Deep Learning Driven Secure Music Traffic Transmission in Consumer Internet of Things

The popularization of Consumer Internet of Things (CIoT) has brought unprecedented convenience. However, rapid development has led to new challenges in the secure transmission of music traffic in CIoT, such as data leakage and privacy infringement. Therefore, this paper proposes a deep learning driven method for secure music traffic transmission in CIoT. This method first adopts bidirectional long short-term memory time space feature fusion structure module (BTSF) to achieve secure detection of music traffic. To address the issue of traffic loss, this model further introduces attention mechanism and constructs music traffic feature enhancement module. Meanwhile, music traffic security discriminator in the model combines bidirectional long short-term memory (Bi-LSTM) and convolutional neural network (CNN) for cascaded network fusion, improving the security detection and recognition capabilities of music traffic. Then, in order to enhance the security of data transmission, this method uses the new traffic encryption transmission scheme based on MQTT (MQTT-TE). It reduces time overhead and effectively resists external threats. Through experimental comparison, BTSF performs better in detecting music traffic security than other models. MQTT-TE has shorter encryption and decryption times. It makes traffic transmission more stable and overall secure in CIoT.