Layered LDPC Decoder Parallelization for Storage Conflict Reduction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing LDPC decoders face challenges in achieving high decoding efficiency due to storage access conflicts during iterative decoding, leading to high decoding delay and low throughput, which hinders their ability to meet the requirements of low delay and high throughput in 5G mobile communication systems.
Innovation Solution
A decoder architecture that acquires and processes soft bit information in advance by determining variable node and check node information in subsequent layers before completing the current layer's computation, allowing for parallel processing and reducing dependency on previous layer results, thereby speeding up the decoding process.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If the standard layered-decoding method is used, then the decoding algorithm is reliable and guarantees BER performance, but storage access conflict occurs in the iterative decoding process, leading to high decoding delay and low throughput rate
Solution Approach 1:
The patent applies preliminary action by pre-calculating and storing check node information for subsequent layers before the current layer's computation is complete. The computing unit determines check node information of a (k+1)th layer in advance, before finishing the current kth layer, thereby preparing decoding data ahead of time to avoid storage access conflicts and reduce decoding delay while maintaining reliability
Solution Approach 2:
The patent transforms the traditional sequential layer-by-layer decoding into a multi-dimensional parallel processing approach. By introducing the layer dimension k and enabling computations across different layers simultaneously, the system breaks the sequential dependency constraint and achieves parallel processing, increasing throughput rate while maintaining the reliability guaranteed by the layered structure
2Reliability
If the standard layered-decoding method is used, then the decoding algorithm is reliable and guarantees BER performance, but storage access conflict occurs leading to high decoding delay
Solution Approach 1:
The computing unit performs preliminary computation by determining check node information of the (k+1)th layer in advance, before the current kth layer computation is complete. This ahead-of-time preparation eliminates waiting time during iterative decoding and reduces overall decoding delay while preserving the reliable BER performance through the maintained layered decoding structure
3Productivity
If parallel processing is enabled by determining subsequent layer information in advance, then throughput rate and decoding efficiency are improved, but device complexity increases
Solution Approach 1:
The patent segments the decoding process into distinct computational stages: acquiring soft bit information for the current layer, determining check node information for the current layer, and pre-determining check node information for subsequent layers. This segmentation allows parallel execution of independent stages while managing complexity through structured organization of the computing unit's operations
Data Source
Figure 1~2
Figure 3~4
Figure 5~6
AI summary
Disclosed are a decoder, a decoding method and a computer storage medium. The decoder includes least one computing unit configured to: acquire soft bit information to be decoded of a variable node n in a basic matrix; determine whether an amount of decoding iterations i is less than a preset decoding iteration threshold; if i is less than the preset decoding iteration threshold, determine whether an amount of decoding layers k is less than an amount of rows a of the basic matrix; if k is less than a, determine soft bit information in a kth layer according to the soft bit information in a (k-1)th layer, and determine variable node information in a (k+1)th layer according to the soft bit information of the variable node n in the (k-1)th layer; determine check node information in the (k+1)th layer according to the variable node information in the (k+1)th update k to k+1, and re-determine whether k is less than a; and in response to k being greater than or equal to a, update i to i+1, and re-determine whether i is less than the preset decoding iteration threshold, until i is equal to the decoding iteration threshold. Also disclosed are a decoding method and a computer storage medium.