Advanced Convolutional Neural Network and Long Short-Term Memory Model for Real-Time Spam Detection in Internet of Things Devices

Abstract

The proliferation of the Internet of Things has seen higher vulnerability to different cyber threats, most notably spam attacks, represents a major risk for device functionality and user privacy. In this study, Authors presents an advanced hybrid deep learning model with the combination of Convolutional Neural Networks and Long Short-Term Memory networks for real-time spam detections on IoT devices. This CNN-LSTM model is designed to deal with the complex and changing IoT environments. This model handles the complexities of spam detection in the sequence data streams common in IoT networks by combining spatial feature extraction offered by CNNs and potential temporal pattern recognition capabilities at which LSTMs excel. One of the fundamental parts of this model lies in its suitability for device strain and limited computational resources. It works well enough that spam can be effectively filtered out, but with a nearly zero processing load on IoT systems. This is important because performance like this will be equivalent to the latency constraints in many real-time applications. Comprehensive testing with real-world datasets shows that this CNN-LSTM model performs better than traditional detection methods, achieving high accuracy and low latency. This move adds to the wider effort of creating more cost-effective, real-time cybersecurity solutions for IoT ecosystems while boosting security and reliability for large-scale IoT deployments.