Generalized Tree Architecture for Efficient Successive-Cancellation Polar Decoding
Hye-Yeon Yoon, Tae‐Hwan Kim · 2018
This study presents an efficient architecture for the successive-cancellation polar decoder. The proposed architecture is designed based on the conventional tree architecture with the multibit decoding and the overlapped scheduling, but the architecture is generalized considering the high-radix processing, wherein several successive radix-2 kernels are processed within a cycle. The complexity and the throughput are formulated for the generalized tree architecture based on the high-radix processing. A case study is conducted for designing a 1024-bit SC decoder. When the design parameters are set to maximize the figure of merit (FOM), which is defined as the (throughput/complexity), the FOM of the 1024-bit SC decoder designed based on the proposed architecture is estimated to be 2 times higher than that of the SC decoder designed based on the conventional tree architecture.