizens for the purpose of issuing
National IDs.
The model demands that other service providers
requiring KYC confirm with the KYC Agency if the
requesting party is genuine and the KYC agency
provides assurance without sharing the PII of the
requesting party. This enables the requesting party to
have access to services without risking their PII and
privacy.
The approach will operate as follows:
4.1 User Registration
The user will first register with the KYC Agency in
their country of residence. They will submit Personal
Identifiable Information (PII) such as their National
Identification documents, Residential address,
contact details such phone numbers and email
addresses.
It is recommended that the KYC Agency be the
same institution that issues citizens with their
National Identification Documents such as passports.
This will ensure that whenever a citizen is issued with
an ID even for the first time, they are issued with one
that can also identify them electronically.
Once the user has satisfied requirements for
registration with the KYC Agency, the KYC Agency
creates a record with full Identifying Information of
the user and appends a universally unique ID on the
record. The data is kept on the “Offline” system that
is not accessible from the internet.
Figure 6 below gives a detailed flow of the
registration process.
After the eID has been appended to the new user
record, the eID is pseudonymized (eIDs) using a
predetermined algorithm and sent to the online
system for the creation of an online record for the
user. The pseudo version of the unique ID, eIDs, is
not appended to the record sitting on the offline
system. This is to minimize the possibility of
associating offline data to online pseudo IDs if they
are leaked for some reason by insiders.
It must be noted that the only communication
between the Offline system and only system will be
the automatic transmission of the Pseudo ID, eIDs, to
the online system. The transmission will be
determined by the firmware sitting on the
microcontrollers as will be explained under the
operations of the Data protector system.
When the online system receives the pseudo ID,
eIDs, it will automatically create a record with the
eIDs as the primary key. It will then create an
anonymous email box for the user. The mail box is to
be used for delivering Random eIDr to the user. In
addition, the emails will be automatically destroyed
after a predetermined period to prevent formulation
of the Key/algorithm being used for the generation of
random IDs in case the online system with mail boxes
is compromised. A chain of emails with various
random IDs might be used to crack the key and
algorithm used to generate the random IDs.
4.2 Proposed Universally Unique
Electronic ID (eID)
The following format of the universally unique ID is
being proposed:
The eID will comprise 10 digits representing the
unique ID for the person being created on the system
and 3 digits representing the country the person is a
citizen of as shown below:
XXX.XXXXXXXXXX
The 10 digits for the universally unique portion of
the ID is to accommodate for the growth of
population for counties like China and India. The 3
digits for country is to accommodate the number of
existing countries in the world and the new countries
that might image. The first Citizen to be registered for
example would have the eID shown below:
KYCAgency
EcommerceWebsite
3.Requestaccesstowebsite
WebsiteUser
4.RequestforRandomIDeID
R
5.SubmiteID
R
8.RespondAppropriately
1.Authenticate&RequestforeID
R
2.GenerateandshareeID
R
OnlineSystemwithPseudonymizedData OfflineSystemwithFullKYCdata
Pseudonymi zeddata
KYCAg encyUsercreatingNewrecords
X
UserVisits KYCAge ncyforIDc reation
Collect
Personally
Identifiable
Datafromthe
person
ICE-B 2021 - 18th International Conference on e-Business
120
Figure 6: KYC Registration Process.
4.3 Data Protector Operations (RMS)
The Data protector that will safeguard the Personally
Identifiable Information will connect the offline
system to the online system and operate as follows:
Data exchange between the two systems will only
flow in one direction; that is, it will only flow from
the offline system towards the online system as
depicted in Fig1. The aim of this restriction is to
ensure that no one is able to access the PII from the
Internet. This is to reduce the possibility of a hacker
accessing the PII without needing physical access to
the server hosting the sensitive data (Kangwa et al.,
2021).
Figure 7 shows the summary of how the Data
protector (Restricted Memory System) will be
operating:
Data Minimization
System (Offline)
Online System
Restricted
Memory
system
Ecommerce
1
Ecommerce
2
Ecommerce
X
KYCPrivacyAgency‐KPA
Internet
User
One‐waydataflow
Two‐waydataflow
Key
Figure 7: Data Protector Operations.
Furthermore, despite data being able to flow
towards the online system from the offline system, to
prevent huge amounts of data from being sent by
disgruntled elements inside the KYC Agency using
the offline system, there is a bandwidth restriction
imposed between the two systems. We propose using
the lowest possible serial data speed available. For
Enhanced Protection of Ecommerce Users’ Personal Data and Privacy using the Trusted Third Party Model
121
example, if we wanted to send 10gigabyte of data
from the offline system to the system via a serial
connection of 9600bps, it would take more than 100
days to complete the transfer of data. Slower speeds
would take even longer. However, transferring
pseudo IDs would take few milliseconds as the strings
would only constitute few kilobytes of data per
unique record created at a given time. The slow rate
of data transfer would discourage a hacker or
disgruntled element from attempting to do so if they
somehow managed to attempt, they would be
discouraged by the estimated time of data transfer and
hence abandon the theft.
Moreover, the system would periodically reset the
connection between the two systems hence disrupt
any exploitive data transfer in session as legal
sessions will be expected to only last few
milliseconds.
(i) User Transaction with Ecommerce Sites.
The user will either access the KYC Agency to
generate a universally unique random ID, eIDr, or
first access an ecommerce site or request to transact.
The site will request the user to submit their random
ID issued by the KYC Agency. The sites will not be
allowed to collect PII from users to prevent data
leakage prevalent with online services.
The user will need to Logon to the KYC Agency
system via a website or app and request a unique
random ID. The KYC will authenticate the users via
existing identification methods such as Google
authenticator or any other multifactor authentication
method. The user remains anonymous using the
records kept by the online system.
User Logons on to the Website or App for the
KYC Agency
Once authenticated, the user generates Random
ID eIDr. The KYC system sends the ID, r, to
Anonymous email or is displayed on an App. Then
the user enters the eIDr on the website. The website
verifies with KYC Agency if they issued the eIDr
supplied by the user. Depending on the feedback of
the Agency, the website either grants or denies the
user access to their services.
Figure 8 below gives a pictorial view of how the
transaction will flow from the beginning to the end.
The KYC Agency system will host the mail boxes
for the users and will destroy emails containing
random IDs after the predetermined validity period
elapses.
There is need to ensure that only pseudonymized data
is made online while raw identifying data is kept
unreachable from the internet. Furthermore,
necessary internal controls must also be put in place
to ensure data is not leaked by internal parties.
Controls such as ensuring that the hardware system
hosting PII is not fitted with external media devices
such as writeable DVD drives, USB drives, Bluetooth,
wireless and so on. External tape can be connected for
mass backup. The contents on tape must be encrypted.
Figure 8: Anonymous Ecommerce Transaction.
5 EXPERIMENTAL
METHODOLOGY
To ascertain the effectiveness of the proposed
solution, tests were conducted using various methods
and tools. Only the Data protector was built as other
components proposed could be easily built using
existing approaches. That’s, components such as
websites do exist already while the offline system for
storing data is similar to other systems except it will
be kept offline focus on the proposed algorithms for
pseudonymization and creation of traceable random
IDs.
The Data protector otherwise, known us
Restricted Memory System (RMS), was built using
the following components; two Arduino UNO
microcontrollers, copper cables, serial ports, serial
monitors, python programming language, Arduino
UNO IDE and Proteus Simulation software.
The design used serial communication to build the
RMS in preference to parallel communication. The
aim was to ensure that data could only flow in one
direction at limited amount of bandwidth. Two way
communication would require two cables physically
connected between the two devices communicating to
ICE-B 2021 - 18th International Conference on e-Business
122
each other. To ensure one way communication, one
cable was physically removed as shown in Figure 9
below.
Figure 9: Simplified Circuit Diagram (Kangwa et al., 2021).
This would ensure that even if the online
component wanted to communicate to the offline
component, the communication would not be
successful as there would be no means of reaching the
other side.
The design uses two microcontrollers instead of
one to control the security of the system. Hardware is
susceptible to hardware Trojans. These viruses can be
put into the hardware at manufacturing stage as the
manufacturer can modify the design to put in place
Trojans that can be used to deliberately leak
information. For the Trojans to be activated, a hacker
or disgruntled manufacturer would need access to the
hardware either physically or remotely (Ali,
Chakraborty, Mukhopadhyay, & Bhunia, 2011). One
Arduino UNO connecting the offline side was
configured as a Master while the one facing the online
system was configured as a slave as shown in the
circuit in Figure 10 below:
Figure 10: Physical Configuration of RMS.
Even if the Arduino facing the online system was
compromised, the hacker would not be able to breach the
entire connection as they would need access to the
Master Arduino to change configurations and enable
two-way communication thus making it impossible to
achieve without having physical access. This is what
makes the solution simple, yet very effective, if built and
deployed as designed.
Tests were conducted by sending data in both
directions. That is, data was sent from the offline
system to the online system via the RMS as per
configuration outlined. Data was also sent from the
online system to the offline system. Six scenarios
were tested. In scenario 1 and 2, both the Transmitting
and receiving PINs were physically connected while
in Scenario 3 and 4, the cable connecting
Transmitting PIN for the Master Arduino to the
receiving PIN of the Slave Arduino was
disconnected. In scenario 5 and 6, the receiving PIN
of the Master Arduino connected to the Transmitting
PIN of the slave Arduino was disconnected.
The Bandwidth between the two Arduinos was set
at 9600bps. This speed can be adjusted as desired.
The lower the speed, the more time will be required
to send huge amounts of data hence the more
frustrating to the hacker.
6 RESULTS AND DISCUSSION
Table1 below gives a summary of results that were
obtained from various scenarios.
In scenario 1 and 2 covered by Test number 1 in
the results of Table 1, Data was successfully sent both
ways with the Transmitting and receiving PINs
connected correctly on the Master and Slave
Arduinos.
In Scenario 1 and 2 confirmed by test 1, with both
Transmitting and receiving PINs connected correctly
on the Master and Slave Arduinos, data was
successfully sent both ways. That is, data was able to
flow from the offline system holding PII towards the
online system susceptible to hacking and vice versa.
This was a fail as the objective was to ensure that
data could only flow in one direction. That is, from
the offline system towards the online system. In this
scenario data successfully flowed in both directions
hence the Data protector cannot protect PII by
preventing access by users connecting from the
Internet. It is vital that data cannot flow from the
online system to the offline system even if bandwidth
is restricted as malware such as ransomware can be
created as a very small payload and yet cause serious
damage to data once deployed into the offline system
holding PII. The chances of one infecting the offline
system, if they cannot access it from the internet, is
very slim as they would need physical access to the
offline system. Hence the test results for the first
scenario renders the configuration undesirable and a
fail.
Enhanced Protection of Ecommerce Users’ Personal Data and Privacy using the Trusted Third Party Model
123
Table 1.
In the Scenario covered by test 2 with transmitting
PIN on the Master Arduino not connected to the
receiving PIN on the Slave Arduino while the
receiving PIN on the Master Arduino remained
connected to the transmitting PIN of the Slave
Arduino, data could not flow in any direction.
In the scenario depicted in test 3, with the
transmitting PIN of the Master Arduino connected to
the transmitting PIN of the receiving Arduino, while
the receiving PIN.
While the receiving PIN is disconnected from the
transmitting PIN of the Slave Arduino, data was sent
successfully from the Master Slave Arduino
connecting the offline system towards the Slave
Arduino connecting the Online system but data could
not be sent from Slave Arduino connecting the online
system towards the Master Arduino connecting the
offline system.
With the speed/bandwidth of 9600bps configured
between the two Arduinos, 10 GB of data would take
about 103days to transfer across the Data protector
from the offline system holding PII to the online
system susceptible to hacking.
Test 2 resulted in data failing to flow in any
direction. This too was a fail as the main objective of
the proposed solution was to allow automatic creation
of online pseudonymized records for users while
preventing access to the PII data by users with access
to the internet.
With data not flowing to the online system from
the offline system, it would require another approach
of transferring pseudo data matching records on the
offline database to the online system. That approach
might be manual hence introducing another risk
where if data has to be moved using external media
then malware can be introduced onto the offline
system using that media. Therefore, this
configuration is not desirable and was a fail.
Test 3 was successful as data could only flow in
one direction. That is, data could only flow from the
offline system towards the online system. This was
the desired configuration as it would prevent hackers
successfully accessing PII data sitting on the offline
database. It would also prevent malware from being
introduced from the online system to the offline
system as it would corrupt the PII data sitting on the
offline system. This result shows that it is possible to
keep sensitive data “offline” while allowing real-time
connection between the offline system and online
system for the creation of corresponding records for
the user to access online services anonymously once
created on the KYC system.
The restriction of the bandwidth between the
offline and online system to 9600bps ensured that
only minimal data could pass across at any given
time. For example to transmit 10GB of PII would take
more than 100days. It is very possible to reduce the
bandwidth further and increase how long it would
take to transmit reasonable amount of data across the
data protector component. The restricted bandwidth
would discourage both external and internal
disgruntled elements from attempting to steal PII
ICE-B 2021 - 18th International Conference on e-Business
124
sitting on the offline system. In addition, the valid
data transmissions across the two systems are short
bursts of few characters. The restriction helps prevent
theft of sensitive PII by both internal and external
parties.
It must be noted that for this solution to be
effective, it must be used in-conjunction with other
techniques such as the pseudonymizing of offline data
before corresponding records are created online as
outlined in Fig5. In addition, online data must be
anonymous so that if the records are leaked, no
identifying information would be part of the leaked
records. Furthermore, random IDs must be used for
the online system to ensure privacy of users is
maintained.
In Our next paper will detail the algorithms to be
used for the pseudonymization of PII as well as the
generation of Random electronic IDs for anonymous
use of electronic services such as ecommerce.
7 CONCLUSION
The experiment results show that it is possible to
protect PII from hackers by not presenting any
possibility of accessing the data regardless of the
security configurations of the systems holding that
data. The fact that no online user can reach the offline
system holding sensitive data makes the system more
secure. Enhanced protection comes in because no one
would be able to access the offline system from the
online system as the separation is physical. In
addition, even if someone breached the security of the
online system, they would need physical access to the
offline side of the data protector to configure it to
accept and allow transfer of data towards the offline
system. The Restricted amount of data that can be
sent via the data protector is a huge deterrent to
would-be data criminals as the time it would take
would render the exercise futile.
To make the proposed solution effective, it must
be implemented as recommended in Fig5 as well as
the detailed process flows in Fig6 and Fig8. The
implementation of the proposed solution in the
manner outlined would create a layered defence
mechanism to protect PII as well provide privacy to
the user. It would further, make it possible for
authorities to trace users who would commit fraud
online if need arises. The approach is good to the
good elements and bad to the bad elements.
REFERENCES
Aarthy, D. K., Aarthi, M., Farhath, K. A., Lakshana, S., &
Lavanya, V. (2017). Reputation-based trust
management in cloud using a trusted third party.
ICONSTEM 2017 - Proceedings: 3rd IEEE
International Conference on Science Technology,
Engineering and Management, 2018-Janua, 220–225.
https://doi.org/10.1109/ICONSTEM.2017.8261418
Ali, S. S., Chakraborty, R. S., Mukhopadhyay, D., &
Bhunia, S. (2011). Multi-level attacks: An emerging
security concern for cryptographic hardware.
Proceedings -Design, Automation and Test in Europe,
DATE, 1176–1179. https://doi.org/10.1109/date.2011.
5763307
Bao, Z., Wang, Q., Shi, W., Wang, L., Lei, H., & Chen, B.
(2020). When Blockchain Meets SGX: An Overview,
Challenges, and Open Issues. IEEE Access, 8, 170404–
170420. https://doi.org/10.1109/access.2020.3024254
Coppolino, L., D’Antonio, S., Mazzeo, G., & Romano, L.
(2019). A comprehensive survey of hardware-assisted
security: From the edge to the cloud. Internet of Things,
6, 100055. https://doi.org/10.1016/j.iot.2019.100055
Frank A Cona, M. D. P. (2019). Patent No. US 2019 /
0333054 A1. United States of America.
Hauer, B. (2015). Data and information leakage prevention
within the scope of information security. IEEE Access,
3, 2554–2565. https://doi.org/10.1109/ACCESS.2015.
2506185
II-Agure, Z., Belsam, A., & Yun-ke, C. (2019). The
Semantics of Anomalies in IoT Integrated BlockChain
Network. IEEE, 144–146.
Innab, N., & Alamri, A. (2018). The Impact of DDoS on E-
commerce. 21st Saudi Computer Society National
Computer Conference, NCC 2018, 1–4. https://doi.org/
10.1109/NCG.2018.8593125
Jamshiya, P. K., & Menon, D. M. (2018). Design of a
Trusted Third Party Key Exchange Protocol for Secure
Internet of Things (IoT). Proceedings of the
International Conference on Inventive Communication
and Computational Technologies, ICICCT 2018,
(Icicct), 1834–1838. https://doi.org/10.1109/ICICCT.
2018.8473281
Kangwa, M., Lubobya, C. S., & Phiri, J. (2021). Prevention
of Personally Identifiable Information Leakage in E-
commerce via Offline Data Minimisation and
Pseudonymisation. International Journal of Innovative
Science and Research Technology, 6(1), 209–212.
Retrieved from https://scholar.google.com/scholar?
hl=en&as_sdt=0%2C5&q=mukuka+kangwa&oq=
Locher, T., Obermeier, S., & Pignolet, Y. A. (2018). When
Can a Distributed Ledger Replace a Trusted Third
Party? 2018 IEEE International Conference on Internet
of Things (IThings) and IEEE Green Computing and
Communications (GreenCom) and IEEE Cyber,
Physical and Social Computing (CPSCom) and IEEE
Smart Data (SmartData), 1069–1077. https://doi.org/
10.1109/Cybermatics_2018.2018.00197
Pawar, H. R., & Harkut, D. G. (2018). Classical and
Quantum Cryptography for Image Encryption
Enhanced Protection of Ecommerce Users’ Personal Data and Privacy using the Trusted Third Party Model
125
Decryption. Proceedings of the 2018 3rd IEEE
International Conference on Research in Intelligent
and Computing in Engineering, RICE 2018, 1–4.
https://doi.org/10.1109/RICE.2018.8509035
Ribeiro, S. L., & Nakamura, E. T. (2019). Privacy
Protection with Pseudonymization and Anonymization
in a Health IoT System: Results from OCARIoT.
Proceedings - 2019 IEEE 19th International
Conference on Bioinformatics and Bioengineering,
BIBE 2019, 904–908. https://doi.org/10.1109/
BIBE.2019.00169
Štarchoň, P., & Pikulík, T. (2019). GDPR principles in data
protection encourage pseudonymization through most
popular and full-personalized devices - mobile phones.
Procedia Computer Science, 151(2018), 303–312.
https://doi.org/10.1016/j.procs.2019.04.043
Ye, F., Dong, X., Shen, J., Cao, Z., & Zhao, W. (2019). A
Verifiable dynamic multi-user searchable encryption
scheme without trusted third parties. Proceedings of the
International Conference on Parallel and Distributed
Systems - ICPADS, 2019-Decem, 896–900.
https://doi.org/10.1109/ICPADS47876.2019.00131
Zhan, J., Fan, X., Cai, L., Gao, Y., & Zhuang, J. (2018).
TPTVer: A trusted third party based trusted verifier for
multi-layered outsourced big data system in cloud
environment. China Communications, 15(2), 122–137.
https://doi.org/10.1109/CC.2018.8300277
Zhang, P., Alkubati, M., Bao, Y., & Yu, G. (2021).
Research advances on blockchain-as-a-service:
architectures, applications and challenges. Digital
Communications and Networks. https://doi.org/
10.1016/j.dcan.2021.02.001
ICE-B 2021 - 18th International Conference on e-Business
126