IoT botnet detection from software defined network using American zebra optimization algorithm with SSRNN-ELM

Software-defined networking (SDN) offers centralized control over large-scale networks, enhancing flexibility and enabling tailored network applications. However, SDN introduces new security vulnerabilities, including the risk of botnet attacks, which can compromise systems and steal data. This paper proposes a novel approach for detecting botnet attacks in SDN environments by leveraging a hybrid model that integrates recurrent neural networks (RNNs) and extreme learning machines (ELMs). The proposed method utilizes RNNs for feature learning and ELMs for classification, with the American zebra optimization algorithm (AZOA) optimizing ELM weights. This hybrid SSRNN-ELM (supervised subset recurrent neural network—extreme learning machine) approach is evaluated using N-BaIoT dataset and performance metrics, demonstrating effective detection of complex botnet attacks with high accuracy. The results demonstrate that the hybrid model effectively identifies complex botnet attacks, achieving a detection accuracy of 96.60%. The integration of recurrent neural networks (RNNs) for feature learning and extreme learning machines (ELMs) for classification, along with optimization by the American zebra optimization algorithm (AZOA), significantly improves detection performance as compared to the recent existing studies.