2.4 2.

Article

Going Deeper into OSNR

Estimation with CNN

Fangqi Shen, Jing Zhou, Zhiping Huang and Longqing Li

https://doi.org/10.3390/photonics8090402
 hv
photonics

Article
Going Deeper into OSNR Estimation with CNN
Fangqi Shen , Jing Zhou , Zhiping Huang and Longqing Li *

College of Intelligent Science, National University of Defense Technology, Changsha 410073, China;
shenfangqi15@nudt.edu.cn (F.S.); zhou.jing@nudt.edu.cn (J.Z.); kdhuangzp@163.com (Z.H.)
* Correspondence: longqing_li@nudt.edu.cn

Abstract: As optical performance monitoring (OPM) requires accurate and robust solutions to tackle
the increasing dynamic and complicated optical network architectures, we experimentally demon-
strate an end-to-end optical signal-to-noise (OSNR) estimation method based on the convolutional
neural network (CNN), named OptInception. The design principles of the proposed scheme are
specified. The idea behind the combination of the Inception module and finite impulse response
(FIR) filter is elaborated as well. We experimentally evaluate the mean absolute error (MAE) and
root-mean-squared error (RMSE) of the OSNR monitored in PDM-QPSK and PDM-16QAM signals
under various symbol rates. The results suggest that the MAE reaches as low as 0.125 dB and RMSE
is 0.246 dB in general. OptInception is also proved to be insensitive to the symbol rate, modulation
format, and chromatic dispersion. The investigation of kernels in CNN indicates that the proposed
scheme helps convolutional layers learn much more than a lowpass filter or bandpass filter. Finally, a
comparison in performance and complexity presents the advantages of OptInception.

Keywords: OSNR; optical performance monitor; convolutional neural network



1. Introduction


Fiber-optic communication has experienced incredible advances in recent years for
Citation: Shen, F.; Zhou, J.; Huang, higher capacity. Advanced optical modulation formats, pulse-shaping techniques, along
Z.; Li, L. Going Deeper into OSNR with multiplexing techniques contribute to higher spectral efficiency [1]. Moreover, re-
Estimation with CNN. Photonics 2021, configurable optical add-drop multiplexers (ROADMs) bring flexible bandwidth assign-
8, 402. https://doi.org/10.3390/
ment [2]. Nevertheless, the promotion of the optical network in dynamicity, flexibility,
photonics8090402
and better utilization of available transmission capacity comes at the price of a more
complicated communication system with more noise sources introduced by new kinds of
Received: 17 August 2021
hardware units, where the communication link becomes path-dependent and dynamic
Accepted: 17 September 2021
due to the advent of ROADMs [3]. Therefore, transmission in fibers is more prone to be
Published: 20 September 2021
degraded, and real-time, comprehensive, and precise monitoring, referred to as optical
performance monitoring (OPM), on the condition of optical networks is currently being
Publisher’s Note: MDPI stays neutral
urged.
with regard to jurisdictional claims in
published maps and institutional affil-
OPM is considered as a key technology for elastic optical networks (EON) [4]. OSNR
iations.
directly reflects the quality of communication links in fiber by quantifying noise, espe-
cially amplified spontaneous emission (ASE) noises, added into the optical signals. A
direct relation between OSNR and bit-error rate (BER) [5] makes it become one of the
most important parameters for the evaluation of the general health of links and fault
diagnosis. The traditional measurement of OSNR typically relies on the optical spectrum
Copyright: © 2021 by the authors.
analyzer (OSA) that needs to get access to optical fibers for out-of-signal band noise power
Licensee MDPI, Basel, Switzerland.
measurement [6]. However, optical add-drop multiplexers (OADM) in wavelength di-
This article is an open access article
distributed under the terms and
vision multiplexing (WDM) networks filter major ASE noises out of band, which makes
conditions of the Creative Commons
traditional measurements fail [2]. Thus, many in-band measurement techniques are being
Attribution (CC BY) license (https://
put forward, such as spectral analysis after frequency down-conversion [7], polarization
creativecommons.org/licenses/by/ extinction [8], and usage of various interferometers [9,10]. The general drawback of these
4.0/). methods is the limitation on adaptation to dispersion impairment or different symbol

Photonics 2021, 8, 402. https://doi.org/10.3390/photonics8090402 https://www.mdpi.com/journal/photonics

Photonics 2021, 8, 402 2 of 16

rates. Moreover, the Refs. [11–14] estimated carrier-to-noise ratios (CNR) or signal-to-
noise ratios (SNR) to calculate OSNR based on statistical moments of equalized signals.
In [15], the utilization of empirical cumulative distribution function (CDF) of the signal’s
amplitude overcame the degradation for signals employing multilevel constellations in
moment-based OSNR estimation approaches mentioned above. In essence, this algorithm
still estimates the SNR using Newton’s method to fit theoretical CDF to the empirical one.
However, the requirement of equalization on signals leads to these methods being de-
signed for the digital coherent receiver. The cost of receiving and equalization makes these
methods non-economic and even impossible in the intermediate network nodes. Besides,
conventional techniques have limitations in simultaneous and independent monitoring of
multiple transmission impairments, of which the effects are often physically inseparable [4].
Fortunately, machine learning (ML) becomes a promising alternative for realizing highly
impairment-tolerant and adaptive monitoring with easy deployment.
The emergence of deep learning made machine learning flourish in pattern recognition,
speech recognition, and natural language processing. End-to-end deep learning has been
used in waveform-to-symbol decoding in coherent fiber-optic communication [16]. The
deep neural network (DNN) used in deep learning has proven to be effective in OSNR
estimation [3]. By and large, the trend of deep learning methods in OPM can be listed as
follows:
1. Transparency to symbol rate, modulation format and impairments;
2. Joint estimation of multiple parameters;
3. Independence of signal receiving;
4. Robust performance along with low complexity;
5. End-to-end learning.
The majority of monitoring methods utilizing deep learning can be concluded as a
combination of a statistic diagram or a statistic of received signals and one kind of neural
network. The diagram can be an eye diagram [17], asynchronous delay-tap sampling
(ADTS) 2D histogram [18], or amplitude histogram (AH) [19,20], while the statistic can be
an error vector magnitude (EVM) [21] or Godard’s error [22]. The neural network can be
an artificial neural network (ANN), k-nearest neighbors (KNN), support vector machine
(SVM), or convolutional neural network (CNN). These methods have one common flaw
that they all need to do extra information extraction or feature extraction manually. The
Refs. [23–25] showed a CNN method that needs asynchronously sampled raw data from
analog-to-digital converters (ADC) as input. The study of [25] probes the trained kernels
in convolutional layers, which explains why CNN can process raw data directly.
CNN introduced in [26] develops a tradition model of pattern recognition with the
idea of the training feature extractor itself. Based on the idea of CNN, deep convolutional
neural networks (AlexNet) [27], networks using blocks (VGG) [28], network in network
(NiN) [29], and so forth were proposed in succession and outperformed their predecessors
on the ImageNet classification challenge. The trend of CNN has been increasing in size,
including the depth and width of the network, to improve the performance at the cost of a
surge in computation complexity. Overfitting simultaneously becomes another side effect.
The sparsely connected architecture inside networks is the fundamental way to solve both
problems mentioned above. However, the computation of infrastructures nowadays suffers
from low efficiency in numerical calculation on the non-uniform sparse matrix. Therefore,
the main idea of GoogLeNet in [30] focused on finding out how an optimal local sparse
structure in a convolutional vision network can be approximated and covered by readily
available dense components, and the inception module was also raised in this paper. Apart
from that, the structure of parallel branches itself is thought-provoking as well.
In this paper, we dig deeper into OSNR estimation with CNN to realize end-to-end
monitoring. A structure comprised of the Inception block, which is a critical component in
GoogLeNet, is designed and explicated. Section 2 presents a review of the definition of
OSNR and high-level consideration in CNN and Inception block. The ideas and principles
behind the essential components in neural networks helped us to design the proposed
Photonics 2021, 8, 402 3 of 16

scheme in Section 2.4. In Section 3, an experiment is set up to train and test the proposed
scheme. The results are analyzed together with the architecture itself. Section 4 discusses
the extent of training with the help of a learning curve. A comparison between the proposed
scheme with state-of-the-art schemes in performance and complexity is presented as well.
These come to a conclusion in Section 5.

2. Principle of Proposed Scheme

2.1. OSNR Measurement
The optical signal-to-noise ratio (OSNR) represents the transmission performance of
optical links and has a direct relationship with BER [19,31]. A sufficiently high OSNR is
an essential requirement to maintain communication within acceptable error limits. Thus,
OSNR measurement can practically promote automatic fault detection and diagnosis, as
well as in-service characterization of signal quality [32].
OSNR can be defined as the logarithmic ratio of the average optical signal power to
average optical noise power over a specific spectral bandwidth measured at the input of an
optical receiver photodiode [5].

Psig Psig+noise ( B0 ) − Pnoise ( B0 )

OSNR = 10 log = 10 log , (1)
Pnoise Pnoise ( B0 )

where Psig and Pnoise are the signal and noise power bound by the signal spectral bandwidth
B0 . In practice, signal power Psig , which is difficult to be measured directly due to being
obscured by noise (where the same goes for noises at signal wavelength), is calculated
by the difference between total power Psig+noise and noise power bound by the signal
bandwidth B0 . A noise equivalent bandwidth (NEB) filter with a bandwidth of Bm was
applied to a signal skirt to estimate the power of noise in B0 by interpolating the noise
power measured outside the bandwidth, see (3). The instrument, the optical spectrum
analyzer (OSA), calculates OSNR as (2), where the specific spectral bandwidth Br for signal
and noise measurement is referred to as OSA’s resolution bandwidth (RBW) [5].

Psig Bm
OSNR = 10 log + 10 log (2)
Pnoise Br

Pnoise (λ1 ) + Pnoise (λ2 )

Pnoise = (3)
2
Given the definition of the OSNR and OSA measurement method, separation of the
noise and signal is the key point of the algorithm. Besides, OSNR describes the relationship
between the second moment of two parts of the received optical signal, signal, and noise.
Consequently, the trained neural network should learn how to extract the signal part
from noise as well as a mapping from some characteristics of the second moment to the
ultimate OSNR value. The Ref. [33] reveals a convolutional layer which has the potential
of extraction while the mapping process can be realized in a fully connected network.

2.2. Design Principles of Basic CNN

Typical structure of CNN comprises convolutional layers for feature extraction and
pooling layers for downsampling, following which fully connected (FC) layers learn
nonlinear mapping from high-dimensional feature spaces to corresponding value spaces.
Despite different definitions of convolution in neural network and signal processing,
the similarity in the mathematical formula reveals that the kernel in 1-D CNN could be
considered as a finite impulse response (FIR) filter [25] with the taps already flipped around
before doing multiplication with the signal sequence in a convolution operation.

N −1
s [i ] = ∑ x [i − j ] ω [ j ] (4)
j =0
Photonics 2021, 8, 402 4 of 16

K −1 N −1
s [i ] = f ( ∑ ∑ xk [i + j]ωk [ j] + b[k]), (5)
k =0 j =0

where s[i ] is the signal output, x [i ] is the signal input, ω [ j] is the weight in the convolutional
kernel, b[k] is the bias, and k represents the channel number. In an investigation into the
CNN kernels in [25], where two convolutional layers were placed in serial at the front end,
the first trained convolutional layer performed a few types of bandpass filters to separate
the signal from the noise, while the second one worked as a low-pass filter to extract a DC
component for averaging values derived from the previous convolutional layer.
In pattern recognition, of which the aim is doing classification of the input, the pooling
layer plays a role of dimension reduction with retaining basic features of the image. The
pooling layer merges semantically similar features into one after computing the maximum
or average of the local patch of units in feature maps [34]. Meanwhile, the pooling operation
helps to make the representation approximately invariant to small translations of the
input [35]. Namely, pooling brings image features invariance in translation, rotation, and
scaling. The following example vividly shows the feature extraction function of pooling.
In Figure 1, three different kinds of pooling are applied to the Girl with a Pearl Earring
(an oil painting by Johannes Vermeer in 1665. The work has been in the collection of the
Mauritshuis in The Hague since 1902). Downsampling samples the last pixel in each patch
with a stride of 40 when max pooling and average pooling takes the patch’s maximum and
average value in every channel of the image, respectively. It is still quite easy to recognize
this famous girl even when the resolution drops to only 168 pixels from 268,800 pixels for
all three pooling operations.
The success of the pooling layer applied to image recognition benefits from the sparse
feature information in an image. A weak correlation between local patches also contributes
to it. Specifically, features scatter in different regions of the 2D plane of an image while
different regions conceal different features. However, signals in optical fiber have nothing to
do with the property of sparsity. During a limited period, OSNR stays relatively unchanged
and can be measured at any time of the signal. The definition of OSNR in Equation (1)
also suggests that the information of OSNR is consistent at every time point and actually
presents features in the frequency domain [5]. If pooling is applied in the raw signal at the
very beginning, we could not derive much useful information from the time domain. The
features concealed in the spectrum are damaged by pooling operation instead.

Figure 1. Pooling operation can extract main features useful for dimension reduction.
Photonics 2021, 8, 402 5 of 16

Multilayer feedforward networks with as little as one hidden layer are proven to have
the potential of being universal approximators to any desired degree of accuracy, provided
sufficient hidden units are available [36]. In many circumstances, a deeper network is more
efficient to get a lower generalization error with less number of units required than a wider
one when representing the desired function [35].

2.3. Inception Architecture

Given that sparsely connected architecture avoids overfitting without utilizing high
computation efficiency on dense matrix multiplication, the Inception module aims at
improving hardware computation speed when keeping the sparse structure of a network.
The Inception block is a combination of variously sized kernels by concatenating several
parallel branches of convolutional layers. An example structure is depicted in Figure 2.

Input

5×5 Conv 3×3 Conv MaxPooling

1×1 Conv 1×1 Conv 1×1 Conv 1×1 Conv

Filter
concat

Output

Figure 2. Structure of the Inception block.

In addition, the Ref. [30] introduces 1 × 1 convolution for dimension reduction to

alleviate the computation complexity of expensive 3 × 3 and 5 × 5 convolution. The
following ReLU also provides a model with extra nonlinearity.
After various modifications made on Inception-v1 in [37], the latest work of [38]
incorporates residual connections, which is argued to be necessary for deep networks
to improve training speed empirically [39]. Furthermore, though Inception-v4 without
residual connections can achieve similar performance with similar expensive Inception-
ResNet-v2, the Ref. [38] admits that residual connections do accelerate training significantly.
Secondly, residual connections are able to solve the degradation problem where training
errors increase with a deeper network, and accuracy gets saturated or even degrades
rapidly. The shortcut in ResNet endues the network with identity mapping, which is
proved to be the best choice [40].

2.4. Proposed Scheme: OptInception

Figure 3 shows the whole architecture of the proposed network, OptInception, and
its advanced version, OptInception-Pro, in this paper. We deepen the OptInception by
cascading multiple Inception-ResNet modules in OptInception-Pro, which is inspired
by Inception-v4. The input length of the network is 1 × 1024 for four channels, which
correspondingly represents sampled data from the optical field of in-phase and quadrature-
phase components in both horizontal and vertical polarization. The detailed structures of
Inception blocks and Reduction blocks are shown in Figures 4 and 5, respectively. In the
Figures, the sizes of data flowing in the networks and convolutional kernels are in the form
of Channels@1 × Length of data(kernel) in every channel. The same padding is the default
strategy, while ’Valid’ is marked in the convolutional layer if valid padding is used. Layers
Photonics 2021, 8, 402 6 of 16

without an activation function are noted with ’Linear’ in parentheses. Concatenation of

branches in Inception blocks is named ’Filter concat’ for short.

Input Input

Inception-FIR Inception-FIR

Inception-ResNet-1
Inception-ResNet-1
×5

Reduction-1 Reduction-1

Inception-ResNet-2
Inception-ResNet-2
×5

Reduction-2 Reduction-2

Inception-ResNet-3
Inception-ResNet-3
×3

Average Pooling Average Pooling

Dropout Dropout

Fully-connected Fully-connected

Output Output

(a) OptInception (b) OptInception-Pro

Figure 3. Whole architecture of (a) OptInception and (b) OptInception-Pro.

In order to estimate OSNR with neural networks, the key is the acquisition of char-
acteristics of signal and noise power from raw data, as clarified above in (2). As clarified
in [25], convolutional layers fed with four channels of raw data asynchronously sampled
by ADC act as filters to extract various components of signals. Meanwhile, these filters in
convolutional layers can be automatically trained by a back-propagation algorithm without
any manual intervention. On this basis, longer convolutional kernels are applied for signal
processing with higher resolution when feature separation of signal and noise can be more
precise in an Inception-FIR block in a proposed scheme. Inspired by the structure of the
Inception block itself, different lengths of CNN kernels are deployed in parallel and then
concatenated by channels. Furthermore, it maintains integration of signals without the
negative part being cut off, meaning that ReLU or other nonlinear activation functions are
not applied to the convolution results. For max pooling in particular, pooling can lose a
lot of information, ending with a deteriorating feature extraction effect in the successive
Photonics 2021, 8, 402 7 of 16

network section, as explained in Section 2.2. Pooling layers which will not contribute to
the separation were not laid out in this block.

Input Input
4@1×1024 32@1×1024

32@1×1
Conv

32@1×1024

32@1×64 32@1×1
8@1×128 16@1×256 8@1×512
Conv Conv
Conv (Linear) Conv (Linear) Conv (Linear)
32@1×1024 32@1×1024
16@1×1024
8@1×1024 32@1×32 32@1×16 32@1×1
8@1×1024
Conv Conv Conv

32@1×1024
32@1×1024 32@1×1024
Filter concat
Filter concat

96@1×1024
32@1×1 32@1×1
Conv Conv (Linear) 32@1×1024

32@1×1024

Output
ReLU Activation
32@1×1024

(a) Inception-FIR (b) Inception-ResNet-1

Input Input
80@1×512 208@1×64

32@1×1
Conv

32@1×512 1×5 128@1×1

Average Pooling Conv
48@1×5 32@1×1
Conv Conv
208@1×64 128@1×64
48@1×512 32@1×512
64@1×1 72@1×3 120@1×1
64@1×3 64@1×3 64@1×1 Conv Conv Conv
Conv Conv Conv

64@1×512 64@1×64 72@1×64

64@1×512 120@1×64
64@1×512
Filter concat Filter concat

192@1×512 256@1×64

80@1×3 208@1×64
Conv (Linear) Conv (Linear) 208@1×64
80@1×512

80@1×512 208@1×64

ReLU Activation ReLU Activation

(c) Inception-ResNet-2 (d) Inception-ResNet-3

Figure 4. Structures of Inception blocks.

Next, residual connections were introduced into subsequent Inception blocks to

improve computation efficiency in deep network training. Considering that the network
we design is definitely deep with multiple Inception blocks and fully connected layers, this
practical structure is necessary and useful.
Following the Inception-ResNet blocks, reduction blocks named Reduction-1 and
Reduction-2 were designed to help aggregate features and decrease the grid sizes of the
feature maps. In the Reduction module, average pooling and 1 × 1 convolution were both
used. Only one of them should be used in a single branch of the Inception parallel structure
as usual.
Photonics 2021, 8, 402 8 of 16

Input Input
32@1×1024 80@1×512

64@1×1 32@1×1
Conv 32@1×513 1×513 Conv 64@1×8 1×8
Conv Average Pooling Conv Average Pooling
64@1×1024 Valid Valid 32@1×512 Stride=8 Valid Stride=8 Valid
16@1×513 64@1×8
Conv 32@1×512 64@1×64
Conv
Valid 32@1×512 Stride=8 Valid 80@1×64

16@1×512
64@1×64
Filter concat Filter concat

Output Output
80@1×512 208@1×64

(a) Reduction-1 (b) Reduction-2

Figure 5. Structures of Reduction blocks.

After three Inception-ResNet and two Reduction blocks, the trained feature maps
were sent to average pooling before being flattened. Four fully connected layers with two
hidden layers and without an activation function applied at the output of the network
were arranged in order to fit the mapping to OSNR that is a continuous value.
All in all, the general architecture of OptInception can be divided into three procedures.
In the beginning, the Inception-FIR is responsible for the reception of input data and feature
separation. Secondly, feature extraction is conducted in Inception-ResNet and Reduction
blocks. Finally, the features learned previously are mapped to a corresponding OSNR value
in FC networks.

3. Experimental Result and Analysis

3.1. Experimental Setup
Figure 6 demonstrates the experiment platform our OSNR monitoring runs on. On
the transceiver side, an external cavity laser (ECL) with less than 100 kHz linewidth gen-
erates a 194.1 THz carrier. PDM-QPSK and PDM-16QAM in 14 GBaud, 20 GBaud, and
28 GBaud Nyquist shaping optical signals with a 0.01 roll-off factor were generated in the
DP-I/Q-modulator driven by a four-channel DAC with a 64 GSa/s sampling rate and 8-bit
nominal resolution. Gain equalizers inserted in the link were aimed at making cascading
of amplifiers possible so that the signal can be transmitted over long distances without
distorting the envelope. Erbium-doped fiber amplifiers (EDFA) act as the amplified sponta-
neous emission (ASE) noise source in the experiment. Variable optical attenuators (VOA)
along with EDFAs control the power of signal and noise and thereby adjust the OSNR at
the receiver. A span of 80 km standard single-mode fibers (SSMF) with 16.75 ps/(km·nm)
average dispersion simulates the real link for different transmission distances in the recir-
culating loop for the sake of the acousto-optic switches (AOS). These optical switches are
integrated in the fiber-recirculating controller Ovlink IOM-601. After the transmission, the
signal is filtered by an optical bandpass filter (OBPF) before being sent into an integrated
coherent receiver. The signal is sampled at a rate of 50 GSa/s in an oscilloscope with
16 GHz bandwidth. Finally, the four-channel raw data go directly to the OSNR monitoring
network. The measurements of OSA are used as labels in the training process.
Photonics 2021, 8, 402 9 of 16

EDFA
VOA ITLA
50 Gsa/s
ICR
ADC
DP-I/Q Modulator
ECL EDFA Gain EQ EDFA OBPF
Ix Qx Iy Qy AOS

AOS Qy Iy Qx Ix
Gain EQ Coupler
4-channel DAC (64 Gsa/s)
OptInception
OSA
OSNR monitoring network
EDFA
VOA
SMF 80 km

Figure 6. Experimental setup for OSNR monitoring in proposed scheme. ECL: External Cavity
Laser. Gain EQ: Gain Equalizer. AOS: Acousto-optic Switch. VOA: Variable Optical Attenuator.
ITLA: Integrated Tunable Laser Assembly. OBPF: Optical Bandpass Filter; ICR: Integrated Coherent
Receiver; EDFA: Erbium-doped Fiber Amplifier.

3.2. Training Methodology

The number of data with a size of 4 × 1024 (channels × length) is 76,800. Within the
dataset, 25% was classed as the test dataset and 75% served as the training dataset. In order
to investigate monitoring performance under different chromatic dispersions (CD), we
made optical signals travel different distances over the recirculation loop. Part of these data
were included in the training dataset for generalization. We trained OptInception using the
TensorFlow library based on graphics processing units (GPU) [41]. The mean square error
was to be minimized after the forward propagation in every mini-batch training loop. The
update of weights depends on the Adam optimization algorithm [42], where the individual
adaptive learning rate is computed for different parameters and brings better training
performance, higher computation efficiency, and lower computation resource occupation.
In order to enhance generalization ability, batch normalization [43] was deployed inside
convolutional layers while the dropout was put before a fully connected network. Bayesian
optimization was used to find the best hyperparameters, such as learning rate and batch
size [44,45]. A combination of a batch size of 40 and learning rate of 0.0005 seemed to be
suitable in our training.

3.3. Results and Analysis

We trained OptInception and its variants in Figure 7. For variants with mini-Inception-
FIR, the convolutional kernels were truncated to the lengths of 64, 32, and 16, respectively.
The OptInception-AVG adds average pooling with a pooling size of eight in each branch
following the convolutional layer. OptInception-Leaky adds one more layer into the
hidden layer in the FC network and alter the ReLU function with the leaky-ReLU function.
Moreover, Inception-v4 was modified to fit in with the 1D input format of raw data from
ADC.
The 1D version of Inception-v4 and OptInception-AVG performed poorly, as shown in
Figure 7. Violently inputting our data into a popular CNN designed for pattern recognition
did not seem to work. The failure of OptInception-AVG proves that pooling will drop much
useful information at the beginning of the networks, as shown by the design principle
mentioned above. Pooling always plays a vital role in integrating similar features, but not
in an information extraction step. The rest learn better on the training dataset.
Next, the estimation accuracy of OSNR on the test dataset depicted in Figure 8 reveals
the generalization ability and feasibility in the reality of the networks. It is obvious that
the modifications in OptInception-Leaky are not successful for its rigorous fluctuation
and poorer and poorer generalization with training going on. In contrast, OptInception
improves continuously and can finally reach quite a high level of accuracy of estimation
in the test. The ones with mini-Inception-FIR do not deteriorate performance much, but
are not stable in the early stages of training. OptInception-Pro trades much steadier
learning improvement with more expensive learning prices. Metaphorically, OptInception
is more like an intelligent student who learns fast with less time and energy but who
Photonics 2021, 8, 402 10 of 16

also occasionally makes mistakes, while OptInception-Pro is a hardworking student who

polishes up their score patiently, step by step.

OptInception
OptInception with mini-Inception-FIR
101 OptInception-Pro
Loss (Mean Square Error) OptInception-Pro with mini-Inception-FIR
OptInception-AVG
OptInception-Leaky
Inception-v4 architecture 1-D version

100

0 2500 5000 7500 10,000 12,500 15,000 17,500

Steps

Figure 7. Training process for OptInception and its variants. The curves of training losses are
smoothed by an exponential weighted (EW) function and presented in a logarithmic scale.

OptInception
OptInception with mini-Inception-FIR
2.0 OptInception-Pro
Loss (Mean Absolute Error)

OptInception-Pro with mini-Inception-FIR

OptInception-Leaky

1.5

1.0

0.5

2 4 6 8 10 12
Epoch

Figure 8. Evaluation on the test dataset for OptInception and its variants.

The following monitoring results are estimated by OptInception-Pro to exploit the

potential of the proposed scheme, considering its steady learning process.
In Figure 9, the estimation of the proposed OptInception-Pro is generally precise.
However, degradation occurs with OSNR climbing because it becomes harder to measure
the exact noise power from raw data when OSNR is higher. Nevertheless, this phenomenon
does not affect its applications since it is more important to monitor the quality of transmis-
sion links in poor condition when the OSNR is always low in practice.
Figure 10 shows the test performance on different symbol rates and modulations, with
Figure 10a showing mean absolute error (MAE) and Figure 10b showing root-mean-square
error (RMSE). The MAE remains only about 0.125 dB when RMSE is 0.246 dB. The general
performance does not fluctuate much, so the proposed scheme is almost transparent to
modulations and symbol rates as long as the combinations are trained thoroughly.
The tolerance against chromatic dispersion is investigated in Figure 11 as well. The
number of loops is controlled to acquire different CD. 28 GBaud PDM-QPSK signals
when 20 dB OSNR are tested. The errors of estimation basically have a weak relation-
Photonics 2021, 8, 402 11 of 16

ship with chromatic dispersion. Thus, OptInception is also transparent to impairment of

chromatic dispersion.

30

Estimated OSNR (dB) 25

20

15

10
10 15 20 25 30
True OSNR (dB)
Figure 9. True OSNR vs. estimated OSNR by OptInception-Pro.

0.25 Root Mean Square Error (dB)

PDM-QPSK PDM-QPSK
Mean Absolute Error (dB)

0.20 PDM-16QAM 0.3 PDM-16QAM

0.15
0.2
0.10
0.1
0.05

0.00 0.0
14 GBaud 20 GBaud 28 GBaud 14 GBaud 20 GBaud 28 GBaud

(a) Mean absolute error (b) Root mean square error

Figure 10. Histogram of test results on different symbol rates and modulations using metrics of
(a) mean absolute error (MAE) and (b) root-mean-square error (RMSE).

0.4
Mean Absolute Error (MAE)
Estimation Error (dB)

Root Mean Square Error (RMSE)

0.3

0.2

0.1

0.0
0 10,000 20,000 30,000 40,000
Chromatic dispersion (ps/nm)
Figure 11. Estimated OSNR errors under different chromatic dispersions (CD) in a 28 GBaud PDM-
QPSK signal with 20 dB OSNR.
Photonics 2021, 8, 402 12 of 16

When it comes to the effect of the convolutional layers in monitoring, we probed the
weights in convolutional kernels again, as [25] did. For best demonstration, OptInception
with mini-Inception-FIR was selected. The fast Fourier transform (FFT) of the shortest
kernel with 1 × 16 size in the third branch was ultimately presented. The zero-frequency
was shifted to the center of the spectrum. In Figure 12, the OptInception learns more than
bandpass filters. Lowpass filters and highpass filters were also formed during the training
process. Some kernels selected various frequencies simultaneously. The variance of filters
brought out variable characteristics. The functions of many other filters, however, cannot
be explained intuitively. It is indeed true that the neural network is a veritable black box.

1 16 1 16

1 16 1 16

1 16 1 16

1 16 1 16
Figure 12. FFT of eight convolutional kernels in the third branch with a 1 × 16 size of a mini-
Inception-FIR module in OptInception. Four curves in every subplot represent four channels of one
convolutional kernel corresponding to four channels in raw input data.

4. Discussion
4.1. Learning Curve
The learning curves of OptInception are further investigated in this section to show the
extent of training. The term ’learning curve’ has different variables on the x-axis under the
contexts of an artificial neural network and general machine learning. The ANN literature
shows the performance on training and test datasets as a function of training iteration,
while general ML shows the predictive generalization performance as a function of the
number of training examples [46].
Considering the size of the training set is large enough, we investigate the training
error and generalization error as a function of the number of iterations in Figure 13. After
every epoch, the mean absolute error of prediction on the training dataset and test dataset
is evaluated. As expected, the MAE on the training dataset drops slowly at a later stage,
while the MAE on the test dataset fluctuates between 0.15 and 0.20 for a relatively long time.
Overfitting does not become tricky, profiting from the techniques used in the architecture
of OptInception, like ResNet, batch normalization, and dropout. The learning curve proves
that the model has been trained thoroughly.
Photonics 2021, 8, 402 13 of 16

2.0

Mean Absolute Error (MAE)

MAE on training dataset
MAE on test dataset
1.5

1.0

0.5

0.0
0 10 20 30 40 50
Epoch
Figure 13. The learning curve of OptInception as a function of the number of epochs. The dashed
lines show the learning curves after smoothing with an exponential weighted (EW) function. Mean
absolute error is the metric of performance.

4.2. Comparison
In this section, comparisons between our proposed architecture and the state-of-the-art
OSNR monitoring schemes are shown in the following with regard to performance and
complexity. We evaluate the performance with MAE and RMSE for accuracy and precision,
respectively. The complexity of an algorithm can be divided into time complexity and space
complexity. Time complexity, which is defined as the time of calculation on the algorithm,
can be quantitatively analyzed with floating-point operations (FLOPs). Space complexity
describes the memory occupation when the algorithm runs.
In [47], He investigated the relationship between the accuracy of CNNs and the
complexity. The paper calculated the total time complexity of all convolutional layers as (6).
d
O( ∑ nl −1 · s2l · nl · m2l ), (6)
l =1

where l is the index of a conventional layer, while l − 1 represents the index of the previous
one, and d is the number of conventional layers. nl is the number of convolutional kernels
of the l-th layer when nl −1 is regarded as the number of input channels of the l-th layer
received from the output of the previous layer. Considering the kernel and feature map
are both square, sl represents the spatial size of the kernel, and ml is the spatial size of the
output feature map.
When it comes to space complexity, the memory occupation always includes two
parts: the memory for weights, and the memory for output of every layer. For CNN, the
space complexity is proportional to (7).
d
O( ∑ (s2l · nl −1 · nl + m2l · nl )). (7)
l =1

For the FC layer, the time and space complexity is determined by the number of nodes
in output nl and that in input nl −1 . Thus, the FLOPs can be computed as O(2nl −1 nl ) and
the occupied memory is proportional to O(nl −1 · nl + nl ).
Therefore, a comparison is invited among the proposed OptInception scheme, CNN
monitoring scheme in [25], and the CDF estimation algorithm in [15], as Table 1 depicts.
All schemes were tested on our test dataset. The CNN model in [25] was trained with
the same training dataset of OptInception. The parameters in the CDF algorithm were
assigned with the same typical values in [15]. The parameters were stored in the data type
of 32-bit floating-point numbers, and the space complexity was presented in bytes.
Photonics 2021, 8, 402 14 of 16

Table 1 vividly shows that the conventional method based on CDF has much lower
computing complexity and memory occupation than schemes based on neural networks.
However, the cost of low complexity is the relatively low estimation performance and
the difficulty in deployment on devices without equalization. It makes sense that the
deeper model, OptInception, monitors OSNR more accurately and precisely with lower
values in MAE and RMSE than the scheme in [25] only using the basic CNN module.
Surprisingly, the complexity of OptInception is obviously less than the basic CNN scheme
in both time and space. In fact, when the node number in the FC layer grows, the number
of weights becomes considerable along with addition and multiplication operations. The
number of weights in FC layers of [25] accounts for 99.46%, but only 5.54% for Inception.
Thanks to the averaging pooling layer before the FC network, the widths of FC layers
shrink markedly. Last but not least, more than two-fold the space complexity and one-third
of the time complexity increment in OptInception-Pro brought almost no more than a
15% improvement in performance. This phenomenon suggests that the marginal utility is
diminishing as the depth of the network increases.

Table 1. The comparison of various state-of-the-art OSNR monitoring schemes with the proposed
scheme in performance and complexity.

Performance Complexity
Scheme
MAE (dB) RMSE (dB) FLOPs Params (Bytes)
OptInception 0.148 0.277 12,482,975 25,001,732
OptInception-Pro 0.125 0.246 16,833,798 56,378,116
Basic CNN scheme in [25] 0.357 0.449 28,135,393 33,751,492
CDF-based algorithm in [15] 0.478 0.525 71,264 -

5. Conclusions
In this paper, a high-performance neural network scheme, namely, OptInception, was
proposed for OSNR monitoring in OPM applications. We elaborated on the design of
the scheme by reviewing structures and functions we used. Additionally, their principles
or the ideas behind them decided whether and how we used them. An experiment was
set up to verify the transparency of OptInception to the symbol rate, modulation format,
and chromatic dispersion. The mean absolute error in the test dataset was approximately
0.125 dB, while the root-mean-square error was 0.246 dB. The kernels in trained the network
were also investigated to reveal the complexity of neural networks. Finally, a learning curve
was drawn to show the training extent of OptInception. A comparison in performance and
complexity presented the advantage of the proposed scheme. In a nutshell, the training
process and experimental results indicate that the design principles function as expected.

Author Contributions: Conceptualization, F.S.; methodology, F.S.; software, F.S.; validation, F.S., L.L.
and J.Z.; formal analysis, J.Z.; investigation, F.S., L.L.; resources, L.L.; data curation, F.S., L.L.; writing—
original draft preparation, F.S.; writing—review and editing, J.Z.; visualization, L.L.; supervision,
J.Z.; project administration, L.L.; funding acquisition, Z.H. All authors have read and agreed to the
published version of the manuscript.
Funding: This project was funded by the Natural Science Foundation of China under Grant
No. 51575517.
Institutional Review Board Statement: Not applicable.
Informed Consent Statement: Not applicable.
Data Availability Statement: Not applicable.
Conflicts of Interest: The authors declare that they have no known competing financial interest or
personal relationships that could have appeared to influence the work reported in this paper.
Photonics 2021, 8, 402 15 of 16

References
1. Ramaswami, R.; Sivarajan, K.; Sasaki, G. Optical Networks: A Practical Perspective; Morgan Kaufmann: Burlington, MA, USA, 2009.
2. Saif, W.S.; Esmail, M.A.; Ragheb, A.M.; Alshawi, T.A.; Alshebeili, S.A. Machine learning techniques for optical performance
monitoring and modulation format identification: A survey. IEEE Commun. Surv. Tutor. 2020, 22, 2839–2882. [CrossRef]
3. Dong, Z.; Khan, F.N.; Sui, Q.; Zhong, K.; Lu, C.; Lau, A.P.T. Optical Performance Monitoring: A Review of Current and Future
Technologies. J. Light. Technol. 2016, 34, 525–543. [CrossRef]
4. Khan, F.N.; Lu, C.; Lau, A.P.T. Optical Performance Monitoring in Fiber-Optic Networks Enabled by Machine Learning
Techniques. In Proceedings of the 2018 Optical Fiber Communications Conference and Exposition (OFC), San Diego, CA, USA,
11–15 March 2018.
5. Chomycz, B. Planning Fiber Optics Networks; McGraw-Hill Education: New York, NY, USA, 2009.
6. Suzuki, H.; Takachio, N. Optical signal quality monitor built into WDM linear repeaters using semiconductor arrayed waveguide
grating filter monolithically integrated with eight photodiodes. Electron. Lett. 1999, 35, 836–837. [CrossRef]
7. Baker-Meflah, L.; Savory, S.; Thomsen, B.; Mitchell, J.; Bayvel, P. In-band OSNR monitoring using spectral analysis after frequency
down-conversion. IEEE Photonics Technol. Lett. 2007, 19, 115–117. [CrossRef]
8. Rasztovits-Wiech, M.; Danner, M.; Leeb, W. Optical signal-to-noise ratio measurement in WDM networks using polarization
extinction. In Proceedings of the 24th European Conference on Optical Communication, ECOC’98 (IEEE Cat. No. 98TH8398),
Madrid, Spain, 20–24 September 1998; Volume 1, pp. 549–550.
9. Tao, Z.; Chen, Z.; Fu, L.; Wu, D.; Xu, A. Monitoring of OSNR by using a Mach–Zehnder interferometer. Microw. Opt. Technol. Lett.
2001, 30, 63–65. [CrossRef]
10. Qiu, J.; Huang, Z.; Yuan, B.; An, N.; Kong, D.; Wu, J. Multi-wavelength in-band OSNR monitor based on Lyot-Sagnac
interferometer. Opt. Express 2015, 23, 20257–20266. [CrossRef]
11. Faruk, M.S.; Mori, Y.; Kikuchi, K. Estimation of OSNR for Nyquist-WDM transmission systems using statistical moments of
equalized signals in digital coherent receivers. In Proceedings of the Optical Fiber Communication Conference, San Francisco,
CA, USA, 9–13 March 2014; p. Th2A.29.
12. Faruk, M.S.; Mori, Y.; Kikuchi, K. In-band estimation of optical signal-to-noise ratio from equalized signals in digital coherent
receivers. IEEE Photonics J. 2014, 6, 1–9. [CrossRef]
13. Ives, D.J.; Thomsen, B.C.; Maher, R.; Savory, S.J. Estimating OSNR of equalised QPSK signals. Opt. Express 2011, 19, B661–B666.
[CrossRef]
14. Zhu, C.; Tran, A.V.; Chen, S.; Du, L.B.; Do, C.C.; Anderson, T.; Lowery, A.J.; Skafidas, E. Statistical moments-based OSNR
monitoring for coherent optical systems. Opt. Express 2012, 20, 17711–17721. [CrossRef] [PubMed]
15. Lin, X.; Dobre, O.A.; Ngatched, T.M.N.; Li, C. A Non-Data-Aided OSNR Estimation Algorithm for Coherent Optical Fiber
Communication Systems Employing Multilevel Constellations. J. Light. Technol. 2019, 37, 3815–3825. [CrossRef]
16. Gaiarin, S.; Da Ros, F.; Jones, R.T.; Zibar, D. End-to-End Optimization of Coherent Optical Communications Over the Split-Step
Fourier Method Guided by the Nonlinear Fourier Transform Theory. J. Light. Technol. 2021, 39, 418–428. [CrossRef]
17. Wang, D.; Zhang, M.; Li, Z.; Li, J.; Fu, M.; Cui, Y.; Chen, X. Modulation Format Recognition and OSNR Estimation Using
CNN-Based Deep Learning. IEEE Photonics Technol. Lett. 2017, 29, 1667–1670. [CrossRef]
18. Wu, X.; Jargon, J.A.; Paraschis, L.; Willner, A.E. ANN-based optical performance monitoring of QPSK signals using parameters
derived from balanced-detected asynchronous diagrams. IEEE Photonics Technol. Lett. 2010, 23, 248–250. [CrossRef]
19. Khan, F.N.; Zhong, K.; Zhou, X.; Al-Arashi, W.H.; Yu, C.; Lu, C.; Lau, A.P.T. Joint OSNR monitoring and modulation format
identification in digital coherent receivers using deep neural networks. Opt. Express 2017, 25, 17767–17776. [CrossRef]
20. Wan, Z.; Yu, Z.; Shu, L.; Zhao, Y.; Zhang, H.; Xu, K. Intelligent optical performance monitor using multi-task learning based
artificial neural network. Opt. Express 2019, 27, 11281–11291. [CrossRef] [PubMed]
21. Li, J.; Wang, D.; Zhang, M. Low-complexity adaptive chromatic dispersion estimation scheme using machine learning for
coherent long-reach passive optical networks. IEEE Photonics J. 2019, 11, 1–11. [CrossRef]
22. Hao, M.; Yan, L.; Yi, A.; Jiang, L.; Pan, Y.; Pan, W.; Luo, B. OSNR Monitoring Using Support Vector Ordinal Regression for Digital
Coherent Receivers. IEEE Photonics J. 2019, 11, 1–11. [CrossRef]
23. Tanimura, T.; Hoshida, T.; Kato, T.; Watanabe, S.; Morikawa, H. Data-analytics-based optical performance monitoring technique
for optical transport networks. In Proceedings of the Optical Fiber Communication Conference, San Diego, CA, USA, 11–15
March 2018; p. Tu3E.3.
24. Tanimura, T.; Kato, T.; Watanabe, S.; Hoshida, T. Deep neural network based optical monitor providing self-confidence as
auxiliary output. In Proceedings of the 2018 European Conference on Optical Communication (ECOC), Rome, Italy, 23–27
September 2018; pp. 1–3. [CrossRef]
25. Tanimura, T.; Hoshida, T.; Kato, T.; Watanabe, S.; Morikawa, H. Convolutional Neural Network-Based Optical Performance
Monitoring for Optical Transport Networks. J. Opt. Commun. Netw. 2019, 11, A52–A59. [CrossRef]
26. Lecun, Y.; Bottou, L.; Bengio, Y.; Haffner, P. Gradient-based learning applied to document recognition. Proc. IEEE 1998,
86, 2278–2324. [CrossRef]
27. Krizhevsky, A.; Sutskever, I.; Hinton, G.E. Imagenet classification with deep convolutional neural networks. Adv. Neural Inf.
Process. Syst. 2012, 25, 1097–1105. [CrossRef]
Photonics 2021, 8, 402 16 of 16

28. Simonyan, K.; Zisserman, A. Very Deep Convolutional Networks for Large-Scale Image Recognition. In Proceedings of the
International Conference on Learning Representations, San Diego, CA, USA, 7–9 May 2015.
29. Lin, M.; Chen, Q.; Yan, S. Network in network. arXiv 2013, arXiv:1312.4400.
30. Szegedy, C.; Liu, W.; Jia, Y.; Sermanet, P.; Reed, S.; Anguelov, D.; Erhan, D.; Vanhoucke, V.; Rabinovich, A. Going deeper with
convolutions. In Proceedings of the 2015 IEEE Conference on Computer Vision and Pattern Recognition (CVPR), Boston, MA,
USA, 7–12 June 2015; pp. 1–9. [CrossRef]
31. Savory, S.J. Digital Coherent Optical Receivers: Algorithms and Subsystems. IEEE J. Sel. Top. Quantum Electron. 2010,
16, 1164–1179. [CrossRef]
32. Chan, C.C. Optical Performance Monitoring: Advanced Techniques for Next-Generation Photonic Networks; Academic Press: Cambridge,
MA, USA, 2010.
33. Khan, F.N.; Fan, Q.; Lu, C.; Lau, A.P.T. An Optical Communication’s Perspective on Machine Learning and Its Applications. J.
Light. Technol. 2019, 37, 493–516. [CrossRef]
34. LeCun, Y.; Bengio, Y.; Hinton, G. Deep learning. Nature 2015, 521, 436–444. [CrossRef] [PubMed]
35. Goodfellow, I.; Bengio, Y.; Courville, A. Deep Learning; MIT Press: Cambridge, MA, USA, 2016.
36. Hornik, K.; Stinchcombe, M.; White, H. Multilayer feedforward networks are universal approximators. Neural Netw. 1989,
2, 359–366. [CrossRef]
37. Szegedy, C.; Vanhoucke, V.; Ioffe, S.; Shlens, J.; Wojna, Z. Rethinking the Inception Architecture for Computer Vision. In
Proceedings of the 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR), Las Vegas, NV, USA, 27–30 June
2016; pp. 2818–2826. [CrossRef]
38. Szegedy, C.; Ioffe, S.; Vanhoucke, V.; Alemi, A.A. Inception-v4, Inception-ResNet and the impact of residual connections
on learning. In Proceedings of the Thirty-First AAAI Conference on Artificial Intelligence, San Francisco, CA, USA, 4–9
February 2017.
39. He, K.; Zhang, X.; Ren, S.; Sun, J. Deep residual learning for image recognition. In Proceedings of the IEEE Conference on
Computer Vision and Pattern Recognition, Las Vegas, NV, USA, 27–30 June 2016; pp. 770–778.
40. He, K.; Zhang, X.; Ren, S.; Sun, J. Identity mappings in deep residual networks. In Proceedings of the European Conference on
Computer Vision, Amsterdam, The Netherlands, 8–16 October 2016; pp. 630–645.
41. Abadi, M.; Barham, P.; Chen, J.; Chen, Z.; Davis, A.; Dean, J.; Devin, M.; Ghemawat, S.; Irving, G.; Isard, M.; et al. Tensorflow: A
system for large-scale machine learning. In Proceedings of the 12th USENIX Symposium on Operating Systems Design and
Implementation, Savannah, GA, USA, 2–4 November 2016; pp. 265–283.
42. Kingma, D.P.; Ba, J. Adam: A method for stochastic optimization. arXiv 2014, arXiv:1412.6980.
43. Ioffe, S.; Szegedy, C. Batch normalization: Accelerating deep network training by reducing internal covariate shift. In Proceedings
of the International Conference on Machine Learning, Lille, France, 6–11 July 2015; pp. 448–456.
44. Rasmussen, C.E. Gaussian Processes in Machine Learning; Springer: Berlin/Heidelberg, Germany, 2004; pp. 63–71. [CrossRef]
45. Snoek, J.; Larochelle, H.; Adams, R.P. Practical Bayesian Optimization of Machine Learning Algorithms. In Proceedings of the
Advances in Neural Information Processing Systems, Lake Tahoe, NV, USA, 3–6 December 2012; Volume 25.
46. Sammut, C.; Webb, G.I. Encyclopedia of Machine Learning; Springer Science & Business Media: Berlin/Heidelberg, Germany, 2011.
47. He, K.; Sun, J. Convolutional neural networks at constrained time cost. In Proceedings of the IEEE Conference on Computer
Vision and Pattern Recognition, Boston, MA, USA, 7–12 June 2015; pp. 5353–5360.