TY - JOUR
T1 - PoAh-Enabled Federated Learning Architecture for DDoS Attack Detection in IoT Networks
AU - Park, Jin Ho
AU - Yotxay, Sangthong
AU - Singh, Sushil Kumar
AU - Park, Jong Hyuk
N1 - Publisher Copyright:
© (2024) Korea Information Processing Society.
PY - 2024
Y1 - 2024
N2 - Nowadays, the Internet of Things (IoT) has widely influenced many areas of human life; several advanced IoT applications and services are developing in smart cities. However, IoT and smart city applications have various issues and challenges, such as security, privacy preservation, data authentication, decentralization, and latency. Security and privacy are essential issues with distributed denial of service (DDoS) attack detection because of the limitation of security techniques and the heterogeneity of IoT devices. An attack detection system is deployed in the IoT network and classifies it. Blockchain technology is accumulating popularity in many applications, its ability to secure the system while discarding centralized requirements. Proof of authentication (PoAh) is used as a consensus mechanism to maintain secure system authentication, sustainability, and high scalability. Furthermore, federated learning has recently proposed to train local models (gated recurrent unit) and share with a global model for aggregation on IoT devices utilizing numerous user-generated data samples while reducing data loss. Therefore, we propose a PoAh-enabled federated learning architecture for DDoS attack detection in IoT networks. Federated learning is used at the federated layer for privacy preservation to mitigate the negative impacts required for fast processing, accuracy, stability, and low latency. Moreover, blockchain technology is utilized at the authentication layer with PoAh for ensures data authentication and validation, high security, and performance in IoT networks. Finally, we evaluate the proposed architecture with theoretical, quantitative, and security analysis and show that its accuracy, precision, recall, F1-score, and efficiency percentage is approximately 98.6% which is better than existing research studies.
AB - Nowadays, the Internet of Things (IoT) has widely influenced many areas of human life; several advanced IoT applications and services are developing in smart cities. However, IoT and smart city applications have various issues and challenges, such as security, privacy preservation, data authentication, decentralization, and latency. Security and privacy are essential issues with distributed denial of service (DDoS) attack detection because of the limitation of security techniques and the heterogeneity of IoT devices. An attack detection system is deployed in the IoT network and classifies it. Blockchain technology is accumulating popularity in many applications, its ability to secure the system while discarding centralized requirements. Proof of authentication (PoAh) is used as a consensus mechanism to maintain secure system authentication, sustainability, and high scalability. Furthermore, federated learning has recently proposed to train local models (gated recurrent unit) and share with a global model for aggregation on IoT devices utilizing numerous user-generated data samples while reducing data loss. Therefore, we propose a PoAh-enabled federated learning architecture for DDoS attack detection in IoT networks. Federated learning is used at the federated layer for privacy preservation to mitigate the negative impacts required for fast processing, accuracy, stability, and low latency. Moreover, blockchain technology is utilized at the authentication layer with PoAh for ensures data authentication and validation, high security, and performance in IoT networks. Finally, we evaluate the proposed architecture with theoretical, quantitative, and security analysis and show that its accuracy, precision, recall, F1-score, and efficiency percentage is approximately 98.6% which is better than existing research studies.
KW - Blockchain
KW - Federated Learning
KW - Internet of Things
KW - Network Security
KW - Proof of Authentication
KW - and Anomaly Detection
UR - http://www.scopus.com/inward/record.url?scp=85188784098&partnerID=8YFLogxK
U2 - 10.22967/HCIS.2024.14.003
DO - 10.22967/HCIS.2024.14.003
M3 - Article
AN - SCOPUS:85188784098
SN - 2192-1962
VL - 14
SP - 1
EP - 24
JO - Human-centric Computing and Information Sciences
JF - Human-centric Computing and Information Sciences
ER -