Prediksi Mobilitas User Pada Virtual Small Cell Dengan Algoritma Long Short-Term Memory

Firmansyah, Muhammad Rizky (2023) Prediksi Mobilitas User Pada Virtual Small Cell Dengan Algoritma Long Short-Term Memory. Other thesis, Institut Teknologi Sepuluh Nopember.

Text 07111940000099-Undergraduate_Thesis.pdf - Accepted Version Restricted to Repository staff only until 1 October 2025. Download (2MB) | Request a copy

Abstract

Dengan pertumbuhan dari penggunaan peralatan cerdas seperti smartphone menyebabkan peningkatan trafik data dan kebutuhan jaringan yang semakin tinggi. Untuk memenuhi kebutuhan pengguna, dibutuhkan penambahan dari segi kapasitas maupun seperti infrastruktur untuk mengimbangi kebutuhan akan jaringan. Akan tetapi muncul masalah, yaitu apabila distribusi dan mobilitas dari pengguna yang berfluktuasi sehingga kebutuhan cluster hostpot dapat berubah secara fungsi waktu. Dengan kemampuan memprediksi mobilitas berdasarkan trafik internet dapat dilakukan planning yang lebih efisien, mengontrol konsumsi resource jaringan sesuai dengan kebutuhan, dan meningkatkan kualitas keputusan. Oleh karena itu, efektivitas dan keakuratan dari prediksi kemunculan trafik dan hotspot ini dapat memberikan informasi berharga.
Pada penelitian tugas akhir ini akan dilakukan prediksi mobilitas user secara fungsi waktu dengan menggunakan algoritma long short term-memory (LSTM). Dengan informasi pola mobilitas user, selanjutnya dapat dilakukan pengelompokkan cluster hotspot dengan K-Means Clustering. Sehingga nantinya dapat dimanfaatkan untuk mendesain beamforming untuk diarahkan ke tiap cluster dengan mengatur sumber daya jaringan berbasis virtuall cell dalam membuat coverage area atau hotspot secara fleksibel dan adaptif. Kualitas prediksi model LSTM diukur dengan metrics utama MSE dan metrics pendukung MAE dan MAPE. Berdasarkan hasil pengujian, didapati model LSTM dengan performansi yang terbaik adalah penggunaan 50 neuron, 50 epoch dan jenis optimizer Adam dengan skor 0.513 untuk MSE, 0.518 MAE dan 3.13 untuk MAPE. Jumlah cluster yang dibuat dilihat dari pengukuran silhouette score dan metode elbow yang selanjutnya melakukan beamforming ke titik tengah cluster yang terbentuk.
======================================================================================================================================
With the growth of the use of smart devices such as smartphones, it causes an increase in data traffic and higher network necessity. To meet user necessity, additional capacity and infrastructure are needed to compensate the necessity for networks. However, a problem arises, namely if the distribution and mobility of users is fluctuate so that the needs of the hotspot cluster can change as a function of time. With the ability to predict user mobility, it also useful for planning efficiency, controlling resource utilization according to necessity, and improving the quality of decisions. Therefore, the effectiveness and accuracy of predicting the emergence of traffic and hotspots can provide valuable information.
In this final project research, prediction of user mobility will be made as a function of time using the long short term-memory (LSTM) algorithm, because it can analyze user’s patterns sequentially. With information of user mobility patterns, hotspot clusters can be grouped using K-Means Clustering. So that later it can be used to design beamforming to be directed to each cluster by managing network resources based on virtual small cells in making coverage areas or hotspots in a flexible and adaptive way. The predictive quality of the LSTM model is measured by the main MSE metrics and the supporting metrics MAE and MAPE. Based on the test results, it was found that the LSTM model with the best performance was the use of 50 neurons, 50 epochs and the Adam optimizer type with a score of 0.513 for MSE, 0.518 MAE and 3.13 for MAPE. The number of clusters made is seen from the silhouette score measurement and the elbow method which then performs beamforming to the center point of the cluster formed.

Item Type:	Thesis (Other)
Uncontrolled Keywords:	Antena, MIMO, Beamforming, VSC, Deep Learning, LSTM, K-Means Clustering. Antena, MIMO, Beamforming, VSC, Deep Learning, LSTM, K-Means Clustering.
Subjects:	T Technology > T Technology (General) > T174 Technological forecasting T Technology > T Technology (General) > T57.5 Data Processing T Technology > TK Electrical engineering. Electronics Nuclear engineering > TK7871.6 Antennas (Electronics) T Technology > TK Electrical engineering. Electronics Nuclear engineering > TK5101 Telecommunication
Divisions:	Faculty of Intelligent Electrical and Informatics Technology (ELECTICS) > Electrical Engineering > 20201-(S1) Undergraduate Thesis
Depositing User:	Muhammad Rizky Firmansyah
Date Deposited:	27 Jul 2023 01:06
Last Modified:	27 Jul 2023 01:06
URI:	http://repository.its.ac.id/id/eprint/99806

Actions (login required)

View Item