Soft Decoder Reliability Memory for Quantized Channel LLRs
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Soft decoding of signals from quantized channels faces challenges in achieving both high fidelity and fast decoding speeds, as conventional methods rely on iterative and time-consuming calculations of log likelihood ratios (LLRs) without known distributions, leading to inefficient use of power and lower fidelity.
Innovation Solution
An apparatus and method that utilize a reliability memory to store and selectively provide pre-computed or updated LLRs to a soft decoder, allowing repetitive attempts with different reliability measures until successful decoding, and updating the memory with new LLRs for improved decoding efficiency.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional iterative LLR calculation methods are used, then decoding fidelity can be improved, but decoding speed deteriorates
Solution Approach 1:
The patent pre-calculates and stores LLR values in a lookup table before decoding operations. During actual decoding, the system retrieves pre-computed LLR values from the table rather than calculating them iteratively, thus maintaining high decoding fidelity while dramatically improving decoding speed.
Solution Approach 2:
The patent transforms the continuous LLR calculation problem into a discrete lookup operation by quantizing the LLR values and storing them in a table. This parameter transformation allows the system to trade off some calculation precision for significant speed improvement while maintaining acceptable decoding fidelity.
2Measurement precision
If iterative LLR calculation is performed from scratch, then decoding accuracy can be achieved, but power consumption increases
Solution Approach 1:
The patent performs LLR calculations in advance and stores the results in a lookup table. During decoding operations, the system retrieves pre-computed values instead of performing iterative calculations, thereby maintaining decoding accuracy while significantly reducing power consumption during active decoding.
Solution Approach 2:
The system uses the stored LLR values from the lookup table to serve decoding operations without requiring repeated computational efforts. The pre-computed data serves the decoding process directly, eliminating the need for continuous high-power iterative calculations.
3Reliability
If conventional trial-and-error LLR selection is used, then convergence can be achieved, but time consumption increases
Solution Approach 1:
The patent pre-computes LLR values and stores them in a structured lookup table organized by threshold voltage distributions. During decoding, the system directly retrieves appropriate LLR values based on the observed signal characteristics without requiring iterative trial-and-error attempts, thus ensuring convergence while minimizing time consumption.
Solution Approach 2:
The patent transforms the iterative parameter selection process into a direct table lookup operation by organizing LLR values according to threshold voltage distributions. This parameter reorganization allows the system to quickly identify and retrieve the appropriate LLR values without time-consuming iterative searches.
Data Source
AI summary
Systems, methods, and other embodiments associated with soft decoding for a quantized channel are described. According to one embodiment, an apparatus includes a soft decoder configured to decode a signal received from a quantized channel based, at least in part, on one or more log likelihood ratios (LLRs). The apparatus may also include a reliability memory configured to store one or more known LLRs, and a controller configured to repetitively and selectively provide the soft decoder with known LLRs chosen from the reliability memory, to control the soft decoder to decode the signal, and to selectively update the reliability memory upon determining that the soft decoder successfully decoded the signal.