Open Access
Add to BibSonomy
Abstract
Mode-group-division multiplexing (MGDM)-based intensity modulation direct detection (IM/DD) transmission is an attractive approach to increase the capacity for short-reach optical communication. In this Letter, a simple but versatile scheme of mode group (MG) filtering for MGDM IM/DD transmission is proposed. The scheme is applicable to any mode basis in the fiber, and it satisfies the needs of low complexity, low power consumption, and high system performance. By employing the proposed MG filter scheme, a total raw bit rate of a 152-Gb/s multiple-input-multiple-output (MIMO)-free IM/DD co-channel simultaneous transmit and receive system based on two orbital angular momentum (OAM) MGs, each carrying a 38-GBaud four-level pulse amplitude modulation (PAM-4) signal, is experimentally demonstrated over a 5-km few-mode fiber (FMF). The bit error ratios (BERs) of the two MGs are below the 7% hard-decision forward error correction (HD-FEC) BER threshold at 3.8×10-3, using simple feedforward equalization (FFE). Furthermore, the dependability and robustness of such MGDM links are of great significance. Thus, the dynamic evaluation of BER and signal-to-noise ratio (SNR) for each MG is tested over 210 minutes under different conditions. In the dynamic cases, all the BER results using the proposed scheme can be below 1×10-3, which further confirms the stability and feasibility of our proposed MGDM transmission scheme.
© 2023 Optica Publishing Group under the terms of the Optica Open Access Publishing Agreement
With the ever-growing demand for network capacity, techniques for increasing the transmission speed of fiber-optic communication systems are urgently needed. As one kind of space-division multiplexing (SDM) technique, mode-division multiplexing (MDM) based on few-mode fibers (FMFs) or multimode fibers (MMFs) has drawn considerable attention [1–3]. Potentially, the MDM scheme can significantly increase the capacity of optical systems by multiplying the capacity of a single-mode fiber by the number of co-transmitted modes. However, the increase of mode channel counts is limited by modal cross talk caused by mode-coupling in fiber transmission [4]. To address this concern, coherent detection in conjunction with multiple-input-multiple-output (MIMO) digital signal processing (DSP) is frequently resorted to alleviating the cross talk effect. However, this has resulted in high power consumption and huge computational complexity [5,6]. For short-reach transmission systems, low system cost, high energy efficiency, and simple implementation are critical considerations. Thus, a MIMO-free intensity modulation and direct detection (IM/DD) solution is highly preferred over coherent schemes. Until now, two main types of IM/DD mode-multiplexed fiber links have been reported, including MDM and mode-group-division multiplexing (MGDM) transmission [7,8].
For MDM scenarios, each fiber mode is transmitted and received as an independent channel to carry an optical signal. Weakly coupled FMF or specially designed fiber such as the air core fiber (ACF) and ring core fiber (RCF) based on different types of modal basis sets [linearly polarized (LP) mode, orbital angular momentum (OAM), vector mode (VM), etc.] have been proposed to suppress the mode coupling [7,9–11]. However, in the circular-symmetric fiber, polarization degeneracy still exists, which will lead to mutual coupling and random rotation among degenerated modes traveling in the fiber. As a result, a large power fluctuation occurs at the receiver and the system performance is unavoidably degraded. For MGDM scenarios, modes sharing near-degenerate effective refractive index (${n_{\rm {eff}}}$) in the fiber are treated as one mode group (MG) or one super channel which are received simultaneously. Compared with the MDM scheme, the MGDM scheme is more suitable and promising for MIMO-free transmission from a practical perspective since intra-MG cross talk can be neglected. In the MGDM link, a mode/MG demultiplexer (DeMUX)-based scheme and mode-diversity scheme are two representative reception methods [12–14]. The former is a single-input-single-output (SISO) architecture, in which intra-MG modes are demultiplexed by mode/MG DeMUX and are then detected by a multimode photodetector (PD). However, mode/MG DeMUX depends on special design and is generally compatible with certain types of fiber or mode basis. In addition, relatively high inter-MG cross talk introduced by mode/MG DeMUX will also degrade the performance of the system. Mode-diversity-based MGDM transmission is a single-input multiple-output (SIMO) architecture enabled by the corresponding DSP for signal detection. Here, each intra-MG mode (polarization is excluded) is converted to the fundamental mode and is then detected by a PD with a single mode fiber pigtail (SMF-PD). Since more than one receiving channel is required to handle each degenerate mode within one MG, different channel responses need to be compensated by the corresponding DSP, such as maximal ratio combining (MRC) and SIMO algorithms [14,15]. Accordingly, the cost and complexity of the whole system are inevitably increased. Compared with the SMF-based optical system, the FMF/MMF-supported backbone is more vulnerable to environmental disturbance because external perturbations such as wind, temperature, and mechanical stress/bending will cause random mode mixing within and between MGs [13,16]. Thus, apart from the total capacity and cost per bit, the dependability and stability of the MGDM structure are of great importance, especially when applied in practical experiments and commercial networks. Stable polarization-/mode-independent performance is always pursued. However, the dynamic response and robustness evaluation of fiber-based MGDM communication systems have not been widely investigated and discussed to date.
In this Letter, we propose, demonstrate, and evaluate a simple, cost-effective, and versatile MG filter architecture for full reception of MGDM IM/DD transmission. The MG filter is realized in the optical domain and only a single PD is required to receive the signal carried by each MG. Based on this scheme, by using two OAM MGs (${|l|=0}$ and ${|l|=3}$) and one SMF-PD and one MMF-PD, a simultaneous transmit and receive IM/DD system carrying four-level pulse amplitude modulation (PAM-4) over a 5-km few-mode fiber (FMF) is realized, achieving a total raw bit rate of 152 Gb/s. The bit error ratio (BER) of approximately 4 × 10−5 for MG of ${|l|=0}$ and BER of approximately 2 × 10−4 for MG of ${|l|=3}$ are obtained in the static tests, far below the 7% hard-decision forward error correction (HD-FEC) BER threshold at 3.8 × 10−3. To evaluate the practicality and stability of this transmission, the BER performance of each MG under different conditions over a reasonably long period (210 minutes in total) is assessed dynamically, in which all the BERs using the proposed scheme stay below 1 × 10−3.
In detail, the illustration of MG propagating in an optical fiber is shown in Fig. 1(a). For an certain MG${_n}$ in the MGDM transmission, $m_{n,1}$, $m_{n,2}$, …, $m_{n,K_n}$ are the degenerate modes (polarization is excluded) and $m_{n,k}$ is an arbitrary mode within MG${_n}$. Here, $K_{n}$ represents the number of intra-MG modes belonging to MG${_n}$. Figure 1(b) displays a versatile MG filter architecture for IM/DD MGDM transmission. First, an independent optical signal is loaded on an arbitrary mode ($m_{1,k}$, $m_{2,k}$, …, $m_{n,K}$) of each MG. After mode multiplexing and fiber transmission, to separate MG${_n}$ from other MGs, each degenerate mode of MG${_n}$ ($m_{n,1}$, $m_{n,2}$, …, $m_{n,K_n}$) is demultiplexed to the fundamental mode by the corresponding mode DeMUX and then is fed into the SMF, filtering other MG channels. To realize single output for one MG super-channel without signal interference and loss, these parallel channels of fundamental modes are subsequently converted to $K_{n}$ arbitrary orthogonal modes $m_{i_1, j_1}$, $m_{i_2, j_2}$, …, $m_{i_{K_n}, j_{K_n}}$ by the corresponding mode converters. Note that the orthogonality condition is used to avoid co-channel interference. Finally, these orthogonal modes are multiplexed into one output and coupled into an MMF-PD for signal detection. Accordingly, a filter for MG of MG${_n}$ is realized and this approach will be further clarified with the following experimental demonstration.
Fig. 1. (a) Illustration of mode groups. (b) Schematic diagram of the proposed MG filter scheme for MGDM transmission.
Figure 2 illustrates the experimental setup of MIMO-free two-OAM-MG MGDM transmission enabled by the MG filter scheme. At the transmitter, a 38-GBaud PAM-4 pulse shaped by a 0.01 roll-off factor raised-cosine filter is generated by an arbitrary waveform generator (AWG, Keysight, M8196A) with a 92-GSa/s sampling rate and then is boosted by an electrical amplifier (EA, SHF, S807). Next, it is employed to modulate the optical carrier in the intensity by a Mach–Zehnder modulator (MZM, Fujitsu, FTM7938EZ). The carrier at the wavelength of 1550.12 nm is from an external cavity laser (ECL). After amplification by an erbium-doped fiber amplifier (EDFA), the optical PAM-4 signal is divided into two branches by an optical coupler (OC). One branch is delayed by a long SMF jumper for signal decorrelation and is directly employed as the fundamental mode channel (${l=0}$) after being collimated by a collimator (COL). The other branch is converted to an OAM mode channel of ${l=+3}$ by a vortex phase plate (VPP, RPC Photonics, VPP-m1550). These two channels are multiplexed together by a non-polarizing beam splitter (NPBS, the NPBS can be replaced by a PBS if it is available) and then coupled into a 5-km step-index FMF with a core/cladding diameter of 16/125 $\mathrm{\mu}$m (YOFC, FM2012-B). This fiber can support the LP modes of LP01, LP11, LP21, LP02, LP31, and LP12 at 1550-nm wavelength. Cross talk between two MGs is first measured, with MG isolation $> 21$ dB. Insets (I) and (II) in Fig. 2 display the intensity profiles of two OAM channels captured by a charge-coupled device (CCD) camera after FMF transmission. To detect two MGs simultaneously, the output light beams from the FMF are divided into two parts by a 20:80 NPBS.
- • At the receiver 1 (Rx-1), the ${20\%}$ part is directly coupled into the SMF-pigtailed photodetector integrated with a trans-impedance amplifier (TIA) (PD-1, Finisar, XPRV2022A) with a 3-dB bandwidth of ${\sim }$32 GHz for optical-to-electrical (O/E) conversion, and the detected analog signal is denoted as ${I_1(t)}$. Since higher-order MGs ${(l > {0})}$ will be exhausted by the SMF, only MG of ${|l|=0}$ can be received.
- • At the receiver 2 (Rx-2), an MG filter for OAM-MG ${|l|=3}$ is implemented first. The ${80\%}$ output beams from the FMF are separated into two branches by a NPBS. In each branch, the intra-MG OAM mode ${l=+3/-3}$ is re-converted to the fundamental mode by the corresponding VPP and then is fed into the SMF to filter other higher-order modes. After collimation, to satisfy the mutual orthogonality of two such branches, the upper branch is converted to ${l=-3}$ and the lower branch is still employed as the fundamental mode. Here, these two branches could be theoretically converted to two orthogonal modes. Specifically, these two paths are precisely aligned by a $z$ axis rail below the COL of the lower branch before multiplexing, avoiding multi-path interference from another path. Next, two branches are multiplexed into one output by an NPBS. Accordingly, a SISO MG filter is realized. Then the output light is injected into the MMF-pigtailed PD with TIA (PD-2, Newport, 1544-B) with a 3-dB bandwidth of ${\sim }$12 GHz for O/E conversion, and the detected analog signal is denoted as ${I_2(t)}$.
- • Finally, the electrical signals ${I_1(t)}$ and ${I_2(t)}$ detected by PD-1 and PD-2 are sampled simultaneously by the two-channel oscilloscope (OSC, Keysight DSAZ634A) operating at a 160-GSa/s sampling rate and further processed by the off-line DSP, including synchronization, equalization by the simple T/2-spaced feed-forward equalizer (FFE), and BER decision.
Fig. 2. Setup of a MIMO-free two-OAM-MG based simultaneous transmit and receive MGDM system enabled by the MG filter approach. Insets (I)/(II) show the intensity profile of ${|l|=0/3}$ after 5-km FMF transmission. Inset (III) shows eight scenarios for static and dynamic evaluation.
To further evaluate the transmission performance of this two-MG MGDM system, eight scenarios are considered for static and dynamic performance evaluation as presented by inset (III) of Fig. 2. Here, (S.) and (M.) represent the single-MG (${|l|=0/3}$) transmission and dual-MG transmission (${|l|=0}$ and ${|l|=3}$), respectively. Additionally, ${|l|=0/3}$ and ${l=-3/+3}$ at the receiver represent the full-MG reception using the proposed filter approach and partial-MG reception (single intra-group mode, mode ${l=-3}$ or mode ${l=+3}$) when blocking the path of ${l=+3}$ or mode ${l=-3}$, respectively. Specifically, the simultaneous reception of ${l=0}$ and ${l=-3/+3}$ or ${l=0}$ and ${|l|=3}$ is always adopted if two MGs are co-transmitted. First, the BER performance versus received optical power (ROP) is measured using feedforward equalization (FFE) with different numbers of taps, which is given in Fig. 3. Obviously, FFE with more taps is more beneficial to the system performance at the price of higher computational complexity. BER saturation is almost achieved for both MGs when 40-tap FFE is implemented, which is also adopted in the following experiments. In addition, it can be seen that the required ROPs for ${|l|=0}$ and ${|l|=3}$ are approximately ${-7.7}$ dBm and ${-7.3}$ dBm at a BER of ${3.8\times 10^{-3}}$, with power penalty $< 0.5$ dB for ${|l|=3}$. Furthermore, static BER performance versus ROP is shown in Fig. 4(a). Here, the ROP for 3 (M.) is also employed to represent the ROP for $-$3/+3 (M.) for simplicity. For single-MG transmission, the BER of MG ${|l|=3}$ is worse than that of MG ${|l|=0}$ mainly due to the narrower bandwidth of PD-2. For two-MG multiplexing transmission, BERs of approximately 5 × 10−5 and 2 × 10−4 for MG ${|l|=0}$ and ${|l|=3}$ can be realized, respectively. The BERs of MG ${|l|=0}$ and MG ${|l|=3}$ are still different because of the differences in power responses of two PDs and inter-MG cross talk. In addition, it is clearly observed that partial-MG reception (${l=-3}$ or ${l=+3}$) has worse BER performance compared with full-MG reception (${|l|=3}$) with/without inter-MG cross talk. Theoretically, due to random mode coupling/rotation within MG during fiber propagation, the ROP and BER of partial-MG reception vary with time. Accordingly, the performance of ${l=-3}$ or ${l=+3}$ is full of uncertainty and randomness. Meanwhile, full-MG reception is unaffected and its performance is always superior to partial-MG reception since mode power loss within MG is avoided. In addition, the variation of the calculated signal-to-noise ratio (SNR) versus ROP is depicted in Fig. 4(b). It can be seen that even during a short time of static testing, the inter-MG cross talk not only degrades the SNR/BER performance but also enlarges the system variation in the SNR/BER performance, especially for MG ${|l|=0}$ with better SNR performance in this system.
Fig. 3. Measured BER versus ROP using FFE with different numbers of taps for (a) ${|l|=0}$ and (b) ${|l|=3}$ over a 5-km FMF.
To ensure the reliability of the system, dynamic evaluation of BER and SNR under four conditions over five time slots is implemented at the ROP of ${-4}$ dBm, as shown in Fig. 5.
- • First time slot from 23:00 to 23:30 Hong Kong time (HKT). Only scenario of 0 (S.) is carried out. BERs/SNRs are quite low/large and stable during this time slot without inter-MG cross talk.
- • Second time slot from 23:30 to 00:00 HKT, MG ${|l|=3}$ is then co-transmitted with MG ${|l|=0}$ and these two MGs are simultaneously received, namely scenario of 0 (M.) and 3 (M.). It is clearly observed that the BERs for both MGs are always lower than 1 × 10−3, although BER/SNR performance of ${|l|=0}$ has degraded to a certain extent compared with its value in the first time slot. In addition, the fluctuation of BER/SNR for ${|l|=0}$ is increased from the first time slot to this time slot with inter-MG cross talk. The fluctuation of BER and SNR performance for MG ${|l|=0}$ is also larger than MG ${|l|=3}$, probably because the BER/SNR performance for MG ${|l|=0}$ is better and the effect of cross talk is more obvious.
- • Third time slot from 00:00 to 00:30 HKT. Same as in the second time slot, two MGs are simultaneously transmitted. However, at the receiver side, 0 (M.) and 3 (M.) (full-MG reception) or 0 (M.) and ${+3}$ (M.) (partial-MG reception) or 0 (M.) and ${-3}$ (M.) (partial-MG reception) are received sequentially and repeatedly over time. As can be seen, the BER/SNR performance of partial-MG reception is worse than full-MG reception since the partial one loses certain power over undetected modes within the MG.
- • Fourth time slot from 00:30 to 01:00 HKT. MG ${|l|=0}$ is switched off and then only MG ${|l|=3}$ is transmitted. At the receiver, 3 (S.) or ${+3}$ (S.) or ${-3}$ (S.) is sequentially and repeatedly received over time. Obviously, without cross talk from MG ${|l|=0}$, BERs/SNRs of MG ${|l|=3}$, mode ${l=-3}$, and mode ${l=+3}$ are improved on average and have a smaller fluctuation over time.
- • Back to scenario of 0 (M.) and 3 (M.) from 01:00 to 02:30 HKT. BERs/SNRs of both MGs still reach the previous plateau and are close to them in the second time slot. Specifically, all BERs still stay below 1×10-3, which further proves the reliability of the proposed scheme.
Fig. 5. Dynamic evaluation of BER and SNR performance versus HKT under four scenarios at the ROP of −4 dBm.
In conclusion, a simple, flexible, and cost-effective MG filter approach for MGDM IM/DD transmission is proposed, which avoids received power fluctuation by random mode rotation through an FMF with time. By using this MG filter scheme, a MIMO-free MGDM multiplexing two OAM MGs with PAM-4 signal and a total capacity of 152 Gb/s is experimentally demonstrated over a 5-km FMF. The performance of BER/SNR is evaluated statically versus ROP and dynamically over time. BERs of two MGs can meet a 7% HD-FEC BER threshold of ${3.8\times 10^{-3}}$ under both static and dynamic cases. These experimental results adequately verify the robustness and feasibility of the proposed MGDM scheme based on MG filter architecture and indicate that such a scheme has enormous potential in future high-speed short-reach optical interconnects.
Funding
National Key Research and Development Program of China (2018YFB1801701); National Natural Science Foundation of China (62035018, 62101602, U1701661); Project of the Shenzhen Municipal Science and Technology Innovation Commission (SGDX20201103095203030); The Hong Kong Government General 291 Research Fund (PolyU 15217620, PolyU 15220120); Project of the Hong Kong Polytechnic University (G-SB1P); PolyU postdoc matching fund scheme of the Hong Kong Polytechnic University (1-W150).
Disclosures
The authors declare no conflicts of interest.
Data availability
Data underlying the results presented in this paper are not publicly available at this time but may be obtained from the authors upon reasonable request.
REFERENCES
1. G. Li, N. Bai, N. Zhao, and C. Xia, Adv. Opt. Photonics 6, 413 (2014). [CrossRef]
2. P. J. Winzer, Nat. Photonics 8, 345 (2014). [CrossRef]
3. L. Shen, D. Ge, S. Shen, S. Wang, C. Zhao, G. Wang, L. Zhang, J. Luo, X. Lan, L. Deng, M. Zuo, Y. Gao, and J. Li, in 2021 Optical Fiber Communication Conference (OFC) (2012), p. W1I.2.
4. F. Yaman, E. Mateo, and T. Wang, 2012 Optical Fiber Communication Conference (OFC) (2012), p. OTu1D.2.
5. D. Soma, Y. Wakayama, S. Beppu, S. Sumita, T. Tsuritani, T. Hayashi, T. Nagashima, M. Suzuki, M. Yoshida, K. Kasai, M. Nakazawa, H. Takahashi, K. Igarashi, I. Morita, and M. Suzuki, J. Lightwave Technol. 36, 1 (2018). [CrossRef]
6. R. Ryf, N. K. Fontaine, S. Wittek, et al., 2018 European Conference on Optical Communication (ECOC) (2018), pp. 1–3.
7. J. Li, J. Zhang, F. Li, X. Huang, S. Gao, and Z. Li, Opt. Express 26, 18749 (2018). [CrossRef]
8. K. Benyahya, C. Simonneau, A. Ghazisaeidi, N. Barré, P. Jian, J.-F. Morizur, G. Labroille, M. Bigot, P. Sillard, J. G. Provost, H. Debrégeas, J. Renaudier, and G. Charlet, J. Lightwave Technol. 36, 355 (2018). [CrossRef]
9. K. Ingerslev, P. Gregg, M. Galili, F. Da Ros, H. Hu, F. Bao, M. A. Usuga Castaneda, P. Kristensen, A. Rubano, L. Marrucci, S. Ramachandran, K. Rottwitt, T. Morioka, and L. K. Oxenløwe, in 2017 Optical Fiber Communications Conference (OFC) (2017), pp. 1–3.
10. L. Shen, D. Ge, Y. Liu, L. Xiong, S. Chen, H. Zhou, R. Zhang, L. Zhang, J. Luo, and J. Li, in 2018 European Conference on Optical Communication (ECOC) (2018), pp. 1–3.
11. C. Brunet, B. Ung, L. Wang, Y. Messaddeq, S. LaRochelle, and L. A. Rusch, Opt. Express 23, 10553 (2015). [CrossRef]
12. H. Liu, H. Wen, J. C. A. Zacarias, J. E. Antonio-Lopez, N. Wang, P. Sillard, R. Amezcua-Correa, and G. Li, in 2018 Optical Fiber Communication Conference (OFC) (2018), p. W4J.2.
13. K. Benyahya, C. Simonneau, A. Ghazisaeidi, R. R. Muller, M. Bigot, P. Sillard, P. Jian, G. Labroille, J. Renaudier, and G. Charlet, 2018 European Conference on Optical Communication (ECOC) (2018), pp. 1–3.
14. J. Zhang, G. Zhu, J. Liu, X. Wu, J. Zhu, C. Du, W. Luo, Y. Chen, and S. Yu, Opt. Express 26, 4243 (2018). [CrossRef]
15. G. Zhu, X. Wu, J. Liu, Y. Chen, J. Zhang, J. Zhu, C. Du, and S. Yu, 2017 Asia Communications and Photonics Conference (ACP) (2017), pp. 1–3.
16. X. Wu, J. Zhang, Q. Fan, Z. Tan, J. Li, X. Yi, Z. Li, A. P. T. Lau, and C. Lu, J. Lightwave Technol. 40, 1962 (2022). [CrossRef]
