•  
  •  
 

Document Type

Article

Keywords

Federated learning, IOT, IDS, Cloud Computing, SDN

Abstract

The accelerated proliferation of Internet of Things (IoT) apparatuses has rendered intrusion detection and incident response progressively arduous owing to device diversity, constrained resources, and concerns regarding data confidentiality. Addressing these challenges is paramount to sustaining secure and resilient IoT ecosystems. This manuscript introduces an innovative framework that amalgamates software-defined networking (SDN) with segmented federated learning (SFL) to augment the effectiveness and reactivity of anomaly detection within the IoT. The proposed methodology delineates the federated learning (FL) process, facilitating lightweight, localized model training customized to the capabilities of individual IoT devices. The SDN is utilized to dynamically regulate network flows and implement real-time incident response measures. The proposed architecture is structured to reduce communication overhead, safeguard data privacy, and support participation from resource-limited nodes. A simulation-based evaluation strategy is proposed, with both execution and empirical substantiation anticipated in forthcoming stages. This integrated SFL-SDN paradigm provides a scalable and privacy-conscious solution for fortifying IoT infrastructures and is anticipated to surpass conventional centralized and nonsegmented FL methodologies in intricate, real-time threat scenarios

Download

Share

COinS