Enhancing IoT Network Security with Hybrid Deep Learning and Ensemble Models for Botnet Detection
G. Nagaleela, Chilamakuru Nagesh, M Raghavendra, Farhana Bano, N. Srihari Rao, Aelluri Lakshmi · 2025
Exponential growth in IoT networks has made those networks increasingly vulnerable to botnet attacks that can lead to severe security violations. One of the major shortcomings of traditional machine learning based technologies is to use for high false-positive rates and inefficient real-time detection. This proposed work presents a novel hybrid machine learning framework in this paper. Proposes a framework that merges ensemble learning and deep learning to enhance the efficiency of botnet detection tasks. A model has been proposed that implements feature selection via genetic algorithm (GA) and an optimum stacked ensemble classifier comprising a convolutional neural network (CNN) and gradient-boosting decision tree (GBDT). The BoT-IoT and N-BaIoT data sets show that models outperform the SOTA models concerning accuracy, latency, and detection rates. Indeed, the proposed architecture would be a viable solution to Internet of Things security concerns as it combines strong mitigations for botnet threats with real-time detection capabilities.