Memory Subsystem Polar Coding Using Metadata Frozen Bits
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional memory sub-systems face inefficiencies in error correction due to limited frozen bits, leading to weaker error correcting codes and increased execution time during decoding processes, which can result in unnecessary resource usage and decreased performance.
Innovation Solution
Implementing metadata-assisted encoding and decoding using a polar encoder and decoder, where system metadata supplements frozen bits to generate a more robust error correction code, allowing for a more efficient decode process by eliminating invalid candidates based on known metadata and check codes.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If conventional encoding with limited frozen bits is used, then device complexity is reduced, but error correction reliability deteriorates
Solution Approach 1:
The patent applies preliminary action by pre-determining fixed bit positions (frozen bits) in the polar code structure before actual data encoding. System metadata is identified and placed in these predetermined positions in advance, allowing the decoder to use this known information to guide the decoding process and eliminate invalid candidates early, thereby improving error correction reliability without proportionally increasing complexity
Solution Approach 2:
The patent changes the parameter of frozen bit proportion by increasing the number of frozen bits from conventional levels to 50% or more of the total codeword bits. This parameter change fundamentally improves error correction capability by providing more known reference points for the decoder, while the structured approach using system metadata prevents exponential complexity increase
2Reliability
If more frozen bits are used to improve error correction, then reliability improves, but execution time during decoding increases
Solution Approach 1:
The patent implements feedback by using system metadata (known frozen bits) to continuously guide and refine the decoding process. The decoder uses these known bit values to eliminate invalid candidates from the solution space, providing feedback that narrows down possible codewords and significantly reduces execution time despite the increased number of frozen bits
Solution Approach 2:
By pre-identifying system metadata and placing it in frozen bit positions before encoding, the patent enables the decoder to start with known information already in place. This preliminary action allows for faster candidate elimination during decoding, offsetting the time cost of processing more frozen bits
3Productivity
If conventional decoding without metadata assistance is used, then device complexity is lower, but productivity decreases due to increased execution time
Solution Approach 1:
The patent introduces system metadata as an intermediary element that mediates between the encoded data and the decoding process. This metadata acts as a guide or key that helps the decoder efficiently navigate the solution space, improving productivity by enabling faster candidate elimination while adding manageable complexity through the metadata management structure
Solution Approach 2:
The decoder is prepared in advance with knowledge of frozen bit positions and their expected values from system metadata. This preliminary preparation allows the decoding algorithm to operate more efficiently by immediately eliminating invalid candidates, improving productivity without requiring complex runtime adjustments
Data Source
AI summary
Data to be stored at a memory sub-system can be received from a host system. A portion of the host data that includes user data and another portion of the host data that includes system metadata can be determined. A mapping for a data structure can be received that identifies locations of the data structure that are fixed with respect to an encoding operation and locations of the data structure that are not fixed with respect to the encoding operation. The data structure can be generated for the user data and system metadata based on the mapping, and an encoding operation can be performed on the data structure to generate a codeword.