Probabilistic Shaping Design Based on Reduced-Exponentiation Subset Indexing and Honeycomb-Structured Constellation Optimization for 5G Fronthaul Network

A probabilistic shaping method based on reduced-exponentiation subset indexing and honeycomb-structured constellation optimization is proposed to compress the number of signal points in the constellation, so that the total number doesn't fit the traditional pattern of multiples of power exponents of 2. The proposed scheme can significantly reduce the average signal power, enhance the space utilization in view of (identifying signals in) the judgement area of the constellation, as well as improve the mutual information, owing to the combination of probabilistic shaping and constellation optimization. Moreover, an experiment of a 25 km intensity-modulation and direct-detection system (IM/DD) transmission system is successfully demonstrated to present the superiority of our proposed scheme. It is shown that the proposed probabilistic shaping 64-to-31 carrier-less amplitude and phase modulation (CAP) based on honeycomb constellation can achieve the gains of 1.5 dB and 3 dB over receiver sensitivity when compared with uniform 32-CAP and uniform 64-CAP at the bit error rate (BER) of 1*10
^-3
, respectively. The experiment suggests the proposed scheme a promising technique for future 5G fronthaul network.