Variable-Size FEC Repair Packet Sizing for Streaming Bandwidth
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing FEC streaming architectures, such as RFC 2733, are inefficient in handling variable-sized source packets, leading to wastage of bandwidth due to fixed-sized repair packets and are limited to specific FEC codes and protocols like RTP, lacking flexibility in using advanced FEC codes and supporting various streaming protocols without modifications.
Innovation Solution
The implementation of a systematic FEC encoding and decoding method that generates repair packets based on variable-sized source blocks, allowing the use of any systematic FEC code and supporting multiple streaming protocols by adding a signaling field to source packets and using separate streams for source and repair packets, enabling efficient protection against packet loss without modifying the streaming protocol.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If repair packets are sized according to the maximum source packet size, then all source packets can be protected, but bandwidth is wasted when smaller source packets are transmitted
Solution Approach 1:
The patent segments the source packet into multiple fixed-size blocks (e.g., 128 bytes each) before applying FEC encoding. This segmentation allows the repair packet size to be determined by the number of blocks rather than the total packet size, enabling proportional protection and reducing bandwidth waste when source packets are smaller than the maximum size.
2Ease of manufacture
If a fixed FEC code is used for all source packets, then implementation is simple, but flexibility to use advanced FEC codes is limited
Solution Approach 1:
The patent introduces a dynamic FEC code selection mechanism where the FEC parameters (such as the number of repair blocks and encoding scheme) can be adjusted based on the actual source packet characteristics and transmission conditions. This allows the system to adapt between simple fixed codes and advanced dynamic codes as needed.
3Reliability
If protocol-specific FEC modifications are applied, then FEC can be optimized for that protocol, but compatibility with other protocols is reduced
Solution Approach 1:
The patent designs a universal FEC framework that can accommodate multiple streaming protocols (RTP, RTSP, custom protocols) through a protocol-agnostic block-based approach. The FEC encoder and decoder operate on standardized blocks with clear delimiters, allowing the same implementation to protect packets from different protocols without requiring protocol-specific modifications.
Data Source
AI summary
In a packet communications system stream data is transported over a channel over which packet loss or corruption is possible, with forward error correction (“FEC”) information. A transmitter receives source packets comprising source data, generates FEC source packets formatted to allow for identification of lost or corrupted source packets at a receiver, arranges source data from the source packets into a plurality of source symbols wherein at least one source packet is arranged into more than one source symbol, associates a plurality of source symbols with a source block, generates a plurality of repair symbols from the source block according to a predetermined FEC encoding process and groups the plurality of repair symbols into one or more FEC repair packets associated with the source block. A receiver can use the FEC repair symbols from the FEC repair packets to recover source symbols, as needed.