Skip to main navigation Skip to search Skip to main content

Low-complexity equalizer with a hybrid decision scheme for 50 Gb/s/λ PAM4-PON using a low-cost 10 G receiver

  • Xizi Tang
  • , You Wei Chen
  • , Rui Zhang
  • , Shuang Yao
  • , Qi Zhou
  • , Shuyi Shen
  • , Mengqi Guo
  • , Yaojun Qiao
  • , Gee Kung Chang
  • Beijing University of Posts and Telecommunications
  • Georgia Institute of Technology

Research output: Contribution to journalArticlepeer-review

6 Scopus citations

Abstract

In this Letter, we experimentally demonstrate a 50 Gb/s/λ four-level pulse amplitude modulation-based passive optical network system with a 10 G class receiver. A memory polynomial equalizer (MPE) combined with a decision feedback equalizer (DFE) is applied to eliminate channel distortions in the system. To further improve the performance of the MPE-DFE, for the first time, to the best of our knowledge, a low-complexity hybrid decision scheme (HDS) is proposed, which consists of single-symbol decision (SSD) and multi-symbol decision (MSD). The SSD is exactly the conventional hard decision based on minimum Euclidean distance, whereas MSD is based on a simplified maximum likelihood detection principle with M-algorithm. In terms of complexity, MSD requires 19.1% more multiplications than SSD, but the symbol number of MSD only accounts for less than 20% of the total signal frame when the received optical power is greater than −27 dBm. Experimental results show that the proposed MPE-DFE with HDS achieves a 0.7 dB and 1.3 dB sensitivity gain compared with conventional SSD, and up to 35.4 dB and 31.4 dB link power budget, regarding the forward error correction threshold of 10−2 and 10−3, respectively.

Original languageEnglish
Pages (from-to)6278-6281
Number of pages4
JournalOptics Letters
Volume45
Issue number22
DOIs
StatePublished - Nov 15 2020

Fingerprint

Dive into the research topics of 'Low-complexity equalizer with a hybrid decision scheme for 50 Gb/s/λ PAM4-PON using a low-cost 10 G receiver'. Together they form a unique fingerprint.

Cite this