**1. Introduction**

With the exponential growth of data traffic due to emerging bandwidth-intensive services and internet applications such as HD video streaming, cloud computing, automatic driving, 5G and other emerging applications, the optical fiber communication capacity is gradually approaching the Shannon limit [1]. Optical transport endowed with higher bit rates and better network resource utilization is an inevitable step to comply with the demand of ever-increasing capacity. Accordingly, large-capacity and long-haul WDM systems equipped with reconfigurable optical add-drop multiplexers (ROADMs) based on wavelength selective switches (WSSs) have been widely deployed in metro and backbone optical networks. The application of ROADMs could improve the flexibility and efficiency of the WDM system significantly, but the optical network would become larger and larger in scale, more and more complex in topology and dynamic in connection at the same time [2]. Therefore, to ensure its working stability and operation efficiency, low-cost and highly reliable multi-channel OPM technology is indispensable, and therefore, it is attracting a lot of attention. Especially the optical power and OSNR of each WDM channel, as key channel performance indicators, are the most important and meaningful parameters that need to be accurately monitored/estimated [3,4]. In addition, wavelength connection monitoring enables network operators to quickly discover incorrect connections caused by

**Citation:** Yang, T.; Li, K.; Liu, Z.; Wang, X.; Shi, S.; Wang, L.; Chen, X. Optical Labels Enabled Optical Performance Monitoring in WDM Systems. *Photonics* **2022**, *9*, 647. https://doi.org/10.3390/ photonics9090647

Received: 29 July 2022 Accepted: 4 September 2022 Published: 9 September 2022

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2022 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

manual and/or software configuration errors, avoiding manual searching and recognizing, which is often time-consuming and labored. In addition, if the OPM can monitor the node location (hereinafter called "node ID") where the service signal is initially added and obtain specific surveillance information (such as symbol rate, modulation format, route configuration, etc., hereinafter called "wavelength ID"), it will be beneficial to the efficient management and optimization of WDM optical networks. However, the majority of critical monitoring modules of existing solutions, such as the conventional method using optical spectral analysis by spectrometer or optical filter, are generally too expensive, operationally inconvenient, have poor timeliness, and their functionally is limited to applied in performance monitoring of all channels over a WDM system with numerous nodes [5].

The existing mainstream monitoring schemes mainly take advantage of direct-or coherent detection of high-speed service signal and use DSP algorithms to monitor the performance of WDM channels. The monitoring technology based on direct detection without optical filters usually works at the receiver end, which divides a small part (less than 5%) of the service's optical signal for direct detection, and it analyzes the characteristics of the received signal by DSP algorithms to obtain the channel performance in a low-cost way [6,7]. However, most of the existing monitoring schemes based on direct detection inherently lose more information about the channel characteristics, and the accuracy of the monitoring is greatly reduced while the correctness of the wavelength connection also cannot be realized. On the other hand, optical filter-based monitoring technology can coherently acquire spectral information in the channel or a wavelength range through a tunable filter or diffraction grating spectrometer. In addition, due to the natural wavelength selectivity of coherent detection, the analysis of spectral information, optical power, modulation format, OSNR and other monitoring information has high recognition resolution and monitoring dynamic range [8]. However, it requires expensive narrow linewidth tunable lasers, and thereby it is difficult to apply in WDM networks due to its high cost, complex structure and one-by-one channel monitoring restrictions. Moreover, most of the existing monitoring solutions based on machine learning (ML) need to preprocess the high-speed sampling sequence with the help of DSP at the receiving end and then perform data analysis, processing and feature extraction [9,10]. The ML methods are not only difficult to effectively monitor the correctness of wavelength connections but also have higher implementation complexity than conventional DSP methods and require a large amount of data set training, parameter tuning and model generalization in advance [11]. Consequently, for a WDM network with large scale, complex structure and continuous network topology expansion, the OPM technology still presents a prominent contradiction between monitoring capability and implementation cost.

To solve these problems, OPM schemes based on optical labels have been proposed and investigated [12–20]. The typical monitoring technology based on optical label mainly adopts a pilot tone modulation to generate a low-speed label signal corresponding to the target channel, where the pilot-tone-carried label signal is loaded on the service optical signal as optical labels. At monitoring nodes, a low bandwidth direct-detection photodetector (PD) followed by the analog-to-digital converter (ADC) acquires the label signal of all wavelength channels simultaneously. The bandwidth of PD is usually at hundred-MHz order while the sampling rate of ADC is about hundred-MSa/s that is enough to demodulate the labels with relatively low cost. Theoretically, the optical labels are tied up to their corresponding optical channels anywhere in the network, and the node ID as well as wavelength ID can be flexibly loaded as optical labels; it can not only monitor the wavelength channel performance but also can reliably deliver any interested surveillance information of each channel to further sense the working state and optimize resource utilization of WDM networks.

However, when oriented to practical applications, there are still many problems in the existing solutions. On the one hand, the OPM techniques based on only pure pilot tones in [14–16] have a trade-off problem of two negative effects of stimulated Raman scattering (SRS) and chromatic dispersion. When the low frequency pilot tone is used, it will have severe SRS effects that need to be combated by dividing the whole wavelength channels into several sub-channels with the assistance of expensive optical filter. In contrast, when highfrequency pilot tones are used, the large CD caused by long-distance fiber transmission will cause severe power fading of the pilot tones, thereby significantly degrading the monitoring performance. On the other hand, the optical labelling scheme based on subcarrier index modulation (SIM) reported in [17–20] has the advantage of efficient demodulation of optical labelling signals. Nevertheless, since it adopts the SIM scheme for optical labels, not only the modulation and demodulation require large-size (16,384 points in [18]) inverse fast Fourier transform (IFFT) and fast Fourier transform (FFT), respectively, which results in very high circuit realization complexity, but also it is difficult to monitor channel optical power due to the SRS effect, and likewise, it is hard to flexibly adjust the transmission rate of label signals. In our previous work, we used pilot tones of a few MHz and demonstrated a scheme that could suppress the SRS effect on optical power monitoring based on our proposed SRS mitigation algorithm [21]. However, a scheme using such low frequency (several MHz) pilot tone has a serious SRS effect, and the monitoring performance is still difficult to guarantee even using the SRS mitigation algorithms [22,23].

In this paper, we present our recent research activities and progress on optical-label enabled OPM, and a low cost and high efficiency multi-channel optical power monitoring and OSNR estimation scheme based on differential phase shift keying (DPSK) digital optical labels is proposed and demonstrated. The proposed scheme innovatively adopts the service transmitter to flexibly load the time-domain digital label signal modulated on unique pilot tone that arranged for the corresponding wavelength channel, without additional digital to analog converter (DAC) or customized optical modulator, and without complex IFFT operation. The pilot tones have a frequency of tens of MHz that is modulated by low-speed (~K Baud) digital label information, i.e., the node ID, wavelength ID and so no. The channel optical power is monitored using the method of spectral integration, and the OSNR is estimated by calculating the amplified spontaneous emission (ASE) noise accumulation that contributed by all amplifiers. To verify the proposed scheme, we have performed simulations on WDM systems within 20 spans of fiber transmission, and the results show that the maximum error of power monitoring is less than 0.9 dB while the error of OSNR estimation is also less than 1 dB. Therefore, compared with the conventional optical filter-based OPM technology, the implementation of the proposed scheme is of lower cost and higher efficiency, which has great potential for large-scale deployment in practical WDM networks.

### **2. Principle of the Optical-Label Enabled OPM Scheme**

The optical label is known as a top modulation or a small amount of amplitude modulation, also called low-frequency perturbation or pilot tones in some studies. The schematic diagram of the scheme is shown in Figure 1. After the digital surveillance information of node ID and wavelength ID of each wavelength channel is mapped by DPSK coding, it is modulated (~K Baud rate) to the pilot tone with a specific frequency to be the final optical label. It is noteworthy that the pilot tone frequency is larger than 40 MHz to avoid the impact of SRS crosstalk [24]. Thus, the power monitoring error caused by the SRS, resulting in the non-correspondence between the label power and the channel power, can be reasonably ignored.

**Figure 1.** Schematic diagram of the optical-label enabled OPM scheme. (OTU: optical transponder unit, OPA: optical pre-amplifier, OBA: optical boost amplifier, OLA: optical line amplifier, MZM: Mach-Zehnder modulator, PBC: polarization beam combiner, PT: pilot tone, OSC: optical supervising channel). **Figure 1.** Schematic diagram of the optical-label enabled OPM scheme. (OTU: optical transponder unit, OPA: optical pre-amplifier, OBA: optical boost amplifier, OLA: optical line amplifier, MZM: Mach-Zehnder modulator, PBC: polarization beam combiner, PT: pilot tone, OSC: optical supervising channel).

### *2.1. Optical Labels Loading and Detection 2.1. Optical Labels Loading and Detection*

At the transmitter side in Figure 1, wavelength channel 1 is taken as an example to introduce the process of label generation and modulation. In the transmitter-side DSP module, the label data are first mapped into BPSK symbols and then modulated onto pilot tones with a specific frequency to generate label signals. The label signals are subsequently loaded into the high-speed service signal in the electrical domain and thus generate a labeled service signal. It is then converted to analog signal by the DAC and drives the optical I/Q modulator to generate a labeled optical signal. Here, the DAC is the one used by the high-speed service itself, not an additional one. When arriving at optical pre-amplifier (OPA) or optical boost amplifier (OBA) in a service node, or optical line amplifier (OLA) in relay node, the received signal is divided into two parts. Thus, 99% power of the signal is used for the coherent detection of service optical signal, and 1% power of signal is used for performance monitoring and optical label demodulation. The 1% power of signal is sent into a PD to achieve optical-to-electrical conversion. The electrical signal is then converted to a digital signal by an analog-digital converter (ADC). Then, after monitoring the optical power and OSNR estimation, the node ID and wavelength ID are recovered. Finally, the monitoring results of channel optical power, OSNR and node/wavelength ID are transmitted to the control plane through optical supervising channel (OSC), and the control plane thus could not only check the route, symbol rate, modulation format and so on of each wavelength channel, but also analyze the performance of the whole WDM op-At the transmitter side in Figure 1, wavelength channel 1 is taken as an example to introduce the process of label generation and modulation. In the transmitter-side DSP module, the label data are first mapped into BPSK symbols and then modulated onto pilot tones with a specific frequency to generate label signals. The label signals are subsequently loaded into the high-speed service signal in the electrical domain and thus generate a labeled service signal. It is then converted to analog signal by the DAC and drives the optical I/Q modulator to generate a labeled optical signal. Here, the DAC is the one used by the high-speed service itself, not an additional one. When arriving at optical pre-amplifier (OPA) or optical boost amplifier (OBA) in a service node, or optical line amplifier (OLA) in relay node, the received signal is divided into two parts. Thus, 99% power of the signal is used for the coherent detection of service optical signal, and 1% power of signal is used for performance monitoring and optical label demodulation. The 1% power of signal is sent into a PD to achieve optical-to-electrical conversion. The electrical signal is then converted to a digital signal by an analog-digital converter (ADC). Then, after monitoring the optical power and OSNR estimation, the node ID and wavelength ID are recovered. Finally, the monitoring results of channel optical power, OSNR and node/wavelength ID are transmitted to the control plane through optical supervising channel (OSC), and the control plane thus could not only check the route, symbol rate, modulation format and so on of each wavelength channel, but also analyze the performance of the whole WDM optical network comprehensively.

tical network comprehensively. In a typical WDM system, one fiber carries the high-speed service signals of 80 wavelength channels, so different frequency PT with digital labels are modulated on corresponding wavelength channels where the same label signals are modulated on two polarizations to improve the received signal-to-noise ratio (SNR) of optical labels. For instance, the labels with PT frequency of *f1* are assigned to the service optical signal on wavelength channel 1 of *λ*1, and the rest can be done in the same manner so that the digital band-pass filter in the OPM module could extract corresponding digital labels on the PT with different frequencies. The optical field of the dual polarization optical signals with amplitude-In a typical WDM system, one fiber carries the high-speed service signals of 80 wavelength channels, so different frequency PT with digital labels are modulated on corresponding wavelength channels where the same label signals are modulated on two polarizations to improve the received signal-to-noise ratio (SNR) of optical labels. For instance, the labels with PT frequency of *f* <sup>1</sup> are assigned to the service optical signal on wavelength channel 1 of *λ*1, and the rest can be done in the same manner so that the digital band-pass filter in the OPM module could extract corresponding digital labels on the PT with different frequencies. The optical field of the dual polarization optical signals with amplitude-modulated optical labels can be expressed as follows:

$$E\_{i\eta,\mathbf{x},i} = A\_{\mathbf{x},i}(t) \cdot \exp\left\{j(2\pi f\_{o,i}t + \theta\_{\mathbf{x},i}(t))\right\} \cdot \sqrt{\left[1 + m \cdot \cos(2\pi f\_i t) \cdot D\_i(t)\right]}\tag{1}$$

modulated optical labels can be expressed as follows:

$$E\_{\rm in,y,i} = A\_{y,i}(t) \cdot \exp\left\{j\left(2\pi f\_{o,i}t + \theta\_{y,i}(t)\right)\right\} \cdot \sqrt{\left[1 + m \cdot \cos(2\pi f\_l t) \cdot D\_l(t)\right]}\tag{2}$$

where *Ein*, *x y* ,*i* represents the electric field of *x*/*y* polarization on the *i*-th channel, *x* and y represent two polarizations, *n* represents the serial number of the wavelength channel where it is located. *Ax*,*i*(*t*)· exp{*j*(2*π fo*,*<sup>i</sup> t* + *θx*,*i*(*t*))} represents the service date signal, where *Ax*,*i*(*t*) is the amplitude, *fo*,*<sup>i</sup>* is the center wavelength of *i*-th channel, *θx*,*i*(*t*) is the phase of service signals. cos(2*π f<sup>i</sup> t*)·*Di*(*t*) is the optical label, where *m* represents the modulation depth of optical labels and *f<sup>i</sup>* is the frequency of the corresponding pilot tone. It is noteworthy that the square root form of the optical label modulation on the service signal is used to facilitate analysis of label reception during square-law detection using the PD module. When it comes to the detection of optical labels in an OPM module, 1% of the optical power component is isolated and inputted into the OPM module. Direct-detection low bandwidth PD and low sample-rate ADC are used to realize the photoelectric conversion and digital sampling of the optical labels of all wavelength channels simultaneously. The PD detection can be expressed with the following equations.

$$I = \sum\_{i=1}^{n} \left[ \left| E\_{\text{in},x,i} \right|^2 + \left| E\_{\text{in},y,i} \right|^2 \right] = \sum\_{i=1}^{n} \left\{ \left[ A\_{\text{x},i} ^2(t) + A\_{y,i} ^2(t) \right] \cdot \left[ 1 + m \cdot \cos(2\pi f\_i t) \cdot D\_i(t) \right] \right\} \tag{3}$$

where *I* represents the signal after PD square-law detection. The signal contains all the optical labels of each channel. *n* represents the number of PT. Here, the spectrum of optical labels is obtained and analyzed by the FFT algorithm, which is an important step for optical labels-based power monitoring. Afterward, since the digital labels are loaded by different frequency PT on different wavelength channels, digital band-pass digital filters are used here to separate the labels on different channels. Finally, differential demodulation methods are used to realize demodulation of the DPSK-based optical labels and recover the ID information. The filtered optical labels (digital labels on PT) of the *i*-th channel can be expressed as

$$I\_i = \left[A\_{\mathbf{x},i}\,^2(t) + A\_{y,i}\,^2(t)\right] \cdot m \cos(2\pi f\_i t) \cdot D\_i(t) = A\_i(t) \cdot m \cos(2\pi f\_i t) \cdot D\_i(t) \tag{4}$$

where *Ai*(*t*) equals [*Ax*,*<sup>i</sup>* 2 (*t*) + *Ay*,*<sup>i</sup>* 2 (*t*)], representing the optical label amplitude of the *i*-th channel after PD reception.

### *2.2. Channel Optical Power Monitoring*

It is obvious that there is a linear proportional relationship between the channel's optical power of the service signal and the detected power of the optical labels. In addition, the integral value of the received spectrum of optical labels for a wavelength channel could represent the detected power of optical labels on the channel based on pre-tested conversion coefficients. These relationships can be expressed using Equation (5).

$$P\_{ch,i} = \frac{P\_{Label,i}}{m \cdot a} \cdot f(V\_{PD}) \tag{5}$$

where *Pch*,*<sup>i</sup>* and *PLabel*,*<sup>i</sup>* represent the actual optical power of the *i*-th channel and the corresponding monitored optical label power respectively. *m* represents the modulation depth of the optical label. *f*(*VPD*) represents the calibration coefficient of the PD module, and it depends on the photoelectric conversion characteristics of the PD itself and the feature of the optical labels. α represents the splitting ratio of the optical coupler (OC). Typically, a 1:99 (*α* = 1%) OC is used in our scheme to divide the labeled optical signal into OPM module.

After photoelectric conversion and digital sampling by a PD and ADC, a DSP unit is used to calculate the channel optical power. Firstly, FFT is used to obtain the spectrum of optical label, and the reception spectrum of an 80-channel WDM using 0.1 MSa/s digital labels is shown in Figure 2. It is shown that the PT frequency range of 40–55.8 MHz with 0.2 MHz

grid between neighbor channels is adopted. Then, spectrum integration of each optical label spectrum from the beginning PT frequency to the ending PT frequency is carried out, respectively. For example, the channel-1 power is equal to the spectrum integration value from [39.9–40.1] MHz range. Next, a proportional coefficient of photoelectric conversion of PD is used to calculate the final channel optical power of the service signal. Finally, the monitoring results of all network monitoring nodes are transmitted to the control plane through optical supervising channel (OSC), and the control plane analyzes the performance of the whole optical network and optimizes resource utilization comprehensively. carried out, respectively. For example, the channel-1 power is equal to the spectrum integration value from [39.9–40.1] MHz range. Next, a proportional coefficient of photoelectric conversion of PD is used to calculate the final channel optical power of the service signal. Finally, the monitoring results of all network monitoring nodes are transmitted to the control plane through optical supervising channel (OSC), and the control plane analyzes the performance of the whole optical network and optimizes resource utilization comprehensively.

optical label spectrum from the beginning PT frequency to the ending PT frequency is

*Photonics* **2022**, *9*, x FOR PEER REVIEW 6 of 16

**Figure 2.** Spectrum of the detected optical label of 80-channel WDM systems. **Figure 2.** Spectrum of the detected optical label of 80-channel WDM systems.

### *2.3. OSNR Estimation 2.3. OSNR Estimation*

The scheme diagram of OSNR estimation is also shown in OPM Module in Figure 1. In a multi-span WDM transmission system, OSNR will gradually deteriorate due to the accumulation of ASE noise generated by the EDFA. Therefore, the OSNR could be estimated by calculating the ASE noise accumulation. In a practical, optical transmission system, the actual gain of the amplifier may be different from the configured gain, so we can calculate the actual current working gain of each optical amplifier with the help of the optical power monitored by the optical labels. At the same time, because the noise figure of optical amplifier changes with different wavelengths, here we can obtain the value of the noise figure (NF) pattern by measuring it in advance. Finally, according to the EDFA amplification of each wavelength channel, the overall amplification case of each wave-The scheme diagram of OSNR estimation is also shown in OPM Module in Figure 1. In a multi-span WDM transmission system, OSNR will gradually deteriorate due to the accumulation of ASE noise generated by the EDFA. Therefore, the OSNR could be estimated by calculating the ASE noise accumulation. In a practical, optical transmission system, the actual gain of the amplifier may be different from the configured gain, so we can calculate the actual current working gain of each optical amplifier with the help of the optical power monitored by the optical labels. At the same time, because the noise figure of optical amplifier changes with different wavelengths, here we can obtain the value of the noise figure (NF) pattern by measuring it in advance. Finally, according to the EDFA amplification of each wavelength channel, the overall amplification case of each wavelength can be obtained. Thus, based on the EDFA amplification and ASE noise accumulation of each wavelength channel, the OSNR can be estimated by the 58-Formula.

length can be obtained. Thus, based on the EDFA amplification and ASE noise accumulation of each wavelength channel, the OSNR can be estimated by the 58-Formula. The detailed principle of OSNR estimation is as follows. For each optical amplifier in the network, its ASE noise could be expressed as [25]:

$$\text{MSE}(\text{dBm}) = 10 \cdot \log[(h\nu)(\text{B}\_0)] + \text{NF}(\text{dB}) + \text{G}(\text{dB}) = -58(\text{dBm}) + \text{NF}(\text{dB}) + \text{G}(\text{dB}) \tag{6}$$

ASE(dBm) = 10 · log[(ℎ)()] + (dB) + (dB) = −58(dBm) + (dB) + (dB) (6) where ℎ represents the energy of a single photon, represents the reference bandwidth of OSNR which is 12.5 GHz. The *NF* represents the noise figure of optical amplifier. *G* represents the actual gain of optical amplifier which is calculated using the results of optical power monitoring. When the service signal is transmitted through a series of amwhere *hν* represents the energy of a single photon, *B<sup>o</sup>* represents the reference bandwidth of OSNR which is 12.5 GHz. The *NF* represents the noise figure of optical amplifier. *G* represents the actual gain of optical amplifier which is calculated using the results of optical power monitoring. When the service signal is transmitted through a series of amplifiers, the ASE noise will be added linearly. At this time, on the service transmission link, the ASE noise accumulation by the *i*-th amplifier in *j*-th channel can be expressed as:

ୀଵ

$$\text{ASE}\_{\text{sum\\_u},i,j}(\text{dBm}) = 10 \cdot \log \sum\_{i=1}^{n} 10 \frac{\frac{-\text{Sb}(\text{dBm}) + \text{N}\_{\text{nJ}}(\text{dB}) + \text{P}\_{\text{ant}\_{\text{nJ}}}(\text{dBm}) - \text{P}\_{\text{in}\_{\text{nJ}}}(\text{dBm})}{10} \tag{7}$$

ଵ

(7)

where the *Poutn*,*<sup>i</sup>* a and *Pinn*,*<sup>i</sup>* represent the output-power and input-power of *n*-th amplifier in *i*-th channel, respectively. Finally, the OSNR of each channel can be estimated by Equation (8).

$$OSNR\_{\mathbf{n},i}(\mathbf{dB}) = P\_{out\_{\mathbf{n},i}}(\mathbf{dBm}) - A\mathbf{SE}\_{\mathbf{sum\\_n,i}}(\mathbf{dBm}) \tag{8}$$

### *2.4. CAPEX Analysis*

Capital expenditure (CAPEX) is an important issue in optical network monitoring. How to monitor optical network with low cost and high efficiency has become the goal pursued by mainstream manufacturers and operators. We conduct a detailed cost analysis of the proposed scheme and two typical existing schemes. We compare the CAPEX of OPM schemes based on optical filter (OF) (hereinafter referred to as scheme A) [15], SIM labels (scheme B) [19] and the proposed method (scheme C). We learn the market price from a number of related device/module manufacturers and customers, and the comparison of CAPEX is mainly based on the market price of all devices/modules used in different OPM schemes. It should be noted that because the labeling rate is low, the loading process is simple, and the hardware resources required by the DSP unit are relatively small, so their cost can be reasonably ignored. Here, the cost of OPM over an 80-channels WDM system is taken as an example to compare the CAPEX. The reference cost units of different devices used in the three schemes are shown in Table 1.

**Table 1.** The cost of different modules used in the three OPM schemes.


Obviously, the proposed OPM scheme adopts only the service transmitter to flexibly load the time-domain digital label signals and uses a PD to detect the optical labels of all channels simultaneously. It is shown in Figure 3 that, compared with scheme A, the proposed scheme greatly reduces the CAPEX by 96% because the variable optical attenuator (VOA) and OF are completely removed. Compared with scheme B, the complex IFFT operation and VOAs are both removed in the proposed scheme, and thereby, the CAPEX is further reduced by 75%. Therefore, it could be concluded that the proposed scheme has the advantages of very low cost and high efficiency, and therefore, it has great potential in WDM network monitoring. *Photonics* **2022**, *9*, x FOR PEER REVIEW 8 of 16

With the help of VPI TransmissionMaker 9.0 created by VPIphotonics(Berlin, Ger-

structed a simulation platform of WDM transmission with OPM module based on DPSKmodulated digital optical labels to investigate the performance of the proposed scheme for channel power monitoring and OSNR estimation. Here we simulate 8, 32 and 64 channels within 20 spans WDM transmission systems. The 2 Mbps rate digital labels modulated by different PTs with 4 MHz grid are arranged to 16 G Baud polarization multiplexing (PM) QPSK and PM-16QAM transmission systems. The center wavelength of service signal is form 193.1–196.25 THz with 50 GHz interval. A photodetector with 300 MHz −3 dB bandwidth and ADC with 600 MSa/s sample rate and 10-bit resolution are used to detect the optical labels in a WDM system. Table 2 shows important simulation parameters. Here we firstly investigate the performance when using different label modulate depth of 0.05, 0.10 and 0.15. The result of optical label demodulation, such as signal noise ratio (SNR) and label power (result of spectrum integral) in QPSK and 16QAM systems,

**Figure 3.** CAPEX comparison of three schemes. **Figure 3.** CAPEX comparison of three schemes.

**3. Simulation and Discussion** 

is shown in Table 3.

**Service Signal** 

QPSK

**Table 2.** Simulation parameters**.** 

**Service Modulation Format** PM-QPSK/16QAM **Modulation Depth** 5%/10%/15% **Service Baud Rate** 16 G Baud **EDFA Noise Figure** 5 dB **Launch Power** 0 dBm **Length of Each Span** 100 km **Digital Label Format** DPSK **PMD Coefficient** 0.1 ps/(km)^1/2 **Digital Label Rate** 2 Mbps **PD Bandwidth** 300 MHz **Bit Number of Digital Labels** 32 bit **ADC Sample Rate** 600 MSa/s **Carrier Wavelength** 193.1–196.25 THz **ADC Resolution** 10 bit **Channel Spacing** 50 GHz **Fiber Attenuation** 0.2 dB/km

**Table 3.** SNR and power of optical labels in with different modulate depth**.** 

**Format Modulation Depth SNR of Optical Label (dB) Label Power (mW)** 

16QAM 5% 6.95 0.91

5% 10.59 0.96 10% 13.71 1.99 15% 15.10 2.98

10% 10.29 1.88

### **3. Simulation and Discussion**

With the help of VPI TransmissionMaker 9.0 created by VPIphotonics (Berlin, Germany) and MATLAB R2020a created by MathWorks (Natick, MA, USA), we have constructed a simulation platform of WDM transmission with OPM module based on DPSKmodulated digital optical labels to investigate the performance of the proposed scheme for channel power monitoring and OSNR estimation. Here we simulate 8, 32 and 64 channels within 20 spans WDM transmission systems. The 2 Mbps rate digital labels modulated by different PTs with 4 MHz grid are arranged to 16 G Baud polarization multiplexing (PM) QPSK and PM-16QAM transmission systems. The center wavelength of service signal is form 193.1–196.25 THz with 50 GHz interval. A photodetector with 300 MHz −3 dB bandwidth and ADC with 600 MSa/s sample rate and 10-bit resolution are used to detect the optical labels in a WDM system. Table 2 shows important simulation parameters. Here we firstly investigate the performance when using different label modulate depth of 0.05, 0.10 and 0.15. The result of optical label demodulation, such as signal noise ratio (SNR) and label power (result of spectrum integral) in QPSK and 16QAM systems, is shown in Table 3.

**Table 2.** Simulation parameters.


**Table 3.** SNR and power of optical labels in with different modulate depth.


According to Table 3, we find that the optical label performance in QPSK system is better than that in 16QAM system, and the optical label power basically exhibits a linear relationship as the modulation depth increases. Optical label power and modulation depth *m* are basically linear, and the label information could be recovered accurately. In addition, the SNR of optical label under different *m* is sufficient for error-free demodulation.
