Neural networks-based turbo equalization of a satellite communication channel

Abstract

This paper proposes neural networks-based turbo equalization (TEQ) applied to a non linear channel. Based on a Volterra model of the satellite non linear communication channel, we derive a soft input soft output (SISO) radial basis function (RBF) equalizer that can be used in an iterative equalization in order to improve the system performance. In particular, it is shown that the RBF-based TEQ is able to achieve its matched filter bound (MFB) within few iterations. The paper also proposes a blind implementation of the TEQ using a multilayer perceptron (MLP) as an adaptive model of the nonlinear channel. Asymptotic analysis as well as reduced complexity implementations are also presented and discussed.

A novel support vector machine robust model based electrical equaliser for coherent optical orthogonal frequency division multiplexing systems

“…Several equalisers such as classification tree, artificial neural network (ANN), K‐nearest neighbours, W‐H‐NLE model and Volterra‐NLE model are proposed to compensate dispersion, non‐linearity and polarisation channel effects in SMF [20–23].…”

Long Short-Term Memory Neural Equalizer

“…Some representative works [11], [12], [13] were using a Bayesian statistical model for channel equalization. Neural network-based approaches are used by some prior works [14], [15], [16], [9] for channels equalization. Some uncommonly used low-speed equalization technologies such as Support Vector Machine based equalizer [17], [18], Fuzzy based networks equalizer [19] were more difficult for hardware implementation and are not elaborated in detail here.…”

A survey of applied machine learning techniques for optical orthogonal frequency division multiplexing based networks

“…Deep neural networks have been applied to predict unknown image transmission at the end part of MMF [48]. On the other hand, several digital signal processing (DSP) equalizers including deterministic algorithms and machine learning such as artificial neural network (ANN) [49,50], K-nearest neighbors (KNN) [39], Wiener-Hammerstein nonlinear equalizer (WH-NLE) model [51], Volterra-NLE model [52] and classification tree (CT) [53] have been proposed to compensate dispersion, nonlinearity and polarization channel effects in both singlemode fiber (SMF) and MMF/FMF using also a dual-polarization scheme with an effective 2x2MIMO at the receiver. In long-haul transmission system, another important noise effect that be tackled only with machine learning is the stochastic parametric noise amplification (PNA) which is essentially the interplay between optical amplification and nonlinearity.…”

“…The most effective ML techniques in terms of performance but with a computational cost are the supervised artificial neural networks (ANNs) and support vector machines (SVMs). [40][41][42] Blind ML-based nonlinearity compensators have also shown great potential with reduce computational cost, using unsupervised ML clustering, such as fuzzy-logic C-means (FL), K-means, hierarchical clustering, 43 density-based spatial clustering of applications with noise (DBSCAN), 44 affinity propagation (AP), 45 and Gaussian mixture. 46 Both supervised and unsupervised ML algorithms have shown benefits over deterministic algorithms due to the fact they can tackle stochastic nonlinear distortions such as the interplay of polarization mode dispersion (PMD) with nonlinearity, transceiver nonlinearities and electro-optical noise.…”

“…55 On the other hand, both deterministic and ML-based DSP have been applied in both SMF and MMF/FMF based optical systems using either simplified 2 × 2 MIMO (dual-polarization, DP, in SMF) or sophisticated MIMO (FMF). Some examples are the following: ANN, 40,56 K-nearest neighbors (KNN), 39 Wiener-Hammerstein NLE (WH-NLE) model, 57 Volterra-NLE model, 58 and classification tree (CT). 59 The benefit of ML approaches for both SMF and MMF/FMF-based is their ability to tackle stochastic nonlinear phenomena.…”