Turbo Equalizer Buffer Management Using Codeword Discard Metrics
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Turbo equalization systems face buffer overflow issues due to fixed buffer sizes in hardware implementations, where one soft output decoder's high local iterations cause buffer overflow, preventing additional codewords from being stored.
Innovation Solution
Implementing a buffer management system that uses metrics such as unsatisfied parity checks and LLR magnitudes to select and preemptively discard codewords far from being successfully decoded, thereby managing buffer overflow by prioritizing codewords closest to completion for processing.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If one soft output decoder performs multiple local iterations on a codeword due to noise or errors, then decoding accuracy is improved, but buffer overflow occurs preventing additional codewords from being stored
Solution Approach 1:
The patent changes the parameter of buffer management by introducing dynamic prioritization based on codeword metrics. Instead of fixed buffer allocation, the system adjusts which codewords are retained or discarded based on their decoding progress indicators (such as number of unsatisfied parity-checks), allowing flexible adaptation when buffer space is constrained while maintaining decoding accuracy for critical codewords
Solution Approach 2:
The patent implements selective discarding of codewords from the buffer based on priority metrics. When buffer overflow is detected, codewords with lower priority (those farther from successful decoding) are discarded to make space for higher priority codewords. This selective discarding prevents buffer overflow while preserving the most critical decoding tasks that need buffer retention
2Device complexity
If fixed buffer size is used in hardware implementation, then device complexity is reduced, but buffer overflow prevents additional codewords from being processed
Solution Approach 1:
The patent introduces dynamic buffer management into a previously static fixed-buffer system. The buffer allocation and codeword retention decisions are made dynamically based on real-time metrics such as decoding progress and buffer occupancy levels. This dynamic approach allows the system to adapt to varying workload conditions without changing the physical buffer size, thereby maintaining hardware simplicity while improving processing throughput
Solution Approach 2:
The patent implements feedback mechanisms that monitor buffer occupancy and codeword decoding status. Based on this feedback, the system adjusts which codewords are retained or discarded from the buffer. This feedback-driven approach enables the fixed buffer to operate more efficiently by making informed decisions about resource allocation, thereby preventing overflow and maintaining productivity
3Reliability
If codewords are prioritized based on proximity to successful decoding, then buffer overflow is prevented, but processing time for slower decoders is reduced
Solution Approach 1:
The patent applies preliminary action by evaluating codeword metrics and determining priority levels before buffer overflow occurs. Codewords are assessed based on indicators such as the number of unsatisfied parity-checks, and priority assignments are made in advance. This preliminary prioritization allows the system to proactively manage buffer space and prevent overflow before it happens, rather than reacting after overflow occurs
Solution Approach 2:
The patent segments the buffer management process by dividing codewords into priority groups based on their decoding status and proximity to successful decoding. This segmentation allows differential treatment of different codeword groups, with higher priority given to those closest to completion. By segmenting the workload in this way, the system can prevent buffer overflow while managing processing time through prioritized handling of critical codewords
Data Source
AI summary
A plurality of metrics associated with a plurality of partially decoded codewords is obtained. The plurality of partially decoded codewords has been processed at least once by a first soft output decoder and a second soft output decoder and the plurality of partially decoded codewords is stored in a memory. At least one of the plurality of partially decoded codewords is selected based at least in part on the plurality of metrics; the memory is instructed to vacate the at least one selected codeword.