ANALISIS TRAFFIC UNTUK MONITORING JARINGAN DENGAN METODE CLUSTERING K-MEANS DAN DBSCAN
Hasna Rafika, Mardhani Riasetiawan, M.T., Dr
2023 | Skripsi | ILMU KOMPUTER
Currently, a lot of data storage and files are stored online in cloud computing, including
network traffic data. Besides that, there is big data technology which is large data that can
potentially be processed as information for machine learning. In addition, the existence of cloud
computing and big data causes data to overflow. Then it requires management of increasing
resource requirements. So the challenges faced are classification in real-time, minimum
resources, and high accuracy and reliability. The Computer and Network Systems Laboratory
has many engines, namely Gama Cloud, Gamabox Big Data, data engines, sensors, big energy
data. All of those engines carry out running processes that enter through the network in the
Laboratory. With so much data crossing the network, it is necessary to analyze what kind of
computer network traffic transaction patterns are in order to monitor computer networks in the
SKJ Laboratory data center more precisely.
In this study using traffic data at Laboratory SKJ server that capture using Wireshark.
The steps are selecting the network you want to capture, capturing the network, performing the
Wireshark sniffing process, viewing and analyzing the contents of the packet, and analyzing
the flow of network packets and the type of data captured. The amount of data taken is 1010
data. After the data is collected, it will be clustered using the K-Means and DBSCAN
algorithms, where the results of these two methods will be compared.
The output of this research is the result of traffic analysis for network monitoring in the
form of clustering the types of traffic running on the network. Tests were carried out using the
DBI and Silhouett Coefficient. Based on the DBI results, the M-Means method obtains more
optimal cluster results, whereas based on the value of the silhouette coefficient, the two
methods have varying values with different interpretations of the cluster structure. The
strongest cluster structure with K-Means is at a value of 0.801 while for a cluster with a strong
structure the DBSCAN method is at a value of 0.972.
Kata Kunci : Jaringan, DBSCAN Clustering, K-Means Clustering, Monitoring, Traffic,Wireshark