FEDERATED LEARNING APPROACHES FOR IOT INTRUSION DETECTION BASED ON FEDAVG AND FEDPROX ON IID AND NON-IID DATA - Dalam bentuk buku karya ilmiah

SATRIA WINEKAS HERLAMBANG

Informasi Dasar

36 kali
25.05.733
000
Karya Ilmiah - Thesis (S2) - Reference

The increasing complexity of cyberattacks in the Internet of Things (IoT) architecture requires an Intrusion Detection System (IDS) that is not only accurate in detecting threats, but also capable of maintaining user data privacy. Conventional IDS approaches based on a centralized architecture are considered ineffective because they are vulnerable to single points of failure and pose a risk of data leaks. To address this, Federated Learning (FL) emerges as an innovative solution that enables collaborative model training among devices without need to send raw data to a central server. However, the effectiveness of FL is significantly influenced by the choice of aggregation method, which impacts model convergence, detection accuracy, and system resilience, especially when handling data with non-independent and identically distributed (non-IID) characteristics. Several researchers have discussed aggregation methods in FL related to non-IID data, but without explicitly evaluating the effect of regularization strength on FedProx under non-IID data distribution conditions. This thesis proposes the development of a Collaborative Intrusion Detection System (CIDS) based on FL using two aggregation methods, FedAvg and FedProx to evaluate attack detection performance on the CICIoT2023 dataset. Experiments were conducted on IID and Non-IID data scenarios using a Deep Neural Network (DNN) model and feature selection techniques based on the Feature Importance Gain values from XGBoost. The results show that on IID data, the combination of FedAvg and feature selection provides the highest accuracy of 97.76%. Meanwhile on non-IID data, the combination of FedProx (? = 1.0) and feature selection achieves the best accuracy of 96.92%. 

Subjek

DEEP LEARNING
 

Katalog

FEDERATED LEARNING APPROACHES FOR IOT INTRUSION DETECTION BASED ON FEDAVG AND FEDPROX ON IID AND NON-IID DATA - Dalam bentuk buku karya ilmiah
 
xi, 54p.: il,; pdf file
English

Sirkulasi

Rp. 0
Rp. 0
Tidak

Pengarang

SATRIA WINEKAS HERLAMBANG
Perorangan
Favian Dewanta, Yudha Purwanto
 

Penerbit

Universitas Telkom, S2 Teknik Elektro
Bandung
2025

Koleksi

Kompetensi

  • TTI6D3 - INTERNET OF THINGS
  • TTI6T3 - KEAMANAN INFORMASI
  • TTH6M3 - KEAMANAN JARINGAN
  • AAK6FFB3 - Keamanan Jaringan Lanjut
  • TEI6G3 - PEMBELAJARAN MESIN LANJUT
  • TTI6A3 - PEMBELAJARAN SECARA STATISTIK DAN OPTIMISASI

Download / Flippingbook

 

Ulasan

Belum ada ulasan yang diberikan
anda harus sign-in untuk memberikan ulasan ke katalog ini