SINERGI Vol.
No.
June 2022: 145-154 http://publikasi.
id/index.
php/sinergi http://doi.
org/10.
22441/sinergi.
The process mining method approach to analyze usersAo behavior of internet in the Local Area Network of Sriwijaya University Muhammad Irfan Jambak1.
Amrifan S.
Mohruni2.
Muhammad Ihsan Jambak3*.
Endy Suherman4 Department of Electrical Engineering.
Faculty of Engineering.
Universitas Sriwijaya.
Indonesia Department of Mechanical Engineering.
Faculty of Engineering.
Universitas Sriwijaya.
Indonesia Department of Informatics Management.
Faculty of Computer Science.
Universitas Sriwijaya.
Indonesia Information & Communication Technology Technical Implementation Unit.
Universitas Sriwijaya.
Indonesia Abstract The Sriwijaya University internet network management unit does not yet have a standard formulation for implementing Bandwidth Management & Bandwidth Allocation.
To provide the best service, they apply the Best-Effort Service concept.
As a result, it requires a relatively large network capacity and bandwidth provision so that it has an impact on costs.
Therefore, it is necessary to know how users use internet bandwidth as the basic principle of Bandwidth Management & Bandwidth Allocation.
This study has completed how to determine the behavior of internet users on the campus LAN as a reason to evaluate internet bandwidth usage.
With the Process Mining method, process mapping has been carried out for all access to internet usage from all faculties.
As a result, the factors that characterize and need to be considered in bandwidth management are obtained.
In order of significance are Total Access Length.
Average Variance.
Number of User Case IDs.
Number of NonAcademic Ports.
Number of Academic Ports.
Number of Access Frequency.
Number of Events, and Number of Ports of Service.
Keywords:
Bandwidth Management.
Business Process.
Internet.
Process Mining.
Article History:
Received: June 26, 2021 Revised: October 22, 2021 Accepted: October 29, 2021 Published: June 2, 2022 Corresponding Author:
Muhammad Ihsan Jambak Department of Informatics Management.
Faculty of Computer Science.
Universitas Sriwijaya Email: jambak@unsri.
This is an open access article under the CC BY-NC license INTRODUCTION Quality of Service (QoS) of the internet network is determined mainly by network management, defined as the various activities, methods, and knowledge required to manage computer networks.
QoS parameters consist of Bandwidth.
Throughput.
Latency/Delay.
Jitter, and Packet Loss .
, 2, 3, .
QoS is also a term used to define the ability of a network to provide different levels of service assurance.
Or it can be said that QoS is a network mechanism that allows services to operate according to the needs expected by users.
Thus.
QoS can make bandwidth, latency, and jitter predictable and tailored to the needs of the network users.
There are three levels of QoS that can be used:
Best-Effort service: makes all efforts to deliver a packet to a destination.
Integrated service: provides applications with a level of service assurance that parameters have been negotiated end-to-end.
Differentiated service: provides a set of classification tools and a queuing mechanism for protocols with specific priorities on different networks.
Bandwidth is the maximum amount of data transferred from one point to another within a certain period, so the unit is bits per second .
Bandwidth management on the network is essential because bandwidth on a computer network is valuable.
Several bandwidth management methods include Traffic Shaping (Rate Limitin.
Scheduling Algorithms.
Congestion Avoidance.
Bandwidth Reservation Protocols, and Traffic Classification.
Jambak et al.
The process mining method approach to analyze usersAo behavior of A SINERGI Vol.
No.
June 2022: 145-154 Traffic Shaping or bandwidth limitation based on data flow or data stream by increasing or decreasing the priority of the flowing packets is the easiest and most frequently used way.
However, the 3rd level of QoS is Differentiated Service, so management should be the most effective and efficient QoS level.
However, this service requires accurate and comprehensive data and information about the needs and behavior of network users, especially on bandwidth parameters due to the application of Traffic Classification techniques.
Therefore.
Bandwidth Management becomes essential because it will impact computer network Sriwijaya University (Unsr.
also has an internet network in the form of a Local Area Network (LAN) that serves all the needs of the academic community, which are students, lecturers, and employees, including the increasing demand for high-speed computer laboratory systems.
Overall, the total bandwidth capacity of Unsri for both the Bukit Besar and Indralaya campuses is 5.
5Gbps.
Therefore, the availability of internet access services at Unsri should be utilized well by the academic The distribution of bandwidth in Unsri, which has been done so far, is by way of BestEffort-Service, where users can access in the form of sending and receiving extensive data at any time without the need to ask permission from the network manager.
However, this has the consequence that it requires a relatively large network capacity and bandwidth provision, thus impacting the cost.
In this case, the Unsri campus internet network manager, the Information & Communication Technology Technical Implementation Unit (UPT TIK), does not have a standard formulation for implementing Bandwidth Management & Bandwidth Allocation.
It is necessary to know how the users utilize internet bandwidth usage from UnsriAos campus LAN to fulfil principles of bandwidth management & bandwidth allocation.
Moreover, this utilization provides an overview of how users use internet network facilities in the Unsri LAN.
Thus, the formulation of the problem that is solved in this research is: How to know the behavior of internet users on the Unsri LAN as rational to evaluate that internet bandwidth usage is effective and efficient? METHOD Regardless of the type of business operation, business is dynamic and must continue to evolve to keep up with the changing needs of the business environment to remain relevant.
One of the attributes that do a great business is efficient business processes.
Business processes define the tasks that must be performed to achieve predetermined business goals and objectives, such as producing products and providing services to end-users.
Therefore, streamlining business processes is essential to make them more adaptable to the changing needs of the business environment, gain a competitive advantage, achieve operational excellence and improve customer experience.
Business Process Management (BPM) is a discipline in operations management in which various methods are used to discover, model, analyze, measure, improve, optimize, and automate business processes .
, 6, 7, .
Any combination of methods used to manage a companyAos business processes is BPM .
Operations can be structured and iterative or unstructured and variable.
BPM looks at how the business processes in an organization can be assessed and improved, and it involves continuous evaluation of the process with actions being taken to optimize it.
BPM itself, two primary areas consist of Business Process Improvement (BPI) and Business Process Modeling (BP Modelin.
Both of them are aim to create an organization to look again at how organizational elements .
uch as people, processes, and technolog.
are aligned.
optimize core business processes.
By mapping and analyzing business processes, they can be redesigned, improved, and better managed.
As an approach.
BPM sees processes as an essential asset of an organization that must be understood, managed, and developed to announce and deliver value-added products and services to clients or customers.
This approach is very similar to other total quality management or continuous improvement process methodologies.
In addition, the BPM approach can be supported or activated through technology .
, 11, .
Because of this.
BPM was often discussed from one of two perspectives: people and technology.
Business process modeling allows an organization to confirm the current state of its processes and provide opportunities to improve them, then make it more effective and efficient.
With every detail of the process mapped, showing end-to-end activities, it will be easier to align the process with the goals and values of the In addition to process improvement, organizations model business processes, among others: complies with regulatory bodies.
assists in employee orientation/training.
and facilitates internal audit.
Jambak et al.
The process mining method approach to analyze usersAo behavior of A p-ISSN: 1410-2331 e-ISSN: 2460-1217 Many businesses have adopted modeling business processes for improvement.
Business Process Modeling allows organizations to graphically document their business processes, including business activities, events, flow control, stakeholders, and relations .
While a business process, in general, is a combination of operational steps and management control that, together, produce a product or provide a service.
BP Modeling involves describing the current state of business processes (Auas-isA.
as well as analyzing and improving them .
rocess Auas-isA.
to create more efficient business processes (Auto-beA.
In addition, diagrams are used for better visualization to facilitate easier understanding to capture current business processes.
Although many organizations have used diagrams such as flowcharts, data flow diagrams, , to describe their processes, it is not a model.
A model is an abstraction that contains all the elements needed to clarify the intent of the process being modeled.
In contrast, a diagram is a specific view of what we try to understand in a particular context .
Behaviors are actions that are always used or performed.
The process is the action stages of thus, modeling a process will reflect the For example, one way to find out the behavior of internet users on the Unsri LAN is by knowing the processes carried out by these internet users.
Then these processes are modeled to provide a complete picture of how these activities take place.
Process Mining is a technical group that supports business process analysis based on event logs.
Using specific algorithms applied to the event log data can identify trends, patterns, and details in the event logs recorded by the information system.
Process Mining aims to increase process efficiency and understand processes .
, 17, .
The term Process Mining comes from the data mining field.
The concept is AuminingAy data for understanding, answering questions, or solving problems.
The search is usually specific to identified challenges or obstacles in data mining.
Data mining has some similarities, particularly in analyzing big data for business decision support.
Process Mining uses specific algorithms against event log data to identify trends, patterns, and details of how the whole process is going.
Process Mining is an analytical discipline for discovering, monitoring, and improving actual processes by extracting knowledge from event log data available in todayAos information systems.
Process Mining offers an objective, fact-based understanding of the existing event logs, which helps audit, analyze and improve business processes by answering questions related to compliance and performance.
There are many ways to describe Process Mining: direct, comprehensive, visual, objective.
The focus is usually on extracting data from transparency about how the business operates.
With the insights gained from Artificial Intelligence and Machine Learning, organizations can then identify opportunities for process optimization.
Process Mining bridges the gap between traditional model-based process analysis and data-centric analysis techniques such as Machine Learning and Data Mining .
, 19, 20, .
Process Mining technology isnAot just about seeing where things can be improved.
Figure 1 explains that, which generates a process model from event log data with no additional, nonempirical input, there are two other types of Process Mining: Conformance and Enhancement .
In Conformance, an existing process model is compared with an event log of the same process to check for alignment.
In other words, confirming whether the process carried out follows the process model.
While Enhancement will improve or refine the model using data derived from the event log.
Rather than disclosing or comparing process operations, improvement aims to modify or enhance existing models.
We tend to think of models as ideal scenarios of how something But in the context of Process Mining, the final model should be viewed less as a fixed state and more like a map.
Its purpose is to guide the user to a destination using the best route, knowing that things will change over time.
So.
Process Mining technology uses a model with empirical data derived from event logs in organizational systems .
The research unit used was the Unsri Local Area Network Management System in this study.
The research stages are as shown in Figure 2 with six steps .
, namely:
Planning.
at this stage, the object data events log is planned to be downloaded by determining the school year and semester.
Data Extraction.
is the process of downloading data from the server and preprocessing.
Data Processing.
is a data processing algorithm selected by the Process Mining Mining & Analysis.
is the stage to produce a research output in the form of a Process Model.
Evaluation.
the stage of diagnosis and review of the Process Model generated in the previous Process Improvement.
the stage of compiling follow-up suggestions.
Jambak et al.
The process mining method approach to analyze usersAo behavior of A SINERGI Vol.
No.
June 2022: 145-154 Figure 1.
Process Mining: Discovery.
Conformance, & Enhancement .
Start Planning Data Format Data Extraction Data Even Logs Data Processing Mining & Analysis Business Process Models Evaluation Process Improvement & Support Stop Figure 2.
Stages of Research Jambak et al.
The process mining method approach to analyze usersAo behavior of A p-ISSN: 1410-2331 e-ISSN: 2460-1217 In this study, the even log data were extracted from records on the Sangfor firewall server in the Network Operating Control room, recording from 1-7 October 2020, of all internet access users per faculty at the Unsri campus Bukit Besar.
Palembang.
Each user access even is logged and tracked with 12 attributes consisting of Date and Time.
Services/Applications used.
Protocol.
Service Zone.
Group.
source IP.
Source Port.
Destination Zone.
IP destination.
Destination Policy Name.
and Description.
Thus, it can be known where and what the purpose of each access is.
the obstacles faced are the large data size and the limited number of rows of data for recording downloads in the Microsoft Excel For this reason, each faculty is recorded separately.
RESULTS AND DISCUSSION
Following the process mining method using Fluxion Disco software, the event log data is then processed to produce a model that shows how the processes of using internet bandwidth all take However, because the data used is an event log that traces all activities carried out from the beginning of access to the end, the resulting model is a complex Auspaghetti mapAy that is too complicated to interpret and use.
Fluxion Disco software miner is based on ChristianAos Fuzzy Miner, which is the first mining algorithm to introduce a Aumap metaphor,Ay highlighting frequent activities and paths through color and thickness .
The complete picture of the process is shown, the same as when it This is very important to understand the reference point of the process map because it shows a one-to-one match of data.
However, if no simplification strategy is applied, the entire process is usually too complicated to view in 100% detail.
To get more manageable pieces, we need to break this down and simplify the process map.
The resulting model as the output of data processing in the Fluxion Disco software for each faculty is shown in Figure The numerical data derived from the process maps are shown in Table 1.
Each activity on the process map is accompanied by information about how often the process is repeated .
as well as the length of access time.
Thus, it can be clearly seen which processes or activities dominate bandwidth The Faculty of Engineering has the highest number of users and access by defeating 22.
of the total.
Followed by the Faculty of Medicine 7%, the Postgraduate Program 14.
3%, and the Faculty of Computers Science 10.
PearsonAos statistical Test with a coefficient of 0.
905 and a significant value of 0.
001 shows that there is a positive correlation between the number of users and the number of accesses, except for an anomaly in the Faculty of Economics which has a large number of users.
Still, the number of accesses is relatively low.
Meanwhile, the Faculty of Medicine became the internet access user with the longest usage time with a total time of 320,076,300 seconds or an average of 159,639.
052 per user.
The Faculty of Engineering follow them with a total time of 301,956,120 seconds .
verage 104,338.
and the Faculty of Computers with a total time of 215,272,809 seconds .
verage 157,709.
Therefore, frequency of use and length of access time will be the initial consideration factors in bandwidth allocation management.
Comparison of the average length of access time provides a proportional comparison in analyzing the level of bandwidth utilization within Unsri campus.
The number of Event data is recorded based on the destination port accessed by each The number of events shows where and what site the user is going to.
One event means one destination, which is processed by one-time internet usage access by the user.
So, the total number of events is the number of destination ports accessed.
Fluxion Disco software also provides Variants, which shows us the extent to which we tend to underestimate the complexity of the However, we cannot see how individual cases move through the process or how many cases pass through this additional loop once, twice, or even more frequently.
In order to grasp the typical process execution pattern from beginning to end, we need to look at the variants.
Process variants are about changes in a process Process variants are the only path from the beginning to the end of the process.
In other words, a process variant is a specific sequence of activities, just like AutrackingAy from the beginning to the end of the process.
Because variant is how many ways or paths are accessed to reach the destination port, a high or low number of variants will have an impact on internet bandwidth usage.
Jambak et al.
The process mining method approach to analyze usersAo behavior of A SINERGI Vol.
No.
June 2022: 145-154 Process Map of the Faculty of Economics Process Map of the Faculty of Law Process Map of the Faculty of Engineering Process Map of the Faculty of Medicine Process Map of the Faculty of Agriculture Process Map of the Faculty of Teacher Training & Education Process Map of the Faculty of Social & Political Sciences Process Map of the Faculty of Computer Science Process Map of the Program of Postgraduate Study Figure 3.
Process Maps Model for Each Faculty Jambak et al.
The process mining method approach to analyze usersAo behavior of A p-ISSN: 1410-2331 e-ISSN: 2460-1217 Table 1.
Numerical Data Derivate from Process Maps Faculty of Economics Law
Engineering Medical Agriculture Education Social Political Computer Postgraduate Total
Number
of User
Case ID
Sum
1,532
2,894
2,005
1,307
1,365
1,838
12,809
Freq.
Access
Sum
616,532
533,685
1,612,540
1,284,058
359,063
729,692
186,326
1,272,670
1,291,689
7,886,255
Length Time of Access .
Mean 102,129.
112,514.
104,338.
159,639.
116,584.
95,931.
86,034.
157,709.
94,941.
Sum
156,462,902
101,487,742
301,956,120
320,076,300
61,090,228
125,382,764
38,027,461
215,272,809
174,502,880
1,494,259,206
Number of Events Number of Variants Per Case
Mean
Mean
1,224
Sum
616,532
533,685
1,612,540
1,284,058
359,063
729,063
186,326
1,27,2670
1,291,689
Sum
10,208,980
380,683
3,541,290
1,888,856
133,397
677,397
89,709
871,981
1,558,791
Table 2.
Purposes of Access Faculty of Economic Law Engineering Medical Agriculture Education Social & Political Computer Science Postgra-duate Service 486,819 .
421,966
1,275,232
1,005,074
286,615
584,244
142,118
1,058,319
1,036,392
The number and type of event data and variants vary greatly for each faculty, based on the destination port address.
Therefore, we have classified it as Purposes of Access into the Service category .
or the network machine destination por.
Academic category, and Non-Academic category, shown in Table 2.
The data in Table 1 and Table 2 are used as data sources for clustering in order to know the behavioral characteristics of users of Unsri LAN internet bandwidth access.
Clustering is done using the Centroid Linkage algorithm on the Hierarchical Cluster method and the k-Means algorithm on the Partition Cluster method.
Ideally, the results show that on the Unsri campus, users are categorized into three groups.
They are shown in Figure 4 and Figure 5.
The first group is users from the Faculty of Law.
Faculty of Agriculture.
Faculty of Education, and Faculty of Social Politics.
The characteristics of this group are shown by all data attributes of low internet usage.
Next is the second group which has a high level of internet usage on all attributes used, consisting of the Faculty of Medicine.
Academic 26,736 .
19,902
68,345
46,619
12,160
22,181
7,102
46,462
44,299
Non-Academic 102,977 .
91,817
268,963
232,365
60,288
123,267
37,106
167,889
210,998
Faculty of Computers.
Postgraduate Programs, and Faculty of Engineering.
While the Faculty of Economics, in particular, has its own characteristics, which are high in the number of users, low in time of access, low in events of access objectives, but has a very high variant of In the clustering process, factors that significantly affect cluster formation were also To get a statistical test has been carried out with the Analysis of Variance (ANOVA) test, the results of which are shown in Table 3.
Of the eleven attributes used, there are eight attributes that have a significant influence, namely attributes with a significant value less than or equal to 0.
The eight have been sorted by the calculated F value.
Therefore, the determination of bandwidth management will use these eight factors as the basis for setting it.
In comparison, the remaining three attributes are considered not to affect bandwidth usage.
Jambak et al.
The process mining method approach to analyze usersAo behavior of A SINERGI Vol.
No.
June 2022: 145-154 Figure 4.
Clustering result using the Centroid Linkage algorithm Figure 5.
Clustering result using the k-Means algorithm Jambak et al.
The process mining method approach to analyze usersAo behavior of A p-ISSN: 1410-2331 e-ISSN: 2460-1217 Table 3.
Analysis of Variance (ANOVA) Test Result Length of Total Access Mean of Variant Sum of User Case ID Number of Port of NonAcademic Number of Port of Academic Sum of Freq.
Access Number of Events Number of Port of Service Number of Variant Per Case Mean of Length Time of Access Mean of Number of Events Cluster Mean Square Error Mean Square Sig.
CONCLUSION
As the formulation of the problem solved in this research is: How to find out the behavior of internet users on LAN Unsri rationally to evaluate the effective and efficient use of internet bandwidth? This research has succeeded in modeling the behavior of internet users who use bandwidth from the Unsri LAN network by mapping the user access process from the beginning of access to the end of access, with the algorithm contained in the Mining Process Method.
The behavior in question is reflected in the map of the processes carried out by users in accessing the internet.
From the Analysis of the process map obtained, the authors found that there are factors that characterize the behavior of internet users, which can be a standard formulation for the implementation of Bandwidth Management and Bandwidth Allocation on the Unsri campus LAN.
The proposed formulation divides users into three requirements, low bandwidth requirements, and The characterize and need to be considered in bandwidth management, in order of significance, are Length of Total Access.
Mean of Variant.
Sum of User Case ID.
Number of Port of NonAcademic.
Number of Port of Academic.
Sum of Frequency of Access.
Number of Events, and Number of Port of Service.
ACKNOWLEDGMENT
The research and publication of this article were funded by the DIPA of Public Service Agency of Universitas Sriwijaya 2020.
SP DIPA023.
677515/2020, on March 16, 2020.
accordance with the RectorAos Decree Number:
0685/UN9/SK.
BUK.
KP/2020, on July 15, 2020.
We thank our colleagues from Malaysia Institute of Information Technology (MIIT).
Universiti Kuala Lumpur, who provided insight and expertise that greatly assisted the research.
REFERENCES