A Robust ACLR Hybrid Approach for Real-Time Botnet Threat Detection in IoT Devices

A crucial component of cybersecurity is the detection of botnet attacks, especially when it comes to protecting Internet of Things (IoT) networks from constantly changing threats. ANN, CNN, RNN, and LSTM are examples of deep learning models or standalone machine learning that are trained on historical attack data and are commonly used in existing IoT botnet detection systems. Their autonomous functioning, however, reduces the detection accuracy against complex and ever-changing botnet attacks. This research suggests a strong Hybrid Federated Learning-based framework, called ACLR, to overcome this constraint. It combines ANN, CNN, RNN, and LSTM models into a single ensemble architecture. Moreover, an attention technique is used to highlight the most pertinent traffic characteristics, improving feature representation and detection efficiency. The suggested ACLR model is appropriate for real-time IoT security applications since experimental results show that it greatly increases botnet detection accuracy and robustness when compared to individual learning methods.