Abstract
We present a digital signal processing (DSP) scheme that performs hyperparameter tuning (HT) via Bayesian optimization (BO) to autonomously optimize memory tap distribution of Volterra series and adapt parameters used in the synthetization of a digital pre-distortion (DPD) filter for optical transmitters. Besides providing a time-efficient technique, this work demonstrates that the self-adaptation of DPD hyperparameters to correct the component-induced nonlinear distortions as different driver amplifier (DA) gains, symbol rates and modulation formats are used, leads to an improvement in transmitter performance. The scheme has been validated in back-to-back (b2b) experiments using dual-polarization (DP) 64 and 256 quadrature amplitude modulation (QAM) formats, and symbol rates of 64 and 80 GBd. For DP-64QAM at 64 GBd, it is shown that the DPD scheme reduces the required optical signal-to-noise ratio (OSNR) at a bit error ratio of 10 <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">-2</sup> by 0.9 dB and 0.6 dB with respect to linear DPD and a heuristic nonlinear DPD approach, respectively. Moreover, we show that the proposed approach also reduces filter complexity by 75% in conjunction with the use of memory polynomials (MP), while achieving a similar performance to Volterra pre-distortion filters.
Highlights
W ITH the increase of demand in bandwidth flexibility [1] in optical networks, there has been an emerging need for efficient resource management tools in order to address the diversity of capacity and reach demands required by users and services
These devices perform the self-adaptation of transmission parameters (e.g., modulation format, symbol rate, forward error correction (FEC) coding schemes) to automatically correct variable channel conditions originated from fiber and, especially, transceiver impairments (e.g., in-phase and quadrature (IQ) skew, IQ imbalance [4] and nonlinearities [5])
In sub-section B, we introduce the testbed and procedures used in our experimental validations and propose a methodology for the autonomous operation of the digital pre-distortion (DPD) scheme
Summary
ML-based algorithms have shown great compatibility to solve standard optical communication problems, while reducing complexity of traditional approaches [12] In this regard, we envision strong similarities between the optimization of the orders and memory taps of a Volterra filter for DPD of optical transmitters and a design problem often coped within ML applications, so-called hyperparameter tuning (HT) [13]. This work is an extension of our previous contribution [15], in which we show that this approach enables autonomous identification and mitigation of transmitter impairments, helps reduce filter complexity and improves system level performance This studr ydeexetpeenndisntghethreesleuvltesl reported in [15] by: of discussion on the technical details r of BO, presenting the computational gain that justifies the use of r the proposed approach, experimentally assessing Bayesian-based SI and Bayesianbased ILA under different setup configurations and DPD filter design scenarios. We recurrently express the conditional probability density function of a random variable Y given the occurrence of X as p(Y |X)
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have