Trellis Equalization for FTN Decoding Complexity
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Faster-Than-Nyquist (FTN) data transmission systems face challenges in managing Inter-Symbol Interference (ISI) and channel frequency response, particularly in achieving efficient decoding of modulated symbols, as existing methods like BCJR are complex and unstable for varying channels.
Innovation Solution
A method for decoding modulated symbols using trellis equalization, which involves identifying a best path in a trellis, computing secondary trellises for neighboring constellation points, and calculating Log-Likelihood Ratios (LLRs) to provide to a forward error correction decoder, reducing complexity and instability while maintaining performance similar to benchmark methods.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If BCJR algorithm is used for decoding modulated symbols, then decoding accuracy is improved, but computational complexity increases
Solution Approach 1:
The patent segments the complex BCJR decoding process into two distinct passes: a first pass that identifies the best path and hard decision constellation point, and a second pass that computes LLRs for selected neighboring constellation points. This segmentation divides the computational workload and allows for optimized processing at each stage, reducing overall complexity while maintaining accuracy.
Solution Approach 2:
The patent extracts and processes only the most relevant constellation points (neighboring points that differ by one bit from the hard decision point) rather than processing all possible constellation points. This selective extraction reduces the number of computations required in the second pass while focusing computational resources on the most impactful points for accurate LLR calculation.
2Reliability
If traditional trellis equalization is used, then channel frequency response compensation is achieved, but stability varies across different channels
Solution Approach 1:
The patent performs preliminary identification of the best path and hard decision constellation point in the first pass before computing LLRs in the second pass. This preliminary action establishes a stable reference framework that adapts to different channel conditions, allowing the system to maintain stability across varying channels by first determining the optimal path specific to each channel's frequency response.
Solution Approach 2:
The patent implements a dynamic two-pass algorithm that adapts to different channel conditions by identifying channel-specific best paths and hard decision points in the first pass, then using these adaptive references for LLR computation in the second pass. This dynamic approach allows the system to adjust to varying channel characteristics while maintaining a consistent processing framework, achieving both stability and adaptability.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
There is provided an apparatus for decoding of modulated symbols or compensating channel frequency response, comprising: a processor configured to: identify a best path for a certain trellis stage of a main trellis of a received modulated symbol; identify a hard decision constellation point of the main trellis by performing a traceback according to the best path; select neighboring constellation points of the hard decision constellation point; compute a respective secondary trellis for each selected neighboring constellation points that differ from the hard decision constellation point by one bit; compute a least likelihood ratio, LLR, for each respective bit of the received modulated symbol by calculating using each respective secondary trellis, the difference between the best path metric of the received modulated symbol and the respective secondary trellis of the respective bit; and provide, the computed LLR to a forward error correction, FEC, decoder for decoding the received modulated symbol.