Real-Time Unrepeated Long-Span Field Trial over Deployed 4-Core Fiber Cable Using Commercial 130-Gbaud PCS-16QAM 800 Gb/s OTN Transceivers

Abstract

The space-division multiplexed (SDM) transmission technique based on uncoupled multi-core fibers (MCF) shows great implementation potential due to its huge transmission capacity and compatibility with existing transceivers. In this paper, we demonstrate a real-time single-span 106 km field trial over deployed 4-core MCF cable using commercial 800 Gb/s optical transport network (OTN) transceivers. The transceivers achieved a modulation rate of 130 Gbaud with the optoelectronic multiple-chip module (OE-MCM) packaging technique, which enabled the adoption of a highly noise-tolerant probability constellation shaping a 16-array quadrature amplitude modulation (PCS-16QAM) modulation format for 800 Gb/s OTN transceivers, and could realize unrepeated long-span transmission. The 4-core 800 Gb/s transmission systems achieved a real-time transmission capacity of 256 Tb/s with fully loaded 80-wavelength channels over the C+L band. The performance of different kinds of 800 G OTN transceivers with different modulation formats under this long-span unrepeated optical transmission system is also estimated and discussed. This field trial demonstrates the feasibility of applying uncoupled MCF with 800 Gb/s OTN transceivers in unrepeated long-span transmission scenarios and promotes its field implementation in next-generation high-speed optical interconnection systems.

1. Introduction

The optical transmission network is undergoing a capacity upgrade with the explosive growth of global business. Multiple capacity-enhancement techniques such as high-order modulation format, broadband wavelength-division multiplexing, and polarization-division multiplexing have been widely investigated and applied, effectively meeting the ever-growing demand for network traffic [1,2,3]. However, due to the limited low-loss wavelength windows and nonlinear effects in single-mode fibers (SMFs), the capacity of a standard SMF has approached its nonlinear Shannon limit [4]. Although ultra-low-loss large-effective-mode-area fibers can mitigate the nonlinear impact and enhance the transmission capacity, it is still a formidable challenge to keep up with the exponentially growing traffic demands of future ultra-high-speed coherent optical transmission networks [5,6]. Recently, the space-division multiplexed (SDM) transmission technique has attracted extensive research interest and is considered a promising candidate for next-generation ultra-high-speed optical transmission networks due to its huge capacity potential. So far, there are two main technical routes for achieving SDM optical transmission. One is the mode-division multiplexed (MDM) transmission approach based on few-mode fibers (FMFs) or multi-mode fibers (MMFs), and the other is the core-division multiplexed (CDM) transmission technique utilizing multi-core fibers (MCFs). The fiber modes or fiber cores in SDM fibers can serve as multiplexed transmission channels and can significantly enhance the transmission capacity of a single fiber [7,8]. For the MDM transmission technical route, a key issue is the difficulty in suppressing the modal crosstalk among different spatial modes. This crosstalk severely affects the quality of the received signals. To address the issue of modal crosstalk, it is necessary to utilize multiple-input multiple-output digital signal processing (MIMO-DSP) at the receiver or use only a limited number of circularly symmetric modes to achieve ultra-high-speed transmission, which means there is still some way to go before this technique can be put into practical application [9,10]. In contrast, for the CDM transmission technique, the inter-core crosstalk can be easily suppressed by designing the structure of MCFs [11]. Besides, the CDM transmission system can be made directly compatible with existing intensity modulation/direct detection (IM/DD) or coherent transceivers through fan-in/fan-out (FIFO) devices, which means it exhibits excellent implementation potential in both next-generation short-reach and long-haul optical transmission systems [12,13,14].

For the large-scale implementation of the CDM transmission approach in existing optical transmission networks, MCF-related maintenance techniques such as low-loss connection schemes and high-performance FIFO devices are crucial issues to be addressed. Over the past few years, multiple MCF-related maintenance techniques have witnessed remarkable progress. For low-loss connection approaches, ultra-low-loss fusion splicing techniques utilizing advanced automatic core misalignment algorithms and portable MCF connectors enabled by precise micromachining have been experimentally demonstrated, promoting the field implementation of MCFs in long-haul transmission scenarios [15,16]. Regarding FIFO devices, a wide variety of low-loss FIFO devices have been reported by improving processing approaches such as using micro-structured preform and reduced-cladding graded index fiber, which can support CDM transmission with multiple types of MCF [16,17,18]. Besides, integrated multi-core erbium-doped fiber amplifiers (MC-EDFAs) with core pumping have also been reported, based on which an ultra-long-haul transmission over more than 12,000 km 4-core fiber has been achieved [19]. Real-time CDM transmission utilizing commercial 400 Gb/s optical transport network (OTN) transceivers has also been demonstrated [14]. These remarkable breakthroughs promote the maturity and field implementation of MCFs, and experimental MCF cables have also been field-deployed and trialed multiple times to further explore the application scenarios of MCFs. For example, a 114 Pbit/s∙km transmission using three vendor-installed 60 km 4-core fiber spans has been reported, demonstrating the implementation feasibility of 4-core fibers in the actual outdoor environment with multiple fusion splicing and connector connections [20]. Besides, an integrated communication and sensing system over field-deployed 7-core fiber cable has also been demonstrated, which can monitor the surrounding traffic while maintaining large-capacity CDM communication [21]. Thanks to the inherent advantage of MCFs with multiple fiber core channels, various types of service, such as advanced MIMO distributed acoustic sensing (DAS) and classical/continuous variable quantum key distribution (CVQKD) co-existence transmission systems, have also been reported [22,23]. MCFs also have the potential to be compatible with other advanced transmission technologies such as high-stability optical time and frequency transmission systems [24]. However, for implementation in long-span transmission scenarios such as unrepeated optical interconnections, unrepeated field trials over more than 100 km MCF cable utilizing commercial 800 Gb/s OTN transceivers have not been reported so far, with the main restriction being the performance of MCF cable and OTN equipment. Exploring the high-speed single-span transmission capabilities of MCFs is of great significance, especially for data center interconnections and submarine cable communication systems.

In this paper, based on our previous real-time CDM field trial utilizing 400 Gb/s OTN transceivers [25], we demonstrate the first real-time single-span 106 km CDM field trial using commercial 130 Gbaud 800 Gb/s OTN transceivers over deployed 4-core MCF cable. The 800 Gb/s OTN transceivers adopted a highly noise-tolerant probability constellation shaping 16-array quadrature amplitude modulation (PCS-16QAM) modulation format, which was enabled by the high modulation rate. The adopted modulation format had a greater tolerance to the impacts of span loss compared to previous higher-order formats such as PCS-64QAM and thus enabled the unrepeated long-span transmission. Experimental results show that the 4-core 800 Gb/s CDM transmission system can achieve a real-time capacity of 256 Tb/s with fully loaded 80-wavelength channels over the C+L band. The performance of different kinds of 800 G OTN transceivers with different modulation formats under this long-span unrepeated optical transmission system are also estimated and discussed. To the best of our knowledge, this is the first real-time unrepeated single-span field trial over more than 100 km MCF cable, and it promotes the field implementation of MCFs with 800 Gb/s OTN transceivers in next-generation high-speed optical interconnection systems.

The rest of this paper is organized as follows. In Section 2, we introduce the field-deployed 4-core MCF cable. Section 3 presents the experimental setup of the real-time CDM field trial using 800 Gb/s OTN transceivers over the deployed MCF cable. In Section 4, the experimental results of the field trial are presented. More in-depth discussions and prospects regarding the 800 Gb/s OTN transceivers and the 4-core MCF cable are provided in Section 5. Finally, Section 6 concludes the paper.

2. The Field-Deployed 4-Core MCF Cable

The MCF cable, which is deployed in Jinan, China, with the routing shown in Figure 1a, connects the Jinxiu Chuan and Xiying data centers and has a length of 17.69 km. The entire MCF cable was laid through a combination of direct burial, overhead installation, and pipeline laying, and is formed by splicing 11 segments of short cables. The length of each segment of short cable is depicted in Figure 1c. It should be emphasized that the length of the MCF cable in our field trial was slightly longer than when it was first deployed, and the number of splicing points has also increased. This is mainly because it was damaged by construction work before this field trial and repaired by splicing a spare optical cable. The MCF cable contains eight 4-core fibers, eight 7-core fibers, four G.652.D SMFs, and four G.654.E SMFs, which is equivalent to a total of 96 fiber cores. In our work, 4-core fibers, whose cross-section is shown in Figure 1b, were utilized to conduct the field trial. The diameters of the cladding and the coating of the 4-core fiber are 125 μm and 245 μm, respectively, consistent with those of standard SMFs, which enables the use of standard fiber fabrication and cabling processes. The core-to-core distance is 42.5 μm, which is enough to effectively suppress the crosstalk between the different fiber cores. Each fiber core in the 4-core fiber complies with the ITU-T Recommendation G.652, and low-refractive-index fluorine-doped trenches are applied around each fiber core to further reduce the inter-core crosstalk and bending loss. To ensure the precise alignment of the fiber cores at the two cross-sections during the fusion splicing and to avoid misconnections between different fiber cores, a tiny marker core is also processed in the MCF fiber as depicted in Figure 1b. In addition to the parameters of the MCF already mentioned above, the other main characteristics of the 4-core fiber after cabling at 1550 nm are presented in

Table 1. Main characteristics of the 4-core fiber after cabling.

Mode Field Diameter (μm)	Dispersion Coefficient (ps/nm/km)	Attenuation Coefficient (dB/km)	Polarization Mode Dispersion (ps/km1/2)	Cut-Off Wavelength (nm)	Inter-Core Crosstalk (dB/10 km)	Bending Loss with R = 30 mm (dB/100 Turns)
9.8	≤22	≤0.21	≤0.08	≤1300	≤−50	<0.1

. The characteristics of the 4-core fiber are similar to those of SMFs. The inter-core crosstalk is rather low and has little impact on the transmission performance.

In our field trial, to achieve unrepeated long-span transmission, we cascaded six of the eight 4-core fibers in the MCF cable as depicted in Figure 1c. This cascading approach enabled the establishment of a single-span 106 km MCF link and allowed the OTN transceivers to be installed in the same data center. The entire 106 km 4-core fiber link contains 65 multi-core fusion splicing connections, which is equivalent to 260 single-core fusion splicing connections. We measured the splicing loss of each single-core fusion splicing at 1550 nm using a high-resolution optical time-domain reflectometer (OTDR) and a FIFO device. In this process, we utilized the OTDR to launch a pulsed optical signal into each fiber core of the 4-core fiber one by one through the FIFO device, and we obtained the splicing loss of each single-core fusion splicing from the reflected backscattering curve. The results show that the splicing losses of the 260 single-core fusion splicing connections are no more than 0.4 dB, and the average splicing loss of the entire MCF link is lower than 0.15 dB. We found that the splicing loss of MCFs was larger than that of SMFs, which was mainly caused by the slight alignment errors between the fiber cores of the two cross-sections during the fusion splicing process.

The span loss and the total inter-core crosstalk of the 106 km 4-core CDM link were also measured before the field trial, and the results are shown in

Table 2. Span loss and total inter-core crosstalk of the 106 km 4-core CDM link (unit: dB).

	Core #1	Core #2	Core #3	Core #4
Span loss	33.6	33.3	32.9	33.0
Total crosstalk	−41.4	−44.6	−42.1	−45.3

. The 4-core CDM link consists of a 106 km 4-core fiber and a pair of FIFO devices. The total inter-core crosstalk of each fiber core is defined as the sum of the inter-core crosstalk from the other three fiber cores to the tested core. It was found that the span losses of the four fiber cores were similar and were all no more than 33.6 dB. The total inter-core crosstalk was lower than −40 dB for each fiber core of the 106 km CDM link.

3. Experimental Setup of the CDM Transmission System

In this section, we introduce the experimental setup of the CDM field trial using 800 Gb/s OTN transceivers over deployed MCF cable, which is shown in Figure 2. At the transmitter station, four sets of 80-wavelength C+L band OTN transmitters were utilized to generate four channels of WDM signals for CDM transmission in the 4-core fiber. For each set of OTN transmitters, a pair of C-band and L-band 800 Gb/s optical transponder units (OTU) was employed to generate modulated optical signals for transmission testing. The 800 Gb/s OTUs have a modulation rate of up to 130 Gbaud, which is enabled by the OE-MCM packaging technique. In this packaging approach, the driver is directly mounted on the photonic integrated circuits (PICs) through flip-chip welding, and the DSP die as well as the PICs are packaged on the same substrate, which can effectively reduce the length of high-speed signal transmission and thus ensure enough modulation bandwidth. Thanks to the high modulation rate, the 800 Gb/s OTUs can adopt a more noise-tolerant PCS-16QAM modulation format instead of the previous PCS-64QAM modulation format to achieve the target bit rate, which makes them more suitable for long-span unrepeated transmission and long-haul transmission. The central wavelengths of the C-band and L-band OTUs are tunable within the ranges of 1524.89 nm to 1571.65 nm and 1575.78 nm to 1625.77 nm, respectively, with a channel spacing of about 1.2 nm (150 GHz). The OTN transmission system can transmit up to 80 wavelength channels simultaneously, with 40 C-band and 40 L-band OTUs for each fiber core. Due to the lack of enough OTUs, dummy light (DL) generated by the C-band and L-band amplifier spontaneous emission (ASE) noise sources was utilized to fill the remaining wavelength channels of the C+L band. In addition to simulating fully loaded 80-wavelength transmission, the filling wavelength channels can also maintain the stability of the broadband transmission system under the influence of stimulated Raman scattering (SRS), which is crucial for practical broadband WDM transmission systems where the add/drop operations of wavelength channels occur frequently. This is mainly because the SRS will cause power transfer from short-wavelength signals to long-wavelength signals, and the power transfer is related to the filling situation of the wavelength channels [5]. In practical broadband transmission systems with frequent add/drop operations of wavelength channels, the SRS will cause a dynamic power change of the received signals at different wavelengths. Filling all the vacant wavelength channels using DL keeps the power transfer caused by the SRS stable, and the impact of the SRS can be balanced through power pre-compensation. The high-speed modulated signals and the DL were multiplexed through a C+L band integrated wavelength selection switch (WSS), and a pair of C-band and L-band EDFAs were followed to amplify the signals. The C-band and L-band amplified signals were then multiplexed by an optical band multiplexer (OBM) and further coupled to the 4-core fiber through a FIFO device. At the receiver station, four sets of C+L band 800 Gb/s OTN receivers were employed to demultiplex and coherently detect the four channel-received signals from the 4-core CDM transmission link. The long-span unrepeated transmission link was the above-mentioned 106 km CDM link, which consists of a segment of single-span 106 km 4-core fiber and a pair of FIFO devices. For each set of OTN receivers, a pair of C-band and L-band EDFAs were first utilized to amplify the received signals. The separation of the received C-band and L-band signals was also achieved through an OBM, and another C+L band integrated WSS was followed to demultiplex the modulated signals. The demultiplexed C-band and L-band signals were then sent to the C-band and L-band OTUs, respectively, to conduct coherent detection and perform a real-time bit error rate (BER) calculation. In our field trial, experimental results such as bit error rate, optical power, and optical spectrum could be obtained in real time from the network management system. Since the performance of our experimental system remained relatively stable during the test, the experimental results were directly obtained after the system was stabilized. The 800 Gb/s coherent OTUs utilized in the field trial belong to the ZTE S3F-E series optical transmission equipment, specifically the LB4TF model optical line board. The OTN transmission system was installed on a ZXONE 19700 platform, which was manufactured by ZTE Corporation in Shenzhen, China. And the network management system was a ZTE ElasticNet UME system. The OTN transmission equipment and the network management system are all placed in the Jinxiu Chuan data center as depicted in Figure 2. We conducted the experiment on-site at the Jinxiu Chuan data center.

4. Experimental Results of the Field Trial

In this section, we further introduce the experimental results of our field trial utilizing 800 Gb/s OTN transceivers with the system setup shown in Figure 2. The BER values under different optical signal-to-noise ratios (OSNRs) at the receiver were first measured to evaluate the OSNR threshold of the 800 Gb/s CDM system. The OSNR was measured according to Equation (1) defined by ITU-T Recommendation G.Sup39:

$$OSNR=10log{\displaystyle \frac{P_i}{N_i}}+10log{\displaystyle \frac{B_m}{B_r}},$$

(1)

where P_i is the optical signal power of the i-th channel and N_i is the ASE noise power of the i-th channel measured within the noise equivalent bandwidth B_m. Both P_i and N_i are expressed in the linear unit of watt. B_r is the reference optical bandwidth, which is 0.1 nm, corresponding to 12.5 GHz in the C-band. The OSNR at the receiver was adjusted by adding white noise through an ASE noise source. The measured results at 1549.72 nm and 1589.57 nm are shown in Figure 3a and Figure 3b, respectively. The results in the back-to-back (B2B) configuration are also illustrated for reference. We found that the C-band and L-band signals showed similar BER-OSNR characteristics. Under the soft-decision forward error correction (SD-FEC) limit, which was 3.3 × 10⁻², the OSNR threshold was lower than 23.2 dB for each fiber core channel under 106 km transmission. This relatively low OSNR threshold indicates that the CDM system utilizing 800 Gb/s OTN transceivers has good noise tolerance, which is mainly owing to the PCS-16QAM modulation format. Compared to the results in the B2B configuration, the degradation of the OSNR threshold induced by fiber transmission was no more than 0.3 dB, which demonstrates that the coherent receiver can compensate well for the impacts brought about by fiber transmission. However, it needs to be acknowledged that compared with the results in our previous work, which was conducted over the same MCF cable with dual-polarization quadrature phase shift keying (DP-QPSK) 400 Gb/s OTN transceivers, the OSNR threshold has increased by approximately 7.2 dB [25]. This is mainly because the noise tolerance of the PCS-16QAM modulation format is worse than that of the DP-QPSK modulation format.

We further investigated the BER values of each fiber core at the 80 wavelength channels by adjusting the central wavelengths of the OTUs at both the transmitter and receiver stations. The single-wavelength input power of the C-band and L-band signals was +11 dBm and +9 dBm, respectively. The input power of the C-band signal was 2 dB higher than that of the L-band signal, which is mainly because in our field trial, under an input power of around +10 dBm, approximately 2 dBm of power transferred from the C-band signals to the L-band signals induced by the SRS. Therefore, under the same input power, the received power of the C-band signals would be approximately 4 dB lower than that of the L-band signals. Further, considering that the noise figure of the L-band amplifier is approximately 2 dB worse than that of the C-band amplifier, setting the input power of the C-band signals to be 2 dB higher than that of the L-band signals will approximately balance the transmission performance of the signals in the two bands. The experimental results are shown in Figure 4. It was found that the BERs all remained lower than the SD-FEC limit for each fiber core over the C+L band after unrepeated 106 km CDM transmission. These results indicate that the PCS-16QAM 800 Gb/s OTN transmission system is capable of tolerating a span loss of over 33.6 dB and achieving long-span unrepeated optical transmission. While different from our previous field trial using DP-QPSK 400 Gb/s OTN transceivers over 106 km 4-core fiber cable, which still reserved a greater than 5.5-dB OSNR margin for each fiber core and wavelength channel, the worst channel (the channel at 1524.89 nm of the first fiber core) of the 800 Gb/s CDM-WDM transmission has already approached the FEC limit and there is insufficient OSNR margin left [25]. It should be noted that the experimental results are not only related to the impacts of fiber transmission such as span loss, inter-core crosstalk, and SRS, but also related to the performance of the OTN equipment such as the OTUs, WSS, and EDFAs. The performance of signals at lower wavelengths within the C-band is somewhat worse, which is mainly caused by the performance of the OTUs and the EDFA, as well as the impact of the SRS. Compared to the previous results using 400 Gb/s OTUs, the BER performance of different core channels seems to be different, which is also mainly due to the performance differences between the 400 Gb/s and 800 Gb/s OTUs. The fiber core channel employed with better OTUs usually has better transmission performance. This field trial also demonstrates the feasibility of real-time 256 Tb/s (800 Gb/s × 80 × 4) unrepeated 106 km CDM-WDM transmission over deployed 4-core MCF cable. Compared to previous experimental results using 400 Gb/s OTN transceivers, which achieve real-time bit rates of 128 Tb/s (400 Gb/s × 80 × 4) and 224 Tb/s (400 Gb/s × 80 × 7) over deployed 4-core and 7-core MCF cable, respectively, this work demonstrates a higher bit rate [25]. This improvement is mainly attributed to the higher spectral efficiency of the 800 Gb/s OTN transceivers with the PCS-16QAM modulation format. The received normalized optical spectrums of the 800 Gb/s 4-core 80-λ transmission system are shown in Figure 5. We found that the power of the L-band signals at the receiver was slightly higher than that of the C-band signals, which is also caused by the SRS.

5. Discussion and Technological Prospects

5.1. Discussion and Technological Prospects of the 800 Gb/s OTN Transceivers

In our field trial, we demonstrated a real-time single-span 106 km CDM field trial utilizing commercial 800 Gb/s OTN transceivers over deployed 4-core MCF for the first time. These results are mainly due to the fact that the 130 Gbaud 800 Gb/s OTN transceivers utilizing the PCS-16QAM modulation format are more noise-tolerant and capable of withstanding larger span loss compared to previous PCS-64QAM 800 Gb/s OTN transceivers. To further discuss the differences of 800 Gb/s OTN systems using different modulation formats, the performance of single-span 106 km CDM transmission systems using different modulation formats at 1549.72 nm was evaluated based on the existing experimental and simulation results, which are shown in

Table 3. Performance of single-span 106 km CDM transmission systems using different modulation formats.

Bit Rate (Gb/s)	Baud Rate (Gbaud)	Modulation Format	OSNR Threshold (dB)	Input Power of C-Band Signals (dBm)	OSNR After Transmission (dB)	OSNR Margin (dB)
400	130	QPSK	16	+11	25.6	9.6
800	260	QPSK	19.1	+14	28.6	9.5
800	130	PCS-16QAM	23.2	+11	25.6	2.4
800	118	16QAM	24.6	+10.6	25.2	0.6
800	98	PCS-64QAM	27.4	+9.8	24.4	−3

. The previous experimental results using 130 Gbaud QPSK 400 Gb/s OTN transceivers are also presented for reference [25]. The span loss was assumed to be 33.6 dB, which is the maximum span loss. The single-wavelength input power of the C-band signals was set to be +11 dBm, which is consistent with our experiments. Moreover, for better comparison, the single-wavelength input power of other 800 Gb/s signals using different modulation formats was adjusted to ensure that the power spectral density was consistent with that in our experiment. The input power of the L-band signals was still 2 dBm lower than that of the C-band signals. The OSNR thresholds were obtained from the field trial and previous laboratory tests. It should be emphasized that the OSNR margins presented in Table 3 are the results at 1549.72 nm, which should generally be greater than 2 dB to ensure that the signals in the entire C+L band can be transmitted. This is because there are inevitable differences in the performance of the signals in the C+L band. From Table 3, we can obtain that the 400 Gb/s and 800 Gb/s signals using the QPSK modulation format have the maximum OSNR margin, which indicates that they have the capability to transmit over longer distances. With the help of previous experimental results, i.e., [25], the entire C+L band signals all reserve an OSNR margin of more than 5.5 dB after 106 km transmission, so the transmission distance could be further extended by at least 15 km. Meanwhile, the 800 Gb/s signals using the 16QAM and PCS-64QAM modulation formats cannot achieve a single-span 106 km transmission in the experimental environment of a field trial due to the lack of sufficient OSNR margin, which demonstrates the performance advantage of the 130 Gbaud PCS-16QAM 800 Gb/s OTN transceivers in single-span unrepeated transmission. Therefore, to achieve longer-distance unrepeated 800 Gb/s CDM transmission over MCF cable, it is necessary to conduct research on higher-speed modulators and promote the application of QPSK 800 Gb/s OTN transceivers with a 260 Gbaud modulation rate. While the cost and single-bit power consumption need to be considered comprehensively, the PCS-16QAM 800 Gb/s OTN transceivers may be a better choice in most scenarios such as metropolitan area networks.

5.2. Discussion on Further Improving the Performance of MCF Cables and Their Future Applications

Through our field trial, we also found that the splicing loss of 4-core fiber was too large compared to that of SMF, which thereby increased the span loss of the field-deployed MCF cable. By reducing the splicing losses of MCFs, the transmission distance of the single-span unrepeated transmission in the field trial could be further extended. Thus, in our future work, we will further reduce the splicing losses of MCFs in the following ways to promote their implementation in existing networks. Firstly, it is necessary to optimize the MCF itself. By appropriately designing MCFs with a larger effective mode field area, the splicing losses caused by misalignments during the splicing process can be reduced. Moreover, the alignment accuracy during the splicing process will be improved to reduce the splicing losses. This requires improving the monitoring ability of the MCF interface and enhancing the automatic alignment algorithm during the splicing process. Additionally, considering the fact that the relatively large splicing losses of MCFs are also related to the uneven heating of the fiber cores during the arc discharge process, it is important to optimize the approach of arc discharging, such as by using a 3-electronic arc-discharging system [14]. Finally, as regards the field implementation, it is recommended that we splice the MCFs produced in the same batch to further reduce the misalignments between different fiber cores. It can easily be evaluated that if the average splicing loss of each single core is further reduced by 0.1 dB, the span loss of each fiber core channel of the 106 km MCF link can be reduced by approximately 6.5 dB. Under the same conditions, the transmission distance can be extended by about 25 km. In addition to reducing the splicing losses, the span loss can be reduced by optimizing the MCFs, such as by designing ultra-low-loss MCFs, which will further extend the single-span unrepeated transmission distance. The high-speed CDM transmission system in our field trial has the potential to be practically implemented in existing network scenarios, such as high-speed long-span metropolitan area networks or high-speed unrepeated interconnections between data centers, which can greatly improve the transmission rate of data information and reduce the number of amplifiers at the same time. It should also be noted that in order to scale the system toward widespread deployment, the splicing time of MCFs also needs to be further reduced to ensure that the MCFs can be repaired in a timely manner when they are broken.

6. Conclusions

In conclusion, we have demonstrated a real-time single-span 106 km CDM field trial using commercial 130 Gbaud PCS-16QAM 800 Gb/s OTN transceivers over deployed 4-core MCF cable. Enabled by the high modulation rate and the noise-tolerant modulation format, the 800 Gb/s OTN transceivers could tolerate a span loss of over 33.6 dB and achieve long-span unrepeated optical transmission. The high baud rate was enabled by the optoelectronic multiple-chip module (OE-MCM) packaging technique, which aims to minimize the path of high-speed signals and thus ensure enough modulation bandwidth. The 4-core 800 Gb/s CDM transmission system achieved a real-time capacity of 256 Tb/s with fully loaded 80 wavelength channels over the C+L band. The transmission distance of this single-span unrepeated field trial could be further extended by reducing the splicing losses of the MCF cable and increasing the modulation rate of the modulator. To the best of our knowledge, this work is the first real-time unrepeated single-span field trial over more than 100 km MCF cable with commercial 800 Gb/s OTN transceivers, and we believe it will be beneficial to next-generation high-speed optical interconnection systems.