Adaptive Bayes-Adam MIMO Equalizer With High Accuracy and Fast Convergence for Orbital Angular Momentum Mode Division Multiplexed Transmission

In this article, a novel Bayes-Adam-based multiple-input multiple-output (MIMO) equalizer was proposed and experimentally demonstrated for an orbital angular momentum (OAM) mode-division multiplexed (MDM) optical fiber communication system. In general, MIMO equalization is required to compensate the crosstalk of the random intra-group mode coupling in the OAM mode. However, due to the time-varying characteristics of the OAM-MDM transmission, it is a long-standing challenge for MIMO equalization to achieve the high accuracy and fast convergence. In this work, a Bayes-Adam MIMO equalizer based on a penalty term and the Bayesian principle is developed for MDM transmission using eight OAM modes over a 20 km ring-core fiber, which smooths the cost function to achieve global optimization with fast convergence and high accuracy. Experiments show that the proposed adaptive Bayes-Adam MIMO equalizer outperforms two other conventional equalizers in terms of convergence speed and accuracy. The proposed approach achieved fast convergence with only 3,000 iterations for the <+4, left> OAM mode at an OSNR of 22 dB, which is much faster than that of Adam MIMO equalizer (11000 iterations) and SGD MIMO equalizer (16000 iterations). Furthermore, compared with a conventional Adam MIMO equalizer, the proposed scheme gains the OSNR improvement of 4 dB and 3 dB for OAM mode of l = <-4, right> and l = <-5, left> at the hard-decision forward error correction (HD-FEC) threshold, respectively, indicating its strong potential for OAM-MDM transmission.