Precoding by Priority: A UEP Scheme for RaptorQ Codes

TL;DR

This paper introduces a Priority Based Precode Ratio (PBPR) method to provide Unequal Error Protection in RaptorQ codes, enhancing error correction for important data without significant computational overhead.

Contribution

The paper proposes a novel PBPR approach that achieves UEP in RaptorQ codes by adjusting precode symbols based on data importance, maintaining efficiency.

Findings

PBPR improves error correction for high-importance data.

PBPR maintains similar decoding and encoding times as standard RaptorQ codes.

Consistent error correction improvements across various channel conditions.

Abstract

Raptor codes are the first class of fountain codes with linear time encoding and decoding. These codes are recommended in standards such as Third Generation Partnership Project (3GPP) and digital video broadcasting. RaptorQ codes are an extension to Raptor codes, having better coding efficiency and flexibility. Standard Raptor and RaptorQ codes are systematic with equal error protection of the data. However, in many applications such as MPEG transmission, there is a need for Unequal Error Protection (UEP): namely, some data symbols require higher error correction capabilities compared to others. We propose an approach that we call Priority Based Precode Ratio (PBPR) to achieve UEP for systematic RaptorQ and Raptor codes. Our UEP assumes that all symbols in a source block belong to the same importance class. The UEP is achieved by changing the number of precode symbols depending on the priority of the information symbols in the source block. PBPR provides UEP with the same number of decoding overhead symbols for source blocks with different importance classes. We demonstrate consistent improvements in the error correction capability of higher importance class compared to the lower importance class across the entire range of channel erasure probabilities. We also show that PBPR does not result in a significant increase in decoding and encoding times compared to the standard implementation.