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

net/publication/313138956

PAYSIM: A FINANCIAL MOBILE MONEY SIMULATOR FOR FRAUD DETECTION

Conference Paper · September 2016

CITATIONS READS

3 5,695

3 authors, including:

Edgar Alonso Lopez-Rojas Stefan Axelsson

Simudyne Norwegian University of Science and Technology at Gjøvik, Norway
24 PUBLICATIONS   95 CITATIONS    61 PUBLICATIONS   2,107 CITATIONS   

SEE PROFILE SEE PROFILE

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

PaySim View project

RetSim View project

All content following this page was uploaded by Stefan Axelsson on 31 January 2017.

The user has requested enhancement of the downloaded file.

 PAYSIM: A FINANCIAL MOBILE MONEY SIMULATOR FOR FRAUD DETECTION

Edgar Alonso Lopez-Rojas(a) , Ahmad Elmir(b) , and Stefan Axelsson(c)

(a) ,(b)
Blekinge Institute of Technology , (c) The Norwegian University of Science and Technology
(a)
edgar.lopez@bth.se (b) ahel11@bth.se (c) stefan.axelsson@hig.no

ABSTRACT try. With the help of statistic analysis and social net-
The lack of legitimate datasets on mobile money transac- work analysis PaySim is able to generate congruous re-
tions to perform research on in the domain of fraud detec- sults with the original data set.
tion is a big problem today in the scientific community. The scope of this paper covers the design and construc-
Part of the problem is the intrinsic private nature of fi- tion of the simulator as well as the evaluation of the qual-
nancial transactions, that leads to no public available data ity of the data generated. The injection of malicious fraud
sets. behaviour and the application of different fraud detection
This will leave the researchers with the burden of first har- methods are outside the scope of this paper and are the
nessing the dataset before performing the actual research topics for further work with the PaySim simulator.
on it. This paper propose an approach to such a problem Outline This paper is structured as follows: Section 2.
that we named the PaySim simulator. presents the background and previous work in simulating
PaySim is a financial simulator that simulates mobile financial data. Section 3. states the problem and dur-
money transactions based on an original dataset. In this ing sections 4. and 5. we present the implementation
paper, we present a solution to ultimately yield the pos- of PaySim and the results of the simulations.Finally sec-
sibility to simulate mobile money transactions in such a tion 6. present the conclusions and future work.
way that they become similar to the original dataset. With
technology frameworks such as Agent-Based simulation 2. BACKGROUND AND PREVIOUS WORK
techniques, and the application of mathematical statistics,
In many parts of Africa the adoption of mobile money
we show in this paper that the simulated data can be as
as a means of sending & receiving funds have improved
prudent as the original dataset for research.
the life of merchants and customers alike. In Tanzania
Keywords: Multi-Agent Based Simulation, Financial for instance, which according to the world bank is one
data, Fraud Detection,Retail Fraud, Synthetic Data. of the fastest growing economies in the world, the adop-
tion of mobile money as a solution for creating payments
1. INTRODUCTION has induced a positive effect on the overall economy.
Obtaining access to data sets of mobile transactions for During December 2013 alone, 100 million transactions
research is a very hard task due to the intrinsic private na- were made in total netting a volume of $1.8 billion dol-
ture of such transactions. Scientists and researchers must lars (Seetharam and Johnson, 2015).
today spend time and effort in obtaining permits and ac- The domain of Mobile Money Transfer has grown sub-
cess to relevant data sets before they can research on such stantially in the last few years and have attracted greater
data set. This is time consuming and distracts researchers attention from users, specifically in areas in which bank-
from from focusing on the main problem which is per- ing solutions may not be as procurable as in developed
forming experiments on the data and finding novel ways countries. Many solutions have been employed in many
to solve problems such as the problem that inspired this places for this purpose. There are existing mobile money
paper which is the fraud detection on financial data. services in more than 10 African countries which cover-
The work presented in this paper provides a tool and age of 14% of all mobile subscribers (Rieke et al., 2013).
a method to generate synthetic data with the help of a The ever growing usage of mobile money has increased
simulator that we named PaySim. PaySim generates syn- the chances and likelihood of criminals to perform fraud-
thetic datasets similar to real datasets from mobile money ulent activities in an attempt to circumvent the security
transactions. This will be done by the means of computer measures of mobile money transfers services for personal
simulation, in particular, agent based simulation. Agent financial gain. There is therefore a great amount of pres-
based simulation is of great benefit in this context, this sure on researching the potential security pitfalls that can
partially because the models created represent with ac- be exploited with the ultimate goal to develop counter-
curacy the human behaviour during transactions and are solutions for the attacks.
flexible enough to easily be adapted to new constraints. Due to the large amount of transactions and the ever
PaySim simulates mobile money transactions based on changing characteristics on fraud, the current measures
a sample of real transactions extracted from the logs of a against fraud lack effectiveness. Many current system
mobile money service implemented in an African coun- still base their detection mechanism on simple thresholds
assigned arbitrarily. Therefore there is a need to push the aggregated financial information of payments during
forward and investigate the effect of fraud and stop the 6 months of the two main cities of Spain that was provided
wrongdoers from fraudulent profit. by a bank in Spain with the purpose of developing ap-
With PaySim, we aim to address this problem by pro- plications of different kinds that benefit from this sort of
viding a simulation tool and a method to generate syn- data. Our work differs from this work because the source
thetic datasets of mobile transactions. The benefits of us- of the data and the characteristics of bank payments and
ing a simulator to address fraud detection was first pre- mobile transactions are different as presented later in the
sented by (Lopez-Rojas and Axelsson, 2012b). This re- following sections.
search states the problem of obtaining access to financial The key common aspect on previous work is the use
datasets and propose using synthetic datasets based on of the paradigm of ”Multi Agent Based Simulation”
simulations. The method proposed is based on the con- approach which incorporates into the behaviour of the
cept of MABS (Multi Agent Based Simulation). MABS agents the main customer logic to reach similar results
has the benefits that allows the agents to incorporate simi- as the real world. It is important to recognize that a sim-
lar financial behaviour to the one present in domains such ulation is not an actual ”replication” of the original data
as bank transactions and mobile payments. set. Rather, a simulation will with the aid of statistical
The first implementation of a simulator for financial methods generate a very similar data set of the original
transaction was introduced by (Lopez-Rojas and Axels- data set. The degree in variance will largely be depen-
son, 2012a) with a mobile money transactions simulator. dent on how the data on the original data set is structured,
This simulator was implemented due to the difficulties to hence, different simulations based on different seeds will
implement a proper fraud detection control on a mobile generate different output data sets but consistent with the
money system that was under development. This paper real world.
was the first to present an alternative to the lack of real
data problem. The synthetic dataset generated by the sim- 3. PROBLEM
ulator was used to test the performance of different ma- The problem formulation for this research paper tackles
chine learning algorithms in finding patterns of money the issue of whether the generation of synthetic financial
laundering. data is sufficient to supersede real financial data whilst
The work by (Gaber et al., 2013) introduced another simultaneously yield commensurable results if the syn-
similar technique to generate synthetic logs for fraud de- thetic data is used as the source data set for any research.
tection. The main difference here was that this time there This is of primary concern for any researcher that wish to
was available real data to calibrate the results and com- perform scientific tests but does not or have limited ac-
pare the quality of the result of the simulator. The purpose cess to a real financial data set.
of this study was to generate testing data that researchers The main focus and goal for the simulation is to yield
can use to evaluate different approaches. This works dif- another completely self-sufficient data set with the goal
ferers significantly from our work because we present a of having similar statistical properties as the original data
different method for analysing the data place special at- set. To yield such results, the simulator must go through
tention on evaluating the quality of the resultant synthetic several steps to be able to complete.
data set. In order to simulate the mobile money service, we
There has been some work done in the domain of fi- need to properly simulate the different kind of transac-
nancial transactions for retail stores. The most prominent tions that the system supports. We decided to cover 5 of
of which is the work done by (Lopez-Rojas et al., 2013). the most important transaction types: CASH-IN, CASH-
The work done in that paper is very similar to the work OUT, DEBIT, PAYMENT and TRANSFER.
done in this paper. A large collection of data was gath- CASH-IN is the process of increasing the balance of
ered from Sweden’s biggest shoe-retailer, and techniques account by paying in cash to a merchant.
involved complex machine-learning algorithms in an at- CASH-OUT is the opposite process of CASH-IN, it
tempt to find fraudulent behaviour in clients. The paper means to withdraw cash from a merchant which decreases
showed among other things results from Social Network the balance of the account.
which described the relationship between the clients and DEBIT is similar process than CASH-OUT and in-
the sellers for each store. A definition of what was per- volves sending the money from the mobile money service
ceived as ”fraudulent” was made and based on that the to a bank account.
machine-learning algorithms were trained to detect that PAYMENT is the process of paying for goods or ser-
type of behaviour. vices to merchants which decreases the balance of the ac-
Public databases of financial transactions are almost count and increases the balance of the receiver.
non existent. However the work of (Lopez-Rojas and Ax- TRANSFER is the process of sending money to an-
elsson, 2014) during the implementation of a simulator other user of the service through the mobile money plat-
called BankSim presents a MABS of financial payments. form.
BankSim is implemented in a similar way as the RetSim There are other types of transactions that we decided
simulator and our simulator using in addition to statistical to exclude from the simulation due to the low percentage
analysis a social network analysis. BankSim is based on of data found in the sample.
4. MODEL AND IMPLEMENTATION between agents since there is a probability that at a partic-
PaySim uses the MABS toolkit called MASON version ular step of the simulation, an agent might transfer money
19 which is implemented in Java (Luke, 2005). We se- to another agent and thus alter its and the other agents
lected MASON because it is: multi-platform, supports state.
parallelisation, and fast execution speed in comparison
with other agent frameworks. This is especially impor- 4.2. Inputs
tant for multiple running and computationally expensive There are multiple inputs required in order for the simu-
simulations such as PaySim (Railsback et al., 2006). lator to function smoothly. As initial input, the number of
clients neighbours for each agent is assigned. The profile
4.1. Overview,Design and Details (ODD) for each agent is then further attached based on a proba-
The design of PaySim was based on the ODD model in- bility. Their location on the spatial space along with their
troduced by (Grimm et al., 2006). ODD contains 3 main neighbours is also initialized.
parts: Overview, Design Concepts and Details.
• Parameter File This is the file that contains all of
4.1.1. ODD Overview the needed parameters that the simulator needs to
The purpose of this simulator is to simulate payments initiate. Among these parameters we find the seed
done in the realms of mobile transactions. The simula- and perhaps the most relevant of which is the paths
tor should ultimately perform simulations in such a way for where the input files and the output files are
that synthetic data in regards to mobile transactions can be placed on the current machine.
generated. The simulator should generate synthetic data
• Aggregated Transaction File This file contains the
that is very similar to a batch of real transactional Data
distribution of the transactions from the original data
provided by Ericsson. The goal is to have a generator
set. More precisely, it contains how many transac-
that can generate data on the fly that can later be used by
tions were made at any given day/hour combination
the scientific community in an attempt to research more
(step). what is the average price for that, what type
about fraud detection.
of transaction it was etc. This is of paramount impor-
The model has one primary type of Entity which is
tance for the simulator since statistical data is gener-
Client. Each client has a profile that describes the al-
ated from the information gathered from this file.
lowed behaviour for the client such as the limit on trans-
actions daily/yearly, the transaction limit and the maxi- • Repetitions File This file contains the frequency of
mum balance for the client. Furthermore the number of transactions that the original clients had per type of
transactions, withdrawals, transfers and deposits is stored transaction. This means that some of the agents are
for each client. The client can further be classified by schedule more than others based on a social network
age to be young, adult or senior. Each client has a base analysis of the indegree and outdegree of the cus-
currency in which the transactions are based upon. The tomers.
client can perform transactions in the form of deposits,
withdrawals and transfers. For every transaction that is Since the simulator is using MASON as the framework
made, it is stored and saved within the client. for performing the simulation, it is of paramount impor-
tance to define how each step is to be regarded. For this
4.1.2. Process Overview and Scheduling simulation we defined that each day/hour combination
The client has several processes that alter their inter- represents one step. At each step, a Client that represents
nal states. For each step that is made by the simulator, the agent for the simulator is generated. The client will
based on a random variable that is contingent on calcu- be placed in an environment in which it is to make deci-
lated probabilities, a type of transaction that is to be per- sions based on the information it perceives. The Client is
formed by the client is chosen. A deposit transaction will created with the statistical distribution of the possibilities
increase the balance of the client, a withdrawal will de- to perform each transaction type for a specific day/hour
crease the balance of the client and a transfer transaction combination. The client then randomly perform (based
will withdraw money from the original client and then de- on the distribution initiated) different transaction types in
posit them to the destination client in question. relation to the other clients on the simulator. Also, for
The concepts that are behind the model are based on each client generated, there is a probability P for the client
statistical analysis of a large batch of real data. From this to make future transactions at later steps. This probability
batch of data, probabilities of each action were calculated is gathered from the database of the original data set.
and incorporated into the model to generate synthetic in-
formation as close as possible to the real data. The client 4.3. Initiation Stage
agent has some adaptive behaviours that will alter their In this stage, the PaySim simulator must load the neces-
way of acting; for instance if the client has reached its sary data needed from the original dataset:
daily limit it cannot withdraw money any more for that
day. This adaptive behaviour is a direct result of the • Load The Parameters The first and most important
transfer process mentioned above. There is interaction step is to load the values for each parameter in the
 parameter file. These will among other things con- • Statistical Distribution This is the different proba-
tain the file paths for the source data inputs that the bilities that the client will have loaded into it which
simulator needs to load. entails the probability P of performing each action.

• Load Aggregate File This is the original Aggregate • Initial Balance This will be the initial balance that
File that will be used as a base point for the simu- the client will have once generated.
lator to generate statistically similar results in terms
of ”What to simulate, at what day, at which amount” 4.4.2. Performing the transactions
etc. One such extraction could be for instance: At After each client is generated, the client will make the
Day 1 and hour 15, simulate 8703 transactions of the decision of what type of transaction it will ultimately
type PAYMENT with the average transaction size of make, again this is completely derived from the distribu-
180000 and the standard deviation of 15000. tion loaded. The client is in an environment which allows
it to freely interact with other clients in the simulation.
• Load Initial Balance Container Apart from the sta- There are some types of transaction types that is based
tistical distribution for each transaction type input to on that, like ”TRANSFER” for instance. The ”TRANS-
the client, there is another important input, namely, FER” type is exchange of money from one client to an-
the ”Initial Balance” of the client. Upon the gen- other; hence, the client will have to interact with other
eration of each client in the simulation, there must clients to simulate the actual exchange of funds.
be an initial ”Balance” attached to that client. This
balance is generated with the ”Balance Container” 4.5. Finalization Stage
file as base point. The Balance Container consists After each of the clients have completed their role in the
of the different probabilities that will generate dif- simulation and performed all of the transactions allotted
ferent initial balance ranges. the results must be saved. There are 4 outputs generated
• Load Maximum Repetitions File As mentioned for each simulation made. All of which serve a specific
previously, each client has a probability P of making purpose which will allow for the exact repetition of the
future transactions in future steps in the simulation. simulation with the exact initial properties.
What this file does is to make sure that each client
• Logfile Each transaction that is made will contain
does not make more repetitions that is allowed. Like
a record with the meta-data for that transaction.
the Balance Container, the probabilities in this file
Data such as what client performed which action, to
is also yielded from the database of the original data
which other client, the sum of the transaction, and
set.
the delta in balance for all clients involved. Each
4.4. Execution Stage such record will be saved in a logfile unique for the
specific simulation.
Upon completion of the Initiation Stage and all of the pa-
rameters are successfully loaded, the simulator can now • MySQL Database Apart from the logfile, the record
proceed to the execution stage. It is at this stage that the for each transaction will also be saved into a MySQL
simulator will perform the actual simulation, and yield the database. The purpose of which is to allow for eas-
simulated transaction results: ier queries when the analysis of the resulsts is to be
made.
4.4.1. Generating The Clients
The agents are the founding blocks of the ”Agent Based • Aggregate Dump An aggregatedump that is similar
Simulator”. The agent in this context, resembles the to the original aggregatedump from the original data
”Client”. Upon each step of the simulation, the PaySim set will also be generated. It is these two files that
simulator will convert each step to a ”Day/hour” combi- will be used to generate the plots and graphs resem-
nation. This will then be used as an input to extract the bling the results of the transactions.
statistical distributions from the original data set. Based
on the Aggregated Transaction File, PaySim harness the • Parameter File History This file will contain the
probability P of performing each each transaction in the exact properties needed for the simulation to be able
simulator and save it into the model of the client. With to reproduce the exact same results again. This is
this information, the client now has gained more knowl- important because each simulator must be able to be
edge and will know the following important things: reproduced again, and without the original ”seed”
used, it will not be possible.
• Number Of Transactions This is the total number
of transactions that this generated client will do.
5. RESULTS
• Make Future Steps This is the information of We ran PaySim several times using random seeds for 744
whether the client is to participate in future steps. steps, representing one month of real time data. Each run
Which means scheduling the tasks of performing took around 30 minutes in a i7 intel processor. We se-
more transactions during further steps. lected a datasets that contained the lowest difference in
 Figure 1: Visualization of transaction type CASH-IN Figure 3: Visualization of transaction type TRANSFER
CASH_IN TRANSFER
20000
COUNT

COUNT
Orig Orig

6000
41840 41840
0

0
0 100 200 300 400 500 600 700 0 100 200 300 400 500 600 700

STEP STEP

4e+09
Orig Orig
SUM

SUM
41840 41840
0.0e+00

0e+00
0 100 200 300 400 500 600 700 0 100 200 300 400 500 600 700

STEP STEP
4e+05

5e+05
Orig Orig
AVG

AVG
41840 41840
0e+00

0e+00
0 100 200 300 400 500 600 700 0 100 200 300 400 500 600 700

STEP STEP

0.0e+00 2.5e+07
0e+00 5e+05

Orig Orig
STD

STD

41840 41840

0 100 200 300 400 500 600 700 0 100 200 300 400 500 600 700

STEP STEP

Figure 2: Visualization of transaction type CASH-OUT Figure 4: Visualization of transaction type PAYMENT
CASH_OUT PAYMENT
30000
40000
COUNT

COUNT

Orig Orig
41840 41840
0

0 100 200 300 400 500 600 700 0 100 200 300 400 500 600 700

STEP STEP
1e+08
4e+09

Orig Orig
SUM

SUM

41840 41840
0e+00

0e+00

0 100 200 300 400 500 600 700 0 100 200 300 400 500 600 700

STEP STEP
5e+05

4000

Orig Orig
AVG

AVG

41840 41840
1000
0e+00

0 100 200 300 400 500 600 700 0 100 200 300 400 500 600 700

STEP STEP
8e+05

60000

Orig Orig
STD

STD

41840 41840
0e+00

0 100 200 300 400 500 600 700 0 100 200 300 400 500 600 700

STEP STEP
 due to the introduction of income during the first days of
Figure 5: Visualization of transaction type DEBIT
the month.
DEBIT

6. CONCLUSIONS
COUNT

Orig
1000

41840

PaySim is a simulation of mobile money transactions with

0

0 100 200 300 400 500 600 700

the objective to generate a synthetic transactional data set
STEP
that can be used for research into fraud detection. The
data sets generated with PaySim can aid academia, finan-
8e+06

Orig cial organisations and governmental agencies to test their

SUM

41840

fraud detection methods or to compare the performance of

0e+00

0 100 200 300 400 500 600 700

different methods under similar conditions using a com-
STEP
mon public available and standard synthetic data set for
the test.
Orig We argue that PaySim is ready to be use as a tool to
0 20000
AVG

41840

generate synthetic transactions that resemble the origi-

0 100 200 300 400 500 600 700 nal and private data set supplied. By using PaySim we
STEP
protect the privacy of the customers of the service at the
same time that interesting results are possible to share
with other researchers without the constrains and legal
200000

Orig
STD

41840

boundaries of the original data.

0

0 100 200 300 400 500 600 700 The results presented in the section 5. help to visually
STEP
appreciate that the generated dataset captures the process
and the frequencies of the different transaction types of
the mobile money service.
Future work on the simulator is to add to the model
values according to the original data set provided. The fraudulent agents and run different scenarios to test the
selected synthetic dataset was arbitrary named PS41840. efficacy and accuracy of diverse fraud detection meth-
PS41840 contains around 23 million records divided into ods. We also want to make a synthetic data set available
the 5 types of categories presented before. Table 1 shows to other researchers and be able to compare and share di-
the types of transactions, count and average amount gen- verse results.
erated with the simulator. The amount values are given
in a currency that we can not disclose. ACKNOWLEDGMENTS
This work is part of the research project ”Scalable
resource-efficient systems for big data analytics” funded
Table 1: Simulated PS41840 by the Knowledge Foundation (grant: 20140032) in Swe-
TYPE Count avgAmount den.
CASH-IN 4 496 947 153 019
CASH-OUT 9 014 407 155 989
TRANSFER 2 030 969 630 810 References
PAYMENT 8 955 794 10 793 Chrystel Gaber, Baptiste Hemery, Mohammed Achem-
DEBIT 139 935 5 016
lal, Marc Pasquet, and Pascal Urien. Synthetic logs
generator for fraud detection in mobile transfer ser-
vices. In 2013 International Conference on Collab-
The evaluation of the quality of the database was first
oration Technologies and Systems (CTS), pages 174–
calculated using the sum of square error (SSE) method on
179. IEEE, may 2013. ISBN 978-1-4673-6404-1. doi:
the quantities of the different datasets. The one with the
10.1109/CTS.2013.6567225.
lowest error was PS41840.
In order to verify that the simulation was working prop- Volker Grimm, Uta Berger, Finn Bastiansen, Sigrunn
erly we plotted the distributions to visually identify sig- Eliassen, Vincent Ginot, Jarl Giske, John Goss-
nificant differences between the original and the synthetic Custard, Tamara Grand, Simone K. Heinz, Geir
dataset. Figures 1, 2, 3, 4 and 5 show the visualization Huse, Andreas Huth, Jane U. Jepsen, Christian
per type of transaction. Each figure contains the output Jø rgensen, Wolf M. Mooij, Birgit Müller, Guy Pe’er,
for each step regarding the count of transactions, the total Cyril Piou, Steven F. Railsback, Andrew M. Rob-
sum of transaction, the average and the standard devia- bins, Martha M. Robbins, Eva Rossmanith, Nadja
tion. The red continuous line represent the original data Rüger, Espen Strand, Sami Souissi, Richard a.
distribution and the blue dashed line represent the syn- Stillman, Rune Vabø, Ute Visser, and Donald L.
thetic dataset PS41840. DeAngelis. A standard protocol for describing
Something we noted is that the first 14 days of the sim- individual-based and agent-based models. Ecolog-
ulation the activity in the system is higher compared to the ical Modelling, 198(1-2):115–126, September 2006.
remaining days. This is perhaps a phenomenon present ISSN 03043800. doi: 10.1016/j.ecolmodel.2006.04.
 023. URL http://linkinghub.elsevier.com/ Computer Science from EAFIT University in Colombia
retrieve/pii/S0304380006002043. (2004). After that he worked for 5 more years at EAFIT
University as a System Analysis and Developer and par-
Edgar Lopez-Rojas and Stefan Axelsson. Multi agent tially as a lecturer. He obtained a Masters degree in Com-
based simulation (mabs) of financial transactions for puter Science from Linköping University in Sweden in
anti money laundering (aml). In Audun Josang and 2011 and a licentiate degree in computer science (a de-
Bengt Carlsson, editors, Nordic Conference on Secure gree halfway between a Master’s degree and a PhD) in
IT Systems, pages 25–32, Karlskrona, 2012a. 2014.
Edgar Alonso Lopez-Rojas and Stefan Axelsson. Money
MSc. Ahmad Elmir
Laundering Detection using Synthetic Data. In Julien
Ahmad obtained a master in computer science with spe-
Karlsson, Lars ; Bidot, editor, The 27th workshop of
ciality in security from the Blekinge Institute of Technol-
(SAIS), pages 33–40, Örebro, 2012b. Linköping Uni-
ogy. His master’s thesis was about the design and con-
versity Electronic Press.
struction of PaySim under the supervision of the main
Edgar Alonso Lopez-Rojas and Stefan Axelsson. So- author of this paper. Previously he have studied natural
cial Simulation of Commercial and Financial Be- sciences in the gymnasium for three years. His speciality
haviour for Fraud Detection Research. In Advances in was at computer science and programming. He has a keen
Computational Social Science and Social Simulation, interest for scientific inquiry in the domain of security as
Barcelona, 2014. ISBN 9789172952782. it is in his opinion an ever developing field. He have also
worked for 9 months with software development in a cor-
Edgar Alonso Lopez-Rojas, Stefan Axelsson, and Dan poration.
Gorton. RetSim: A Shoe Store Agent-Based Simula-
tion for Fraud Detection. In The 25th European Mod- Dr. Stefan Axelsson
eling and Simulation Symposium, number c, page 10, Stefan Axelsson is a senior lecturer at NTNU - Norwegian
Athens, Greece, 2013. University of Science and Technology in Norway. He re-
ceived his M.Sc in computer science and engineering in
S. Luke. MASON: A Multiagent Simulation Environ-
1993, and his Ph.D. in computer science in 2005, both
ment. Simulation, 81(7):517–527, July 2005. ISSN
from Chalmers University of Technology, in Gothenburg,
0037-5497. doi: 10.1177/0037549705058073. URL
Sweden. His research interests revolve around computer
http://sim.sagepub.com/cgi/doi/10.1177/
security, especially the detection of anomalous behaviour
0037549705058073.
in computer networks, financial transactions and ship/-
S. F. Railsback, S. L. Lytinen, and S. K. Jackson. Agent- cargo movements to name a few. He is also interested in
based Simulation Platforms: Review and Develop- how to combine the application of machine learning and
ment Recommendations. Simulation, 82(9):609–623, information visualization to better aid the operator in un-
September 2006. ISSN 0037-5497. doi: 10.1177/ derstanding how the system classifies a certain behaviour
0037549706073695. URL http://sim.sagepub. as anomalous. Stefan has ten years of industry experi-
com/cgi/doi/10.1177/0037549706073695. ence, most of it working with systems security issues at
Ericsson.
Roland Rieke, Maria Zhdanova, Jurgen Repp, Romain
Giot, and Chrystel Gaber. Fraud Detection in Mo-
bile Payments Utilizing Process Behavior Analysis. In
2013 International Conference on Availability, Relia-
bility and Security, pages 662–669. IEEE, sep 2013.
ISBN 978-0-7695-5008-4. doi: 10.1109/ARES.2013.
87.

Balachandran Seetharam and Drew Johnson. Mobile

Money’s Impact on Tanzanian Agriculture. 2015.

AUTHORS BIOGRAPHY
MSc. Edgar A. Lopez-Rojas
Edgar Lopez is a PhD student in Computer Science at
Blekinge Institue of Technology in Sweden and his re-
search areas are Multi-Agent Based Simulation, Ma-
chine Learning techniques with applied Visualization for
fraud detection and Anti Money Laundering (AML) in
the domains of retail stores, payment systems and fi-
nancial transactions. He obtained a Bachelors degree in

View publication stats