To achieve high throughput, parallel decoding of low density parity check (LDPC) codes is required, but needs a large set of registers and complex interconnection due to randomly located 1's in a sparse parity check matrix of large block size. This paper proposes a memory-based decoding architecture for low density parity check codes using loosely coupled two data flows. Instead of register, intermediate values are optimally grouped and scheduled to store into the segmented memory, which reduces large area and enables a scalable architecture. The performance of the proposed decoder architecture is demonstrated by implementing a 1024 bit, rate-1/2 LDPC codes decoder.
展开▼