Streaming Data FEC Bitrate Control for Bursty Packet Loss
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for data transport over unreliable protocols, such as UDP, face challenges in efficiently utilizing network bandwidth while avoiding congested conditions that lead to unacceptable packet loss, especially in real-time applications like live audio/video streaming and online gaming, where bursty loss occurs.
Innovation Solution
The method involves encoding current frames at a higher bitrate and previous frames at a lower bitrate for forward error correction, with the encoded source and FEC frames being packetized and transmitted, allowing recipients to reconstruct lost data without retransmission, and adjusting the bitrate based on feedback to optimize data delivery.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If forward error correction packets are transmitted at the same high bitrate as source packets, then data reconstruction quality is improved, but network bandwidth consumption increases and may cause congestion
Solution Approach 1:
The patent applies different bitrates to different types of packets: source packets are transmitted at high bitrate to maintain quality, while FEC packets are transmitted at lower bitrate since they are redundancy data. This local differentiation of quality levels resolves the contradiction by optimizing bandwidth usage without sacrificing reconstruction capability.
Solution Approach 2:
The patent dynamically changes the bitrate parameter for FEC packets based on network conditions and packet type. By adjusting the bitrate parameter locally for FEC packets rather than applying a uniform high bitrate to all packets, the system maintains reconstruction quality while reducing overall bandwidth consumption.
2Reliability
If more FEC packets are transmitted to handle bursty losses, then packet loss recovery is improved, but network congestion increases leading to more packet losses
Solution Approach 1:
The patent transmits FEC packets at reduced bitrate compared to source packets, providing error protection where needed while minimizing the bandwidth impact that could cause congestion. This local quality adjustment allows FEC packets to contribute to loss recovery without proportionally increasing congestion risk.
Solution Approach 2:
The patent provides partial error correction by transmitting fewer FEC packets at lower bitrate rather than full-rate FEC for all packets. This partial action is sufficient to handle typical bursty losses while avoiding the excessive bandwidth consumption that would cause congestion.
3Reliability
If source packets are transmitted at high bitrate for quality, then perceived quality is improved, but bandwidth usage increases reducing capacity for error correction packets
Solution Approach 1:
The patent applies high bitrate only to source packets where quality is critical, while using lower bitrate for FEC packets where quality is less important since they are redundancy data. This local quality differentiation maximizes perceived quality within available bandwidth constraints.
Solution Approach 2:
The patent changes the bitrate parameter based on packet type: high bitrate for source packets to maintain perceived quality, and lower bitrate for FEC packets to conserve bandwidth for error correction purposes.
Data Source
AI summary
Encoded information corresponding to the encoded source frames and the one or more previous frames is unpackaged from each data packet. Each data packet in the plurality contains encoded information corresponding to a source frame in a sequence encoded at a first bitrate and one or more previous frames in the sequence encoded as forward error correction (FEC) frames at a second bitrate that is equal to or lower than the first bitrate. The encoded source frames are decoded to generate corresponding decoded source frames. Encoded FEC frames that correspond to a given source frame for which encoded information is missing are decoded to generate corresponding decoded FEC frames. A reconstructed frame is generated corresponding to the given source frame using the one or more decoded FEC frames. The decoded source to frames and the reconstructed missing frame are stored in a memory and/or presented with a display.