LDPC Check-Node Message Aggregation for Lower-Memory Decoding
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing error correction code (ECC) decoders, such as those using Low Density Parity Check (LDPC) codes, face inefficiencies in memory usage, computational complexity, and power consumption due to the need to store and update all messages received from variable nodes during iterative decoding processes.
Innovation Solution
Implementing a method where check nodes in the decoder store only aggregated information about the messages, such as existence flags and message counts, to efficiently calculate and update messages for transmission to variable nodes, reducing memory and computational requirements.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If all messages received from variable nodes are stored during iterative decoding, then accurate ECC decoding performance is maintained, but memory usage increases
Solution Approach 1:
The patent extracts only the essential information from the full message set - specifically the minimum absolute value and its corresponding variable node index - while discarding redundant message data. This extraction principle allows the decoder to maintain accurate decoding performance by preserving the critical minimum value needed for check node computations, while dramatically reducing memory requirements by not storing all individual messages.
Solution Approach 2:
The patent applies local quality by maintaining different levels of detail for different pieces of information. Instead of uniformly storing all messages with equal detail, the system maintains high precision for the minimum absolute value (which is critical for decoding accuracy) while using compact representations (sign bit and index) for identifying the source variable node. This differentiated storage strategy optimizes both memory efficiency and decoding performance.
2Reliability
If all messages are stored and updated in each iteration, then decoding accuracy is maintained, but computational complexity increases
Solution Approach 1:
The patent extracts only the necessary computational elements - the minimum absolute value and its index - from the complete message set. By focusing computation on these extracted elements rather than processing all messages, the system maintains decoding accuracy while reducing the computational burden of message updates and comparisons in each iteration.
Solution Approach 2:
The patent applies partial action by performing computations only on the essential subset of message data (the minimum value and its index) rather than on the complete set of all messages. This partial processing approach is sufficient to maintain decoding accuracy while significantly reducing the number of computational operations required per iteration.
3Measurement precision
If complete message sets are processed, then decoding precision is maintained, but power consumption increases
Solution Approach 1:
The patent extracts and processes only the critical minimum absolute value and its index from the full message set. This extraction reduces the volume of data that must be read from memory, processed, and written back, thereby reducing the dynamic power consumption associated with these operations while preserving the precision needed for accurate ECC decoding.
Solution Approach 2:
The patent performs partial processing by computing and updating only the essential minimum value information rather than processing the entire message set. This reduced computational scope directly translates to lower power consumption for arithmetic operations, memory access, and data transmission, while maintaining sufficient precision for reliable decoding.
Data Source
AI summary
A method includes, in a decoder of an Error Correction Code (ECC), maintaining only aggregated information regarding a set of messages, a function of which is to be reported from a first node to a second node of the decoder. The function of the set is determined and reported using the aggregated information. After reporting the function, one of the messages in the set is replaced with a new message. The aggregated information is updated to reflect the set having the new message, and the function of the set having the new message is determined and reported using the updated aggregated information.