Optimization of Internet of Things Remote Desktop Protocol for Low-Bandwidth Environments Using Convolutional Neural...

Area of Science:

• Computer Science
• Image Processing
• Network Engineering

Background:

• Remote desktop tools are essential for efficiency but strain bandwidth.
• Traditional JPEG compression for remote desktops causes quality loss.
• Deep learning offers advanced image compression alternatives.

Purpose of the Study:

• To optimize desktop image quality and bandwidth in remote IoT GUI scenarios.
• To introduce an improved Remote Frame Buffer (RFB) protocol using CNNs.
• To enhance visual perception in remote desktop image processing.

Main Methods:

• Developed an optimized RFB protocol incorporating a convolutional neural network (CNN) image compression algorithm.
• Focused on human visual perception for desktop image processing.
• Evaluated performance against unoptimized RFB protocols.

Main Results:

• Achieved 30-80% bandwidth savings compared to unoptimized RFB.
• Demonstrated enhanced remote desktop image quality using PSNR and MS-SSIM metrics.
• Provided superior desktop image transmission quality.

Conclusions:

• The CNN-based optimized RFB protocol offers significant bandwidth reduction.
• The proposed method improves image quality metrics for remote desktop applications.
• This approach enhances the efficiency and user experience of remote IoT GUI systems.