Reducing Video Glitches from Lost Network Packets

This patent describes a method for video streaming to reduce visual errors caused by lost network packets by cleverly identifying specific recovery points within the video data stream.

What it covers

The patent aims to reduce the number of video packets that are negatively affected when some packets are lost during network transmission. It achieves this by analyzing the 'error resilience' built into the video packets. When one or more packets are lost, the system identifies a special 'first packet' that arrives after the loss. This 'first packet' must carry the start of a 'tile' within the video data stream and also contain a 'slice' whose header is found in a packet that was successfully received. This targeted recovery point helps reconstruct the video efficiently, as stated in the abstract, with low 'side information overhead'. For example, if you're watching a video call and your internet briefly cuts out, this method tries to quickly find a good spot in the incoming video data to restart showing a clear picture, minimizing the frozen or distorted frames.

What it doesn't cover

• —Does not cover error handling for general data streams that are not specifically video data streams.
• —Does not cover error recovery methods that do not rely on identifying 'tiles' or 'slices' within the video data.
• —Does not cover techniques where the 'slice header' for the recovery point is also contained in a lost packet.
• —Does not cover error resilience methods that incur high 'side information overhead' compared to the reduction in affected packets.
• —Does not cover error correction techniques that retransmit lost packets rather than recovering from subsequent data.

The clever bit

The novelty lies in precisely identifying a recovery point after packet loss by looking for a packet that simultaneously starts a video 'tile' and has a 'slice header' from an unlost packet. This specific combination allows for efficient video reconstruction with minimal extra data needed for error resilience.

Why it matters

Packet loss is a common issue in network communication, especially for real-time video streaming over the internet. This patent addresses a core problem of delivering smooth, high-quality video even when network conditions are imperfect. By reducing visual artifacts and interruptions caused by lost data, it contributes to a better user experience in video conferencing, online entertainment, and other video-intensive applications.

Real-world examples

• 1.Video conferencing platforms like Zoom or Google Meet
• 2.Video streaming services such as Netflix or YouTube
• 3.Live broadcasting over IP networks
• 4.Cloud gaming services that stream video to users
• 5.Video surveillance systems transmitting over unreliable networks