1. Introduction

Optical performance monitoring (OPM) has emerged as an increasingly practical topic for enabling future robust, stable, and high-performance optical communication systems. As network architecture complexity, channel count, and bit rate increase, the operating range of the network shrinks dramatically. OPM becomes challenging since the networks paths are dynamic and channel degradation mechanisms can change with temperature, component drift, aging, and maintenance [1]. OPM can enable the network itself to agilely provision data degradations and dynamically allocate physical-layer resources to ensure error-free transport. Ideally, OPM should isolate the specific cause that degrades a data stream, such as deleterious CD, PMD, nonlinearities, or low optical signal-to-noise ratio (OSNR). As advances are made in OPM functionality, the optical systems are transmitting increasingly complex data formats. This important trend in advanced data formats is driven by the desire for: (i) enhanced receiver sensitivity and tolerance to nonlinearities using DPSK format, and (ii) better spectral efficiency and tolerance to CD and PMD using differential quadrature PSK (DQPSK).

In order to fit more bits/s/Hz on any channel, a technique that has gained much interest for the ability of doubling the system capacity and spectral efficiency is the utilization of polarization-multiplexing (Pol-Mux) of two independent data streams on two orthogonal polarization states [2]. High-capacity transmission systems tend to pol-mux two DPSK or DQPSK data channels for high spectral efficiency, robustness to fiber dispersive effects, and relaxed limitation on the available operation bandwidth of the RF/optical components although this format is inherently more sensitive to the PMD effects than other types of formats [3,4]. It could be quite valuable to investigate OPM methods to monitor PMD in a high-speed pol-muxed system, especially in a manner that is insensitive to CD.

We note that some OPM methods have been previously demonstrated to efficiently monitor the PMD effect of on-off-keying (OOK) and D(Q)PSK data channels [5–12]. However, some methods are susceptible to the CD effect [7–11] and require high-speed components [5,7–11] and complex post-processing algorithms [8,10,11] in the measurements, which might have limitations for monitoring of high-speed (i.e., >40 Gb/s) data channels. Additionally, since the D(Q)PSK data is vulnerable to tight optical filtering, some method might not be applicable. We previously demonstrated the PMD monitor for high-speed D(Q)PSK data using the DGD-induced interferometric filtering [12]. However, the filtering effect cannot be observed for pol-muxed channel due to the existing of another orthogonal data stream. As a consequence, it becomes quite challenging to monitor and isolate the PMD effect for the high-speed pol-muxed data since two independent data streams are transmitted on two orthogonal polarization states simultaneously. To date, there has been little reported on PMD monitoring in a pol-muxed system.

In this paper, we propose and experimentally demonstrate a new CD-insensitive technique for monitoring PMD on high-speed pol-muxed data channel. The PMD monitor consists of a polarizer, a low-speed detector, and an RF power measurement. By rotating the polarization controller, the polarizer is aligned along one data stream of the pol-muxed RZ-DPSK channel. Therefore, the detected RF power increases with the increasing accumulated DGD due to the crosstalk caused by the DGD-induced depolarizations of two data streams, which can be used for monitoring. It is noted that the RF power is measured at low frequency (i.e., 250 MHz), indicating that the monitored RF power is less sensitive to CD [13]. The monitored power increment of 16.2 dB, which is insensitive to CD of 0~100 ps/nm, is achieved in the presence of up to 12-ps DGD value for an 80-Gb/s pol-muxed RZ-DPSK data channel. The proposed technique is simple and potentially applicable to 100-Gb/s pol-muxed RZ-DQPSK channel with low-speed components for measurements.

2. Concept

The 80-Gb/s pol-muxed channel carries two 40-Gb/s RZ-DPSK data streams on two orthogonal polarizations, denoted by Data 1 and Data 2 in Fig. 1(a) . It is noted that the two data streams are synchronized in time. By adjusting the polarization controller, the pol-muxed channel is launched at 45 degrees relative to the tunable DGD element in order to emulate the worst case of the first-order PMD. Each of the two RZ-DPSK data streams decomposes its own polarization state into two orthogonal components along the slow and fast axes of the DGD element and experiences the DGD of Δτ. As a result, the degree of polarization of each RZ-DPSK data stream is reduced (i.e. depolarized) due to DGD. At the output of the DGD emulator, both of the DGD-distorted data streams are combined together on the two orthogonal polarizations. In the PMD monitor, the direction of the polarizer is fixed and thus the second polarization controller (PC) is used to adjust the polarization rotation of the pol-muxed channel. The extinction ratio of the polarizer needs to be high in order to avoid the unwanted coupling between Data 1 and Data 2, which enhances the monitoring sensitivity. We also note that the direction of the polarizer can be adjusted by automatic control systems, which will provide fast alignment and long-term stabilization [14].

 

Fig. 1 (a) Conceptual diagram of the proposed PMD monitoring technique. (b) The polarizer is aligned along one RZ-DPSK data stream, and the extracted Data 1 is sent to the low-speed detector for measurements. A CD-insensitive RF power increment at low frequency component is used for monitoring the PMD.

Download Full Size | PDF

Specifically as shown in Fig. 1(b), when the pol-muxed channel is exactly aligned to the polarizer, the channel is extracted by the polarizer, detected by the low-speed detector and sent to the RF spectrum analyzer for measurements. Without the DGD, both Data 1 and Data 2 remain orthorgonal and thus the polarizer extracts only one polarized RZ-DPSK Data 1, resulting in minimum RF power since Data 2 is completely removed. We note that the effect of the crosstalk between Data 1 and Data 2 occurs and is different when the angle between the pol-muxed channel and polarizer is misaligned. The minimal RF power occurs when the angle is well aligned (i.e., 0°) while the maximum RF power results from the angle of 45° [14,15]. As a consequence, the polarizer is controlled to track the minimum RF power for monitoring purposes. On the other hand, the DGD depolarizes both orthogonal Data 1 and Data 2, which means that both data streams’ degrees of polarization are reduced. The DGD-distorted pol-muxed data channel interacts with the aligned polarizer and then results in RF power increment due to the DGD-induced crosstalk (i.e., interference effect) between Data 1 and Data 2 along the direction of the polarizer. As a consequence, the RF power increases significantly as the DGD value is increased, which can be measured for monitoring the accumulated PMD effect. It is noted that the measured RF power particularly in the low frequency region is less sensitive to CD and thus remains almost the same [13], which means that low-speed detection is sufficient to provide the monitored PMD information of high-speed pol-muxed data channel. We emphasize that the minimal RF power for each of the monitored DGD values occur when the angle is well aligned.

3. Experimental setup

The experimental setup for PMD monitoring of an 80-Gb/s pol-muxed RZ-DPSK data channel is schematically shown in Fig. 2 . At the transmitter, a continuous-wave laser at 1550.920 nm is externally modulated by a 40-GHz phase modulator with 2¹⁵-1 pseudorandom binary sequence and then an RZ pulse carver driven by a 20-GHz clock source to generate the 40-Gb/s RZ-DPSK signal. The signal is then split, decorrelated and combined into an 80-Gb/s pol-muxed RZ-DPSK data channel using a 50/50 coupler, a piece of single-mode fiber, and a polarization beam combiner (PBC), respectively. Two polarization controllers (PCs), an optical tunable delay module and an optical attenuator (Att) are adjusted to maintain the orthogonal polarization states, time alignment and equal power between two RZ-DPSK signals. The polarization extinction ratio of PBC in use is 23 dB, which can maintain the orthogonal states and suppress the unwanted crosstalk.

 

Fig. 2 Experimental setup of the proposed PMD monitoring technique for an 80-Gb/s pol-muxed RZ-DPSK.

Download Full Size | PDF

The pol-muxed channel is coupled with amplified spontaneous emission (ASE) noise and amplified by an erbium-doped fiber amplifier (EDFA) to adjust the OSNR value. The channel then passes through the tunable CD and DGD emulators. The worst case PMD is generated by aligning the PC before the DGD emulator. The PMD monitor consists of a polarization controller, a polarizer, a 10-GHz receiver and then an RF spectrum analyzer. The RF powers under different DGD/CD values are measured at low frequency content (i.e., 250 MHz). The RF frequency that we utilize depends on the resolution and bandwidth of the RF spectrum analyzer, which might not be tuned finely. RF filter and power meter can be used to replace the RF spectrum analyzer so that the measurable RF power will depend on the available filter frequency. The commercial RF filter can enable the measurements at very low frequency.

4. Results and discussion

The measured RF power spectra and optical waveforms with different alignments between the pol-muxed channel and polarizer are shown in Fig. 3(a) . In this case, the DGD and CD values are tuned to zero. Due to the crosstalk between two orthogonal RZ-DPSK data streams, the measured RF power increases significantly as the misaligned angle increases. We observe that the minimum crosstalk and RF power is achieved when the polarizer is well aligned to one polarized RZ-DPSK data, indicating that the power change due to DGD can be monitored by tracking the minimum RF power. On the other hand, the maximum crosstalk and RF power occur when the polarizer is aligned at 45° relative to the pol-muxed channel, and the crosstalk results in constructive and destructive interferences, which fluctuate over time. Additionally, Fig. 3(b) shows the measured optical waveforms of the pol-muxed and de-pol-muxed RZ-DPSK data streams in the presence of different DGD values. The distorted pol-muxed data channel is extracted by a polarizer with proper alignment (i.e., with the minimum RF power).

 

Fig. 3 (a) Measured RF power increases when the polarizer is misaligned to the pol-muxed channel (without DGD/CD). (b) Optical waveforms of the pol-muxed and de-pol-muxed data with DGD. (c) RF power (at 250 MHz) increases with the increasing DGD. The insets show the optical waveforms of the pol-muxed data. (d) RF power under 0~100-ps/nm CD for three different DGD values.

Download Full Size | PDF

Moreover, the polarizer is aligned to one data stream of the pol-muxed channel and the RF power is measured at 250 MHz for monitoring. The CD is tuned to zero and different DGD values are emulated from 0 to 12 ps, respectively. As shown in Fig. 3(c), the RF power increment around 16.2 dB is obtained with the increasing DGD value and the insets in the figure show the optical waveforms of the pol-muxed data (i.e., without the polarizer) for 6 and 8 ps DGD, respectively. Since the RF power at low frequency content has small variations in the presence of different CD values as compared to the high frequency content. The RF spectra at the low frequency content for different DGD values also have similar insensitivity to CD. Figure 3(d) shows the monitored RF power at 250 MHz under different combinations of CD and DGD values, and the measured results are less sensitive to CD up to 100 ps/nm. We simulate our scheme to investigate the maximum tolerance to CD and find that CD of 144 ps/nm starts to cause > 1-dB power fluctuation.

The angle between the DGD element and the pol-muxed tributaries determines the effect of depolarization, and the RF power changes with various angles. If the angle is launched at zero degree, DGD only induces the time misalignment between the two tributaries and the measurement is not changing. In contrast, the effect of DGD becomes the worst when the two tributaries are misaligned with a half-bit time and 45-degree angle [16]. In our experiments, the bits of the two polarized tributaries are aligned and we expect that the measured RF power will increase further with the increasing DGD and the time misalignments. For non-pol-muxed channel, the incident angle can be calibrated by using a polarization beam splitter and then analyzing the data of the two output ports. For pol-muxed channel, this effect cannot be observed due to the existing of another polarized data. The data of the two output ports provides the same information, which might not be useful to isolate the incident angle from DGD. A potential approach by rotating the polarizer continuously and analyzing the RF powers measured at different angles might be used to determine the incident angle. Additionally, the measurements depend on other parameters, including OSNR, optical input power, and data pattern. The RF power decreases with the increasing OSNR due to the contribution of ASE noise. The input power sent to the detector is typically fixed to 0 dBm for maximum dynamic range, which can be achieved by using a fixed output power optical amplifier. Pattern dependence results from different effects, such as optical filtering, output signal of semiconductor optical amplifier and so forth. Our scheme is tolerant to the pattern dependence that results from the optical filtering. Other specific pattern dependences might influence the measured RF spectrum, which are beyond the scope of this paper. When considering these parameters, the RF power can be measured at the same frequency with the direct detection of the pol-muxed channel. DGD induces negligible change on the measured RF power with direct detection, and thus the difference of the measurements between our scheme and the direct detection might provide calibrated information that is isolated from other impairments.

Figure 4 shows the monitoring ranges of the simulated results for the exact alignment and 5 degrees-misalignment between the pol-muxed channel and polarizer. The RF power with the exact alignment increases by ~30 dB as the normalized DGD is increased (i.e., up to 25 and 40 ps) for 80-Gb/s pol-muxed RZ-DPSK and 100-Gb/s pol-muxed RZ-DQPSK channels, respectively. The RF power decreases, which results in a limited change for monitoring, when DGD becomes larger than one bit period of the pol-muxed data channel. The monitoring window is determined by the first local maximum (i.e., 25 and 40 ps). The dynamic range of the RF power change becomes smaller when misaligned by 5 degrees, particularly in the low DGD region, which is attributed to the fact that the crosstalk resulting from the misalignment induces the power increment and then affects the dynamic range contributed by DGD. The RF power change will be further reduced as the misaligned angle becomes larger till 45 degrees. The alignment between the pol-muxed data channel and polarizer is important to suppress the crosstalk and then enhance the monitoring sensitivity. The commercial polarization controller can provide ± 0.9-degree accuracy, which will be sufficient for the angle alignment. This PMD monitoring technique is potentially applicable to higher bit-rate pol-muxed data channels, and the monitoring range of DGD is scaled to the bit rate as shown in Fig. 4.

 

Fig. 4 Simulated results of RF power change with the alignments for pol-muxed RZ-D(Q)PSK channels.

Download Full Size | PDF

5. Conclusion

A CD-insensitive PMD monitoring technique using a polarizer and a low-speed detector is presented for an 80-Gb/s pol-muxed RZ-DPSK data channel. Measured RF power increment of 16.2 dB, is measured in the presence of the increasing DGD from 0 to 12 ps. The measurements are less sensitive to CD up to 100 ps/nm. This technique does not require high-speed components for monitoring the accumulated PMD on the pol-muxed data channel, which is potentially applicable to the higher speed pol-muxed data channels.