Denial of Service Attack Detection and Mitigation using Ensemble based ML in Software Defined Network

Software-defined networking (SDN) introduces a novel networking paradigm by decoupling control logic from forwarding operations, resulting in improved network resource management and administration efficiency. However, centralized network design in SDN gives rise to significant security challenges, notably denial-of-service (DoS) attacks. Attackers exploit SDN’s lack of effective message-verification mechanisms to launch DoS attacks by falsifying source address information. In response, this study proposes an innovative approach to Internet of Things (IoT) security leveraging software-based networks and collaborative learning. One key strategy involves partitioning the network domain into smaller subdomains, each with its own controller, facilitating the exchange of security rules among subdomains. However, this topology necessitates routing every packet through the subnet’s control node. To continually monitor network traffic data and detect attacks, each controller node employs an integrated learning model. This model combines statistical insights from data streams with artificial neural networks and the Extreme Gradient Boosting Algorithm to anticipate potential attacks. Additionally, the ICOA model selects relevant characteristics for analysis. Through comprehensive performance evaluation, including empirical values, the ensemble model demonstrates superior performance compared to existing approaches. Furthermore, the study compares various models and methodologies, affirming the efficacy of the proposed strategy in accurately identifying Distributed Denial-of-Service (DDoS) attacks in SDNs, thereby presenting a groundbreaking approach to SDN security.