This study proposes a deep autoencoder-based method for IoT traffic classification and anomaly detection. The method combines Convolutional Neural Networks (CNN) and Variational Autoencoders (VAE) to enhance the detection of security threats in IoT environments. The CNN model achieves a high accuracy of 95.85% in classifying IoT traffic, effectively distinguishing between different types of traffic. The VAE model excels in anomaly detection by capturing abnormal patterns through reconstruction loss and KL divergence. The combined approach provides a comprehensive solution for IoT cybersecurity. The study addresses the challenges of traditional machine learning methods in detecting anomalies in high-dimensional, noisy IoT traffic data. It explores three main deep learning-based anomaly detection methods: deep Boltzmann machines, stacked autoencoders, and CNNs. While CNNs offer strong robustness and high detection performance, they require converting traffic data into images, increasing data processing complexity. The proposed method uses particle swarm optimization (PSO) to optimize the structure of the deep autoencoder (SDA), enabling effective feature extraction and anomaly detection. The study also discusses related work, including machine learning and deep learning-based attack detection technologies. It highlights the limitations of traditional methods, such as high data requirements and data imbalance, which can affect detection accuracy. The VAE model is introduced as a solution that can capture data potential structures and generate new samples, improving data reconstruction and generalization. The experimental results show that the CNN model achieves high accuracy in classifying IoT traffic, while the VAE model effectively detects anomalies. The study concludes that the CNN-VAE approach offers significant advantages in IoT cybersecurity, but further research is needed to improve model generalization and performance in real-world applications. The study emphasizes the importance of continuous research and optimization to enhance IoT security and build a more secure and reliable IoT environment.This study proposes a deep autoencoder-based method for IoT traffic classification and anomaly detection. The method combines Convolutional Neural Networks (CNN) and Variational Autoencoders (VAE) to enhance the detection of security threats in IoT environments. The CNN model achieves a high accuracy of 95.85% in classifying IoT traffic, effectively distinguishing between different types of traffic. The VAE model excels in anomaly detection by capturing abnormal patterns through reconstruction loss and KL divergence. The combined approach provides a comprehensive solution for IoT cybersecurity. The study addresses the challenges of traditional machine learning methods in detecting anomalies in high-dimensional, noisy IoT traffic data. It explores three main deep learning-based anomaly detection methods: deep Boltzmann machines, stacked autoencoders, and CNNs. While CNNs offer strong robustness and high detection performance, they require converting traffic data into images, increasing data processing complexity. The proposed method uses particle swarm optimization (PSO) to optimize the structure of the deep autoencoder (SDA), enabling effective feature extraction and anomaly detection. The study also discusses related work, including machine learning and deep learning-based attack detection technologies. It highlights the limitations of traditional methods, such as high data requirements and data imbalance, which can affect detection accuracy. The VAE model is introduced as a solution that can capture data potential structures and generate new samples, improving data reconstruction and generalization. The experimental results show that the CNN model achieves high accuracy in classifying IoT traffic, while the VAE model effectively detects anomalies. The study concludes that the CNN-VAE approach offers significant advantages in IoT cybersecurity, but further research is needed to improve model generalization and performance in real-world applications. The study emphasizes the importance of continuous research and optimization to enhance IoT security and build a more secure and reliable IoT environment.