No abstract

In big data, noise in data mining is a necessity. Its existence depends on data and algorithm, but it does not mean the algorithm caused noise. Although the advantages of the Density Based Spatial Clustering Application with Noise, DBSCAN algorithm, in executing spatial data (two-dimensional data) have been widely discussed, but it has not been convincing in executing non-spatial data. As an algorithm should perform well on any data for optimizing data mining, this research proposes a trial to convert dimensions of non-spatial data into 2 dimensions for executing with DBSCAN algorithm, and a different input value for epsilon to know about its minimum which begins arising noise in the execution. Method of analysis in trial is with considering the attributes of non-spatial data as variables that represent coordinate points, rather than cardinality. Technically, it is assumed that 2-dimensional coordinate axes as a spot point for coordinate with more than or equal 3 dimensions according to development of Cartesian coordinate system, by first paying attention to relationship of variables (attributes). This way is then called Spatial Coordinate. The different input values are with paying attention to numbers from non-zero minimum distance to the forth of epsilon where the epsilon is in integer. The results of trial and testing on clusters formed, with Silhouette Coefficient, point out that the clusters are well, strong, and quality enough. Therefore, this research gives a new way on how preprocessing non-spatial data for DBSCAN algorithm performance.

LowRate DDoS (LDDoS) is a variation of DDoS attack that sends fewer packets than conventional DDoS attacks. However, by sending a smaller number of packets and using a unique attack period, low-rate DDoS is very effective in reducing the quality of an internet network-based service due to full access. On the other hand, the low-rate DDoS with its nature also makes it difficult to detect because it looks more mixed with normal user access. The Deep Learning model that will be used in this research is the RNN LSTM (Long Short Term Memory) model. LSTM is a neural network architecture which is good enough to process sequential data. This model is better than the simple RNN model. The research method is adapted to the SKKNI  No. 299 of 2020. However, this research will be carried out until the model development stage, namely the evaluation model. From the results of the research that has been done, it can be concluded that the RNN LSTM model can be used to classify low-rate DDOS attacks using feature selection. The accuracy of the training data on the validation data is around 98% and after visualizing the data for accuracy and loss, it can be concluded that the model is quite good, aka there is no underfitting or overfitting. While the accuracy obtained for testing data is 0.97%.

Tak dapat dipungkiri saat ini internet merupakan sesuatu yang sangat penting untuk berbagai kebutuhan. Tidak terkecuali di STMIK Widya Cipta Dharma. Internet banyak digunakan dalam lingkungan kampus, baik oleh mahasiswa, dosen dan juga tenaga kependidikan. Kegiatan belajar mengajar dan juga pekerjaan dalam lingkungan kampus tidak terlepas dari kebutuhan penggunaan internet. Namun waktu penggunaan internet juga terkadang menumpuk dalam jam-jam tertentu dan menyebabkan kecepatan internet menjadi lambat. Hal itu dipengaruhi oleh banyaknya pengiriman paket header pada flow/arus lalu lintas internet sehingga koneksi menjadi berat dan terasa lambat. Oleh karena itu, dibutuhkan suatu metode klasifikasi yang dapat memberikan informasi mengenai aktivitas mahasiswa, dosen dan tenaga kependidikan dalam penggunaan internet. Adapun algoritma klasifikasi yang digunakan adalah klasifikasi Support Vector Machine (SVM). Metode pengembangan yang digunakan adalah SKKNI Nomor 299 Tahun 2020. Parameter yang digunakan adalah arus paket yang dikirim oleh user dan paket yang diterima oleh user. Adapun hasil penelitian ini berupa model algoritma SVM yang dapat mengklasifikasikan arus penggunaan trafik internet dengan empat kategori yaitu Download, Game, SocialNetwork, dan Web yang memiliki akurasi 64% dengan menggunakan kernel Radial Basis Function (RBF). Hasil akurasi yang dihasilkan cukup rendah dan membuat algoritma SVM tidak cocok untuk melakukan klasifikasi trafik internet dan perlunya metode lain untuk mengklasifikasikan trafik internet.

Ddos is an attack method by sending a lot of packets into a network that causes the device not to run according to its function. This attack will result in machine or network resources cannot be accessed or used by the user. Various methods are used to detect DDOS attacks on SDN [4] , namely statistical methods, machine learning, SDN architecture, blockchain, Network Function Virtualization, honeynets, network slicing, and moving target defense. Because so many people use machine learning to detect DDoS attacks, it is necessary to do further research to find out which one is the best and has high accuracy. Therefore, a research entitled “Comparison of Machine Learning Algorithms in Detecting DDoS Attacks was made. In this study, three machine learning algorithms will be compared, namely XGBoost, Decision Tree and ANN. The methods used are data acquisition, data understanding, data preparation, modeling, performance evaluation, and conclusions. In this study it can be said that for accuracy, the highest model is XGBoost in determining attacks, but to execute it requires the longest time among other models tested. While Decision tree also has high accuracy, slightly below XGBoost, but the time required to execute is fast or short. Therefore, in this study it can be said that the Decision Tree is the best model in detecting and classifying DDoS attacks.Keywords: Ddos Attack, Machine Learning, Decision Tree, XGBoost, ANN.