Secure Indoor Positioning System Model Dengan Perlindungan Terhadap Serangan Projected Gradient Descent (PGD)

Sudirlan, Salman Al Farisi (2024) Secure Indoor Positioning System Model Dengan Perlindungan Terhadap Serangan Projected Gradient Descent (PGD). Other thesis, Institut Teknologi Sepuluh Nopember.

Text 5027201056-Undergraduate_Thesis.pdf - Accepted Version Restricted to Repository staff only until 1 April 2026. Download (9MB) | Request a copy

Abstract

Dalam era digital saat ini, sistem pemosisian dalam ruangan (Indoor Positioning System, IPS) menjadi semakin penting, misalnya untuk menemukan produk di pusat perbelanjaan atau pasien di rumah sakit. Namun, serangan dan ancaman keamanan pada teknologi ini semakin meningkat, menempatkan privasi pengguna dalam risiko. Penelitian "Secure Indoor Positioning System Model Menggunakan Metode Serangan Projected Gradient Descent (PGD)" berfokus pada meningkatkan keamanan IPS melalui integrasi dengan Projected Gradient Descent (PGD), sebuah teknik yang mengintervensi dan memodifikasi data masukan untuk mengelabui model machine learning. Dengan mengkombinasikan kecerdasan buatan (Artificial Intelligent, AI) dan PGD, Penulis berniat meningkatkan ketahanan model terhadap serangan eksternal. Sebagai demonstrasi, model dengan sistem pemosisian yang aman dan efisien. Dari hasil penelitian yang dilakukan didapatkan bahwa model yang aman dari serangan PGD dapat direalisasikan dengan menggunakan metode neural structured network. Penielitian ini merupakan pengembangan dari penelitian dengan judul Few-Shot Transfer Learning for Device-Free Fingerprinting Indoor Localization dengan akurasi EGNN sebesar 90.44%. Model terbaik hasil penelitian ini merupakan model LSTM dengan tambahan lapisan CNN dan DNN pada dataset secara utuh dan dapat meraih akurasi sebesar 80.74% pada dataset Tower 2 ITS dan akurasi sebesar 99.8% pada dataset penelitian terdahulu.
=================================================================================================================================
In today's digital age, indoor positioning systems (IPS) are becoming increasingly important, for example to locate products in shopping malls or patients in hospitals. However, attacks and security threats on this technology are increasing, putting user privacy at risk. The research "Secure Indoor Positioning System Model Using Projected Gradient Descent (PGD) Attack Method" focuses on improving IPS security through integration with Projected Gradient Descent (PGD), a technique that intervenes and modifies input data to trick machine learning models. By combining artificial intelligence (AI) and PGD, the authors intend to improve the model's resilience to external attacks. As a demonstration, a model with a secure and efficient positioning system. From the results of the research conducted, it was found that a model that is safe from PGD attacks can be realized using the neural structured network method. This research is a development of research with the title Few-Shot Transfer Learning for DeviceFree Fingerprinting Indoor Localization with EGNN accuracy of 90.44%. The best model resulting from this research is the LSTM model with additional CNN and DNN layers on the dataset as a whole and can achieve an accuracy of 80.74% on the Tower 2 ITS dataset and an accuracy of 99.8% on the previous research dataset.

Item Type:	Thesis (Other)
Uncontrolled Keywords:	Kecerdasan Artifisial, PGD, Pelokalan Dalam Ruangan, Artificial Intelligence, indoor positioning
Subjects:	T Technology > T Technology (General) > T11 Technical writing. Scientific Writing T Technology > T Technology (General) > T385 Visualization--Technique T Technology > T Technology (General) > T57.5 Data Processing
Divisions:	Faculty of Intelligent Electrical and Informatics Technology (ELECTICS) > Information Technology > 59201-(S1) Undergraduate Thesis
Depositing User:	Salman Al Farisi Sudirlan
Date Deposited:	05 Feb 2024 05:21
Last Modified:	05 Feb 2024 08:16
URI:	http://repository.its.ac.id/id/eprint/106083

Actions (login required)

View Item