Optimizing QoE in IoT-Based Video Streaming through Deep Learning Algorithms

The demand for video streaming services over Internet of Things (IoT) networks has surged, yet maintaining a high Quality of Experience (QoE) remains challenging due to network heterogeneity and resource constraints. This paper presents an innovative deep learning-based approach to optimize QoE in IoT-based video streaming. Leveraging convolutional neural networks (CNN) and long short-term memory (LSTM) networks, the proposed framework dynamically adapts video streaming parameters in real-time, such as bit rate, frame rate, and resolution, based on the existing network conditions and device capabilities. Through extensive simulations and real-world deployments, our approach exhibited a 24.5% reduction in re-buffering ratio, a 12.5% increase in average bitrate, and a 15% improvement in video quality measured by SSIM, compared to conventional methods. Furthermore, it reduced network bandwidth consumption by up to 150% without compromising on video quality. These results underscore the effectiveness of applying deep learning algorithms to optimize video streaming in complex IoT environments.