FPGA Implementation of Prefix-Free Code Distribution Matching for Probabilistic Constellation Shaping

Abstract

We implement rate-adaptable prefix-free code distribution matching in an FPGA, demonstrating its real-time feasibility with substantially less hardware resources than low-density parity-check coding.

Probabilistic Shaping for Finite Blocklengths: Distribution Matching and Sphere Shaping

“…Finally, we note that in this section, we restrict our attention to algorithmic implementation, while aspects related with hardware implementation such as throughput, clock frequency, and number of instances are outside the scope of this paper. For a detailed discussion on these concepts, shaping algorithms that are discussed in this work should actually be implemented as in [ 102 , 103 , 104 , 105 ].…”

Energy-efficient implementation of probabilistic shaping

“…Several DM techniques have been proposed, such as constantcomposition DM [6], enumerative sphere shaping [7], hierarchical DM (HiDM) [8], and prefix-free code DM [9]. Among these, only HiDM and prefixfree code DM have been implemented on a fieldprogrammable gate array (FPGA) [10][11][12][13][14][15].…”

FPGA Implementation of Hierarchical Subcarrier Rate and Distribution Matching for up to 1.032 Tb/s or 262144-QAM

“…However, there are only few reports detailing the DM implementation, which is still a challenging issue. Field programmable gate array (FPGA) implementations were reported in [5][6][7]. Real-time demonstrations of both DM encoding/decoding implementations were done in [5] only, and later verified at 400 Gb/s throughput in [8].…”

“…Real-time demonstrations of both DM encoding/decoding implementations were done in [5] only, and later verified at 400 Gb/s throughput in [8]. In contrast, only [6,7] implemented fine rate adaptation at 0.2 bit/channel use (bpcu) for 16-and 64-ary quadrature amplitude modulation (QAM). More precise rate adaptation and subcarrier rate matching are required in practical systems.…”