We investigate the performance of a self-homodyne coherent detection (SHCD) system using a 19 core multi-core fiber (MCF) and 16 wavelength-division-multiplexed channels. We show that SHCD, with the pilot-tone transmitted on a single MCF core and information carrying signals on the remaining cores, is compatible with space-division-multiplexed transmission, potentially relaxing laser linewidth and digital signal processing requirements due to phase noise cancellation. However, inter-core crosstalk can have an impact on performance and core selection.
©2013 Optical Society of America
Self-homodyne coherent detection (SHCD) has long been proposed and investigated as a useful technique to exploit phase noise cancellation, thus reducing receiver complexity and relaxing requirements for narrow linewidth lasers and high-speed digital signal processing (DSP) [1,2]. Previously, SHCD with a pilot-tone transmitted on an orthogonal polarization to the data and used as the local oscillator (LO) signal at the receiver has been demonstrated for a range of multi-level modulation formats [3–8], but suffers from a loss of spectral efficiency of 50% compared to polarization multiplexed systems. Although the spectral efficiency maybe improved by spectrally interleaving the pilot-tone with the signal , recent work on space-division-multiplexing (SDM) [9–12] has opened up the possibility of employing SHCD with the pilot-tone transmitted through 1 SDM channel and the remainder used for signal channels. In an SDM system, it is envisaged that different channels experience the same environmental disturbances during transmission and thus path length variations between channels, are minimized. This is advantageous for SHCD which relies on matching the path length between pilot-tone and information channels for strong phase noise cancellation. Another advantage of utilizing SDM systems with SHCD is that the spectral efficiency cost is inversely proportional to the number of SDM channels. For example, using a 19-core fiber results in a 5.3% reduction of spectral efficiency compared to an equivalent system with intradyne detection (ID), although this figure rises to 14.3% for a 7-core fiber.
Recently, in combination with a 19-core multi-core-fiber (MCF) and free-space coupling system , previously used for 305-Tb/s SDM transmission , we investigated the feasibility of combining SHCD with MCF . Here, we expand on those results with all measurements repeated and extended using an improved experimental set-up of a SHCD system with quadrature-phase-shift-keyed (QPSK) signals on 16 wavelength-division multiplexed (WDM) channels. We transmit the pilot-tone on a single core of the 19-core MCF with the remaining 18 cores used to transmit information carrying signals. We show that, in an MCF based SDM system, phase noise cancellation inherent in SHCD both relaxes the requirement for narrow linewidth lasers and can reduce the amount of DSP required at the receiver enabling cost and power savings. We investigate the transmission penalty of both SDM and WDM cases and finally assess the impact of inter-core crosstalk on system performance. We show that inter-core crosstalk must be considered as a significant, but tolerable impairment, particularly when the pilot-tone signal is transmitted on the center core, and minimizing it is advantageous for optimum performance.
2. Experiment description
The experimental set-up is shown in Fig. 1 . An external-cavity tunable laser (ECTL) with 1- MHz linewidth at 193.5 THz was used for measurement signals. In WDM operation, carriers from the ECTL and 15 distributed feedback (DFB) lasers on the 100-GHz ITU grid between 192.8 THz and 194.3 THz were multiplexed together in a temperature controlled arrayed-waveguide-grating (AWG) at the input to a 3-dB coupler used to split carriers for data modulation from those to be transmitted as the pilot-tone. The additional input port of the coupler was used for an intensity modulated signal used to align the total optical path lengths traversed by the measurement signal and pilot-tone. The signal path was then amplified in an erbium-doped-fiber amplifier (EDFA) and its polarization adjusted at the input to a single polarization QPSK modulator driven by two 101-bit decorrelated 215-1 bit PRBS signals for I and Q at 5 Gbaud. The total data throughput of each 10 Gb/s channel of 16 wavelengths and 18 fiber cores was therefore 2.88 Tb/s or 2.68 Tb/s allowing a 7% overhead for forward error correction. The baud rate was intentionally kept low in order to increase the impact of phase noise in BER measurements using the ECTL. The high power arm of the 10 dB coupler was connected to The input of a 1x20 splitter which split the signal between 17 of the input fibers of the SDM MUX and subsequently in to 17 different cores of the MCF. These acted as dummy data channels and the remaining 3 ports of the splitter were used for power and spectrum monitoring. The low power output of the coupler was connected directly to the SDM MUX input of an outer core used to transmit the decorrelated signal channel upon which BER measurements were made. Variable optical attenuators (VOAs) were used on both coupler outputs to control the fiber launch power of both the dummy MCF cores and the signal channel. The WDM pilot-tone was then connected to the remaining MCF core via the SDM MUX with 3 different cores selected during the measurements. The pilot-tone path also contained an EDFA and VOA with some additional fiber and an optical delay line used to align the optical path lengths. Previously , measurements were performed over a range of temperatures which led to variations in the coupling efficiency. For the BER measurements reported here, thermal isolation was used to minimize thermal deviations and care was taken to ensure minimal variations of coupling efficiency, particularly on the center core which is most sensitive to crosstalk [10, 11].
After transmission across the 19 core MCF, the SDM-DEMUX was used to separate the WDM pilot-tone and measurement channel in to separate fibers for reception and the dummy MCF cores were terminated. The measurement signal was then amplified and filtered by a 1 nm optical band-pass filter (OBPF) to select the wavelength for reception. After a further EDFA, the signal was combined with the output of an amplified spontaneous emission (ASE) source constructed from 2 EDFAs either side of a 2 nm OBPF. In contrast to the filters used in previous measurement , these filters were selected to have low polarization dependent loss (PDL) to ensure OSNR measurements accuracy. Another 2 nm OBPF was used after the noise loading stage and an additional 0.3 nm OBPF was used on the pilot-tone path. For both signal and pilot-tone paths, VOAs and optical taps were used to control and monitor optical power which were both set to 5 dBm at the input to an optical modulation analyzer (OMA). It should be noted that since the LO power at the receiver should be shared between each signal core, a more realistic demonstration should include an additional 12.55 dB loss before reception. To minimize the OSNR penalty on the LO, It is envisaged that this split would occur after the EDFA in pilot-tone receiver path with a second EDFA added afterwards if required.
The OMA was used for constellation analysis and BER measurements and sampled at 40 GS/s with a 13 GHz analogue bandwidth. Chromatic dispersion compensation and polarization tracking were also performed in the OMA for all measurements and the impact of the phase tracking algorithm (PTA) was quantified by BER measurements for ID and SHCD. ID measurements were performed with the OMA’s internal laser with a linewidth of approximately 100 kHz and compared to SHCD with the WDM pilot-tone transmitted across 3 different MCF cores. Since path length is a critical issue for SHCD systems using higher order modulation formats, the relative variation of path length was monitored over temperature variations of 4°C before BER measurements were performed. The largest variation of optical length between pilot-tone and signal path observed was 0.8 cm or 40 ps over the whole transmission system. This was without any active compensation and included optical path length drift from other components. For example, after splitting from the pilot-tone, the signal path included 4 EDFAs, modulator, PCs, VOAs, OBPFs and connecting fibers, in addition to the MCF and coupling system.
3. Experimental results
3.1 Homodyne vs. intradyne detection
Initially, the phase noise cancellation was verified with both a single wavelength and WDM signals transmitted on an outer core of the MCF and the pilot-tone on the center core without signal light in the dummy MCF cores. Figure 2 shows BER comparison of ID and SHCD for measurements with and without the PTA for both the WDM and single channel case.
Figure 2 shows that the best results are achieved with the receiver PTA employed and using ID. However, there is a penalty of almost 2 dB at BER = 10−3 when ID is used without the PTA. With SHCD employed, phase noise cancellation means that it is possible to achieve almost the same performance without the requirement of a potentially DSP intensive PTA [2–4]. It should be noted that since only the signal was noise loaded, meaning the ASE on the LO was negligible compared to the signal, we did not observe the expected OSNR penalty for SHCD . Also, we note that the BER degradation for ID without the PTA is not as severe as expected for the laser linewidth and baud rate used. This is believed to be due to the proprietary DSP of the OMA which, even without the PTA enabled, estimates the phase at the beginning of each acquired trace in order to rotate the constellation for alignment before hard-detection. This, combined with the small size of the acquired traces (4,000 bits), reduces the impact of phase noise.
3.2 Measuring carrier phase estimation
Next, the impact of laser phase noise cancellation was investigated by employing a standard 64 samples-per-block Viterbi and Viterbi algorithm in order to track the carrier phase. Measurements were taken for center-core pilot-tone transmission with both the path length aligned and with path mis-aligned by removal of the entire MCF from the pilot-tone path to ensure decorrelation. Figure 3 shows the estimated carrier phase evolution across a 200-µs span containing 2x106 bits using both the ECTL used previously and a DFB laser with linewidth measured to be 3 MHz at 193.6 THz, as light sources.
From Fig. 3, it is evident that even in the presence of the various noise sources including inter-core crosstalk, only a residual carrier phase correction is required for both laser types. This result verifies that the laser phase noise is cancelled when employing SHCD with goodpath alignment at the receiver. Without the path length alignment the receiver behaves as in the ID case and phase corrections of over a ± 600 rad range can be observed for both lasers. However, it is worth noting that the phase variations of the narrower linewidth ECTL appear smoother than the DFB case and so would reduce the likelihood of phase slips when using the ECTL with ID. Hence, these results show that regardless of the laser linewidth, in addition to frequency offset estimation, which no longer needs to be tracked since signal and LO originate from the same laser, SHCD can also greatly reduce the estimation time-scale required for carrier phase recovery, thus reducing the amount of necessary receiver DSP logic and consumed power compared to ID.
3.3. Measurement of SDM and WDM penalties
Figure 4 shows the measurements of WDM and SDM penalties taken for SHDC with the WDM pilot-tone transmitted in an outer core of the 19-core MCF, but not neighboring the core carrying the measurement signal. Figure 4(a) shows the measured BER as a function of the OSNR for the system with 1, 4, 8 and 16 channels active and Fig. 4(b) shows the BER vs. OSNR plots as a function of the dummy signal fiber launch power relative to the pilot-tone launch power with signal and pilot-tone powers set to 0 dBm for all measurements.
Figure 4(a) shows that for SHCD there is a small additional penalty for WDM transmission compared to the single channel case. The penalty is less than 0.3 dB over the entire BER range measured and is attributed to additional noise from reduced per-channel EDFA input power on both the pilot-tone and signal channel. Since all measurements used the same MCF launch power, 16-channel WDM transmission causes 12 dB lower per channel power for both the signal and pilot-tone. However, it is also possible that inter-channel non-linear effects had some impact. Figure 4(b) shows the impact of inter-core crosstalk from the dummy signal channels on system performance. With equal launch power for signal, dummy channel and pilot-tone, 1 dB BER penalty is observed at BER = 10−3 compared to the case without dummy signal channels. This penalty is reduced to under 0.1 dB with the dummy power reduced by 8 dB showing that a moderate reduction of the inter-core crosstalk can help significantly reduce any SDM penalty. As observed in previous measurements , it is believed that the penalty can also be reduced by increasing the pilot-tone power relative to the signal power.
3.4 Impact of inter-core crosstalk
Next, the impact of inter-core crosstalk on the performance of the system was investigated. This was done by comparing measurements with and without light in the dummy MCF cores and changing the core used for pilot-tone transmission, since crosstalk was previously observed to vary between cores, with center and inner cores experiencing the most severe crosstalk which may be up to 23 dB below the signal power . Hence, for the same outer core signal channel, 3 pilot-tone positions were investigated. These were a neighboring outer core, non-neighboring outer core and center core, shown as A, B and C respectively in the inset of Fig. 5 , which shows BER curves for all 3 cases with the PTA disabled in each.
Figure 5 reveals that the choice of core for pilot-tone transmission has an impact on system performance. Compared to the single core case, the use of the center core caused the largest penalty of almost 1 dB at BER = 10−3. This penalty was not observed for the same case with ID showing that interference with the pilot-tone is the origin of the performance degradation. The penalty is reduced to less than 0.5 dB when using one of the outer cores for pilot-tone transmission. There was no discernible penalty when the pilot-tone was placed in a neighboring channel to the signal, showing that the presence of the un-modulated pilot-tone does not introduce any additional interference compared to the modulated signal channels. The magnitude of the cross-talk induced penalty was smaller than the 2 dB penalty measured in previous measurements  and it is attributed to the improvements in the experimental set-up including thermal isolation, ensuring optimal coupling of the signal and pilot-tone into the MCF and removing the high PDL filters in the OSNR loading stage.
Overall, these results are intended to demonstrate the feasibility of exploiting the phase noise cancellation of SHCD with MCFs and more generally with SDM systems. Although a 19-core fiber was used in these measurements, SHCD should be possible with any MCF. Although higher core numbers enable greater spectral efficiency by sharing the pilot-tones between more signal channels, the core number and indeed the lengths of fiber over which SHCD is possible will be determined by the OSNR requirements of the receiver. Since both the OSNR-limiting inter-core crosstalk and the OSNR penalty of splitting the pilot-tones increase with core number, a clear trade-off exists to be investigated in future work.
We report a first investigation of the performance of transmission system combining self-homodyne coherent detection with an SDM transmission. Using a 19-core fiber and free-space coupling system, we transmitted a 16 wavelength local oscillator through 1 fiber core and 16 WDM information carrying signal channels in the remaining 18 cores. We showed that self-homodyne detection is compatible with SDM transmission, offering advantages of reduced receiver complexity and potentially enabling the use of lower cost transmitters with broader linewidth and reduced DSP requirements at the receiver. However, to exploit the advantages of self-homodyne detection in SDM transmission systems, it is necessary to consider and, where possible, minimize, the fiber’s inter-core crosstalk to achieve the best performance. Similarly, increasing the relative power of the pilot-tone may, also be considered an option to optimize performance since it is critical to successful reception of all signal channels.
The authors would like to thank M. Kurihara and T. Hashimoto for their technical assistance.
References and links
1. T. Miyazaki, “Linewidth-Tolerant QPSK Homodyne Transmission Using a Polarization-Multiplexed Pilot Carrier,” IEEE Photon. Technol. Lett. 18(2), 388–390 (2006). [CrossRef]
2. P. Johannisson, M. Sjödin, M. Karlsson, E. Tipsuwannakul, and P. Andrekson, “Cancellation of nonlinear phase distortion in self-homodyne coherent systems,” IEEE Photon. Technol. Lett. 22(11), 802–804 (2010). [CrossRef]
3. M. Sjödin, E. Agrell, P. Johannisson, G.-W. Lu, P. A. Andrekson, and M. Karlsson, “Filter optimization for self-homodyne coherent WDM systems using interleaved polarization division,” J. Lightwave Technol. 29(9), 1219–1226 (2011). [CrossRef]
4. M. Sjödin, P. Johannisson, M. Karlsson, Z. Tong, and P. A. Andrekson, “OSNR requirements for self-homodyne coherent systems,” IEEE Photon. Technol. Lett. 22(2), 91–93 (2010). [CrossRef]
5. M. Sjödin, E. Agrell, P. Johannisson, G.-W. Lu, P. A. Andrekson, and M. Karlsson, “Interleaved polarization division multiplexing in self-homodyne coherent WDM systems,” Proc. ECOC '10, Mo.1.3.C (2010).
6. G.-W. Lu, M. Nakamura, Y. Kamio, and T. Miyazaki, “40-Gb/s QPSK and 20-Gb/s PSK with inserted pilot symbols using self-homodyne detection,” Opt. Express 15(12), 7660–7666 (2007). [CrossRef] [PubMed]
7. M. Nakamura, Y. Kamio, G.-W. Lu, and T. Miyazaki, “Ultimate linewidth-tolerant 20-Gbps QPSK-homodyne transmission using a spectrum-sliced ASE light source,” Proc. OFC’07, OThD4 (2007).
8. Y. Kamio, M. Nakamura, and T. Miyazaki, “80-Gb/s 256-QAM signals using phase noise and DGD-tolerant pilot-carrier-aided homodyne detection,” Proc. ECOC’07, P089 (2007).
9. B. Zhu, T. F. Taunay, M. Fishteyn, X. Liu, S. Chandrasekhar, M. F. Yan, J. M. Fini, E. M. Monberg, and F. V. Dimarcello, “112-Tb/s space-division multiplexed DWDM transmission with 14-b/s/Hz aggregate spectral efficiency over a 76.8-km seven-core fiber,” Opt. Express 19(17), 16665–16671 (2011). [CrossRef] [PubMed]
10. W. Klaus, J. Sakaguchi, B. J. Puttnam, Y. Awaji, N. Wada, T. Kobayashi, and M. Watanabe, “Free-space coupling optics for multi-core fibers,” in Proceedings of IEEE Phot. Soc. Summer Topicals, WC3.3 (2012).
11. J. Sakaguchi, B. J. Puttnam, W. Klaus, Y. Awaji, N. Wada, A. Kanno, T. Kawanishi, K. Imamura, K. Inaba, K. Mukasa, R. Sugizaki, T. Kobayashi, and M. Watanabe, “19-core fiber transmission of 19×100×172-Gb/s SDM-WDM-PDM-QPSK signals at 305Tb/s,” Proc. OFC Paper PDP5C.1 (2012).
12. B. J. Puttnam, J. Sakaguchi, W. Klaus, Y. Awaji, J.-M. Delgado Mendinueta, N. Wada, A. Kanno, and T. Kawanishi, “Investigating self-homodyne coherent detection in a 19-core spatial-division-multiplexed transmission link,” Proc. ECOC '12, Paper Tu.3.C.3 (2012).