Reverse Viterbi Decoding for Unequal Error-Protected Frames
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing Serial List Viterbi Algorithms (SLVAs) are not optimally applied in communication systems for reconstructing frames with bits having different levels of error protection, leading to suboptimal performance in identifying the most probable sequences, especially when the search boundary does not extend beyond more important bits protected by additional error mechanisms.
Innovation Solution
Implementing a reverse Viterbi decoding followed by a serial list Viterbi algorithm with a boundary that focuses on the more important bits and CRC-protected portions, ensuring that only bit sequences differing within these protected areas are identified, thereby improving the detection of alternative paths that satisfy the CRC and reducing frame error rates.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If traditional SLVA is applied to frames with different bit protection levels, then the decoding process is simplified, but the frame error rate increases due to suboptimal identification of probable sequences
Solution Approach 1:
The frame is divided into two portions: a first portion with additional error protection (CRC-protected) and a second portion without additional error protection. The SLVA search boundary is positioned to separate these portions, allowing the algorithm to focus its computational effort on identifying alternative sequences in the protected portion while maintaining overall decoding simplicity.
Solution Approach 2:
Different decoding strategies are applied to different portions of the frame. The SLVA algorithm with optimized search boundary focuses computational resources on the first portion (bits 0-83) that has CRC protection, while the second portion (bits 84-167) is processed with standard Viterbi decoding. This local differentiation improves frame error rate performance without significantly increasing overall complexity.
2Reliability
If the SLVA search boundary is extended to include all bits, then all potential sequences are considered, but the computational complexity increases significantly
Solution Approach 1:
The SLVA search is extracted and limited to only the first portion of the frame (bits 0-83) that contains the CRC-protected important bits. By taking out the unnecessary computation on the second portion (bits 84-167), the algorithm achieves optimal sequence identification for the protected bits while significantly reducing computational complexity compared to searching the entire frame.
Solution Approach 2:
Instead of performing exhaustive search on all 168 bits, the algorithm performs partial action by limiting the SLVA search to the critical first 84 bits. This partial action is sufficient to identify alternative sequences that differ in the protected portion, which is all that is needed to improve frame error rate performance without the excessive computational cost of searching the entire frame.
3Reliability
If reverse Viterbi decoding is used, then alternative sequences differing in protected bits are identified more effectively, but the decoding algorithm becomes more complex
Solution Approach 1:
The patent employs reverse Viterbi decoding by processing the frame from the end (bit 167) backwards to the beginning (bit 0). This inversion of the decoding direction allows the algorithm to effectively identify alternative sequences that differ in the protected first portion (bits 0-83) by tracing back from the known good sequences in the unprotected second portion, improving alternative sequence detection while maintaining manageable algorithmic complexity.
Data Source
AI summary
A convolutionally encoded frame to be decoded includes a first portion of bits having additional error protection and another portion without additional error protection. The decoding of the frame involves reverse Viterbi decoding or Viterbi decoding on a reversed bit sequence followed by applying a serial list Viterbi algorithm to the first portion. The result is that the list of probable sequences have unique sets of bits in the first portion.