Open Access
Add to BibSonomy
Abstract
Self-homodyne detection (SHD) is a promising approach to realize high-capacity short-reach optical transmission systems with low cost and low power consumption. We experimentally demonstrate single-carrier net 800-Gb/s SHD transmission with low-cost ∼MHz linewidth distributed feedback (DFB) laser over 2 km, 10 km, 25 km, and 40 km single-mode fiber (SMF) using three different quadrature amplitude modulation (QAM) formats, including 80-Gbaud dual-polarization (DP) 64QAM, 100-Gbaud DP-32QAM, and 120-Gbaud DP-16QAM. Among them, net 800-Gb/s DP-64QAM SHD transmission over 25 km SMF using an uncooled DFB laser with a linewidth of 2.6 MHz is experimentally verified. The detailed experimental performance evaluation of net 800Gb/s SHD system is performed, in which various configurations are considered, such as different laser linewidths, three QAM formats, and different transmission distances. DFB lasers with linewidths of 1 MHz and 2.6 MHz lead to negligible penalty when compared to the same SHD system but using an external cavity laser (ECL) with a linewidth of 26kHz in back-to-back (BTB) case. 80-Gbaud DP-64QAM obtains the highest optical signal-to-noise ratio (OSNR) requirement and the highest bit-error rate (BER) floor but the best tolerance of chromatic dispersion (CD). 120-Gbaud DP-16QAM achieves the lowest OSNR requirement and the lowest BER floor but the worst tolerance of CD. The detailed experimental investigation is conducive to promote the practical application of SHD in different short-reach scenarios.
© 2023 Optica Publishing Group under the terms of the Optica Open Access Publishing Agreement
1. Introduction
Driven by the rapid proliferation of emerging communication services such as online interactive maps, online education, and Internet of Things, the global data traffic has been growing explosively. The short-reach optical transmissions such as the data center applications and access networks dominate a major portion of current global data traffic. Intensity modulation direct-detection (IM/DD) systems with the characteristic of low cost and low power consumption appear to be the preferred scheme in the short-reach networks [1,2]. However, further adoption of multi-lane IM/DD in an attempt to meet the data rate of Ethernet interface approaching 800 G and 1.6 T may be confronted with such problems as limited electrical bandwidth, power fading caused by dispersion, and limited spectrum occupancy [3,4]. Consequently, the application of coherent detection technologies with higher spectral efficiency and good tolerance of propagation impairments has been gradually extended from optical back-bone transmission networks to short-reach scenarios. Nevertheless, traditional intradyne-based coherent detection is still viewed as of high cost and high-power consumption for short-reach applications due to its requirements for narrow linewidth lasers with complex laser control circuitries and power-hungry digital signal processing (DSP) algorithms [5–7].
Self-homodyne detection (SHD), on the other hand, has been known to bring much relaxed requirement in terms of component specifications and DSP [8–11]. In SHD systems, the local oscillator (LO) and modulated signal originate from the same laser and are transmitted over parallel channels such as single-mode fiber (SMF) pair [12–14], multi-core fiber (MCF) [15–17] and few-mode fiber (FMF) [18–20]. The preservation of coherence between signal and LO in SHD systems is conducive to eliminate the impairments caused by laser frequency offset and phase noise. Therefore, a simple and low-cost distributed feedback (DFB) laser is sufficient for SHD systems, instead of expensive lasers with narrow linewidth and wavelength locking capabilities that are indispensable in traditional intradyne detection systems. In addition, the requirement of the complex frequency offset estimation and carrier phase recovery algorithms can be much relaxed or even unnecessary. Thanks to these advantages, SHD provides high capacity and spectral efficiency while reduces the cost and power consumption, which is a good compromise for short-reach optical transmission applications. Several recent works on SHD systems have been demonstrated. It has successfully realized the SHD transmission of 400-Gb/s dual-polarization (DP) 16-quadrature amplitude modulation (QAM) over 2.3-km SMF pair [21]. Exploiting a DFB laser with a linewidth of 1.5 MHz, 600-Gb/s DP-64QAM SHD transmission over 5-km SMF has been demonstrated [22]. More recently, 800-Gb/s DP-64QAM SHD transmission over 2-km SMF using a DFB laser with a linewidth of 1 MHz [23] and 800-Gb/s DP-16QAM SHD transmission over 1-km SMF using a DFB laser with a linewidth of 125kHz [24] have been reported. However, single-carrier 800-Gb/s SHD transmission with an uncooled DFB laser that would be significantly cheaper than external cavity laser (ECL) is rarely reported. On the other hand, for short-reach optical interconnect applications with longer transmission distances such as LR (10 km), and ER (10 km ∼ 40 km), the chromatic dispersion (CD) is non-negligible for longer transmission distance, especially in the high rate communication systems. It is not clear about the CD tolerance of 800-Gb/s SHD systems with different configurations. Furthermore, the applicable modulation formats may be inconsistent for different 800 G short-reach optical communication applications. In these scenarios, a laudable goal would be to investigate in detail the transmission performance of single-carrier 800-Gb/s SHD system with various configurations such as several types of laser with different linewidth, different QAM formats, and transmission distances ranging from a few kilometers to dozens of kilometers.
In this paper, we report the experimental demonstration of net 800-Gb/s SHD transmission over 2 km, 10 km, 25 km, and 40 km SMF with different QAM formats, including 80-Gbaud DP-64QAM, 100-Gbaud DP-32QAM, and 120-Gbaud DP-16QAM, using a low-cost ∼MHz linewidth DFB laser. Moreover, we successfully demonstrate 80-Gbaud DP-64QAM using an uncooled DFB laser with a linewidth of 2.6 MHz. In order to evaluate the effect of laser linewidth on SHD transmission performance, the ECL with a linewidth of 26kHz is employed for comparison. For these net 800Gb/s SHD systems with various configurations, the transmission performance is characterized in detail. The impacts of laser linewidth and CD are also evaluated. Particularly, the frequency offset compensation and carrier phase recovery algorithms are not employed. There are no distinct penalties among the three types of lasers with different linewidths in back-to-back (BTB) case, which verifies the phase noise elimination property of SHD. In terms of optical signal-to-noise ratio (OSNR) requirement, 80-Gbaud DP-64QAM is the highest and 120-Gbaud 16QAM is the lowest. While in terms of CD tolerance, on the contrary, 80-Gbaud DP-64QAM reaches the highest tolerance, and 120-Gbaud DP-16QAM reaches the lowest tolerance. The favorable experimental results suggest that SHD system with a large linewidth DFB laser is a potentially attractive scheme for high capacity short-reach optical interconnect applications.
2. Experimental setup
Figure 1 illustrates the experimental setup of 800-Gb/s SHD transmission system based on SMF pair with various configurations. The light at a wavelength of 1550 nm from a laser is pre-amplified by an erbium-doped fiber amplifier (EDFA) due to low output optical power of the DFB laser used in our experiment, and then a tunable filter (TF) is used to reduce the amount of amplified spontaneous emission (ASE) noise from the EDFA. The optical power after amplification and filtering is about 18dBm. If a DFB laser with high output power is obtained, the EDFA and TF can be removed. A 90:10 coupler is employed to divide the light into two separate channels, one of which acts as remote LO and the other is modulated to carry signals. The simple digital pre-distortion is carried out by using the arcsine function to compensate the inherent nonlinear sinusoidal characteristics of the optical in-phase and quadrature modulator (IQM) [25]. An arbitrary waveform generator (AWG, Keysight M8199A) is used to generate 80-Gbaud DP-64QAM RF signals (AWG operating at 1.6 Sa/Symbol), 100-Gbaud DP-32QAM RF signals (AWG operating at 1.25 Sa/Symbol), and 120-Gbaud DP-16QAM RF signals (AWG operating at 1 Sa/Symbol). Four SHF S804B RF amplifiers with 60-GHz 3 dB bandwidth drive the RF signals to a dual-polarization IQM with 35-GHz 3 dB bandwidth. The gross data rate of both 80-Gbaud DP-64QAM and 120-Gbaud DP-16QAM are 960-Gbit/s, and gross data rate of 100-Gbaud DP-32QAM is up to 1-Tbit/s. Assuming that the pre-forward-error correction (pre-FEC) threshold has 15% overhead, the net data rate is about 800-Gbit/s. At the receiver, the remote LO that travels the same distance as modulated signal is sent to a coherent receiver which consists of a state-of-the-art 2 × 8 optical 90°-hybrid and four 70-GHz balanced photodiodes (BPDs). Note that an optical delay line placed in the LO link is used to eliminate the relative length difference between the signal and remote LO, and we finely calibrate the lengths of signal and LO links to be nearly identical. A real-time oscilloscope (Keysight UXR0594A) operating at 256 GSa/s digitizes the electrical signal. The remote LO and modulated signal co-propagate in the SMF (standard G.652 fiber) pair, and thus they have same central frequency and reference phase, which minimizes the impact of laser phase noise and frequency offset. The portion of DSP algorithms related to frequency offset compensation and carrier phase recovery are no longer needed. In receiver DSP, only equalizer block using conventional 4 × 4 real-value MIMO is adopted after resampling. Note that the offline DSP bypasses the IQ non-orthogonality compensation algorithm in our experiment.
Fig. 1. The experimental setup of 800-Gb/s SHD transmission system based on SMF pair. DFB: distributed feedback; EDFA: erbium-doped fiber amplifier; TF: tunable filter; AWG: arbitrary waveform generator; IQM: in-phase/quadrature modulator; VOA: variable optical attenuator; LO: local oscillator; OSC: oscilloscope; Coh. Rx.: coherent receiver.
A DFB laser and an uncooled DFB laser with linewidths in the order of MHz are employed in the SHD system, while an ECL with a narrow linewidth is also used for comparison. Moreover, we measure the linewidths of the three lasers by delayed self-homodyne method with a coherent receiver [26,27]. The light from the laser being tested is evenly split into two paths, both of which are then fed into a coherent receiver. Note that one of the paths transmits over 20-km SMF as delayed path. By using a coherent receiver, the complex amplitude of a beat between the two beams originating from a laser is acquired, from which we calculate differential phase noise of the laser though offline DSP. The FM-noise spectrum which is defined as the power-spectral-density (PSD) function of the instantaneous-frequency fluctuation can be estimated from differential phase noise. The linewidth $\Delta f$ of laser is related to the white noise region ${S_w}$ of the FM-noise spectrum, which is estimated as $\Delta f = \pi {S_w}$. Figure 2(a), (b) and (c) display the measured FM-noise spectrum of the ECL, DFB laser, and uncooled DFB laser, respectively. The corresponding white noise region are marked by the red dashed line. Consequently, the measured linewidths of the ECL, DFB laser, and uncooled DFB laser are confirmed to be 26kHz, 1 MHz, and 2.6 MHz, respectively.
3. Experimental results and discussions
The performance of 800-Gb/s SHD transmission with various configurations is experimentally characterized. First, in the case of 80-Gbaud DP-64QAM, we investigate the pre-FEC bit-error rate (BER) performance as a function of the received OSNR of the transmitted signal using three types of lasers with different linewidth in back-to-back (BTB) and 25 km SMF transmission scenarios. Figure 3(a), (b) and (c) show the BER performance of 80-Gbaud DP-64QAM SHD transmission using an ECL with a linewidth of 26kHz, a DFB laser with a linewidth of 1 MHz, and an uncooled DFB laser with a linewidth of 2.6 MHz, respectively, where the MIMO equalizer uses 61 taps and CD compensation algorithm is not employed. In BTB scenario, the corresponding OSNR at a BER threshold of 1.25e-2 (take the concatenated forward error correction (CFEC) threshold of 400 Ze Best Range (ZR) as a reference) for three types of lasers are all about 32 dB. There is no distinct OSNR penalty for a SHD system using a larger-linewidth DFB laser, an uncooled DFB laser, or a narrow-linewidth ECL. It is worth noting that the lasers with such wide linewidths cannot be operated in traditional coherent systems. At a BER of 1.25e-2, the corresponding OSNR of 25 km SMF transmission with ECL, DFB laser, and uncooled DFB laser are about 33.8 dB, 36 dB, and 37 dB, respectively. The OSNR penalty at a BER threshold of 1.25e-2 between 25 km SMF transmission and BTB for using ECL, DFB laser, and uncooled DFB laser are 1.8 dB, 4 dB, and 5 dB, respectively. Compared with the SHD system with a narrow-linewidth ECL, the use of a 1 MHz linewidth DFB laser and a 2.6 MHz linewidth uncooled DFB laser lead to OSNR penalties of 2.2 dB and 3.2 dB at BER of 1.25e-2, respectively. In addition, the pre-FEC BER floor raises as the laser linewidth increases in the case of 25 km SMF transmission.
Next, for short-reach SHD transmission applications with different distances such as 2 km (FR), 10 km (LR), and 40 km (ER), the pre-FEC BER performance versus OSNR is characterized, as shown in Fig. 3(d). 61 taps are also used in the MIMO equalizer. Here, the CD compensation algorithm in frequency domain is adopted for 40 km SHD transmission. Compared with BTB scenario, the OSNR penalties at BER of 1.25e-2 for 2 km, 10 km, and 40 km are about 0.9 dB, 1.9 dB, and 5.5 dB, respectively. Longer fiber transmission distance brings greater signal impairments. Such penalty in BER performance mainly comes from the additional distortions introduced by optical fiber, such as CD. The CD cannot be neglected in high-capacity single-carrier 800-Gb/s fiber transmission system. The pulse broadening delay Δτ caused by CD is proportional to the spectral width of the light pulse Δλ and the transmission distance L, expressed as $\mathrm{\Delta}\tau=D\cdot\mathrm{\Delta}\lambda\cdot L$, where D is dispersion coefficient of fiber.
Fig. 3. The performance of 800-Gb/s SHD transmission with 80-Gbaud DP-64QAM. The measured BER performance versus the received OSNR for 25 km SMF transmission using (a) an ECL with a linewidth of 26kHz; (b) a DFB laser with a linewidth of 1 MHz; (c) an uncooled DFB laser with a linewidth of 2.6 MHz. (d) The measured BER performance versus OSNR for 2 km (FR), 10 km (LR), and 40 km (ER) SMF transmission. LW: linewidth.
A simulation is carried out to quantitatively analyze the OSNR penalty caused by the combination impacts of CD and laser linewidth in SHD system. Figure 4 shows the simulated results of OSNR penalty of 80-Gbaud DP-64QAM at a BER of 1.25e-2 versus CD with various laser linewidths. The penalty is calculated with reference to the BTB scenario. Here, the simulation model and employed DSP are configured the same as the experiment. The dispersion coefficient of SMF at 1550 nm is set to be -17 ps/nm·km. The corresponding CD values for 2 km, 10 km, 25 km, and 40 km are calculated to be 34ps/nm, 170ps/nm, 425ps/nm, and 680ps/nm, respectively. Note that the CD compensation algorithm in frequency domain is also adopted for 40 km in simulation. It can be seen from the simulated results that the expected OSNR penalties at 25 km for laser linewidths of 26KHz, 1 MHz, and 2.6 MHz are 1.6 dB, 3.7 dB, and 4.7 dB, respectively. The experimental and simulated OSNR performance of 80-Gbaud DP-64QAM with different lasers are summarized in Table 1. The expected OSNR penalty using 1 MHz linewidth laser for 2 km, 10 km, and 40 km are 0.6 dB, 1.6 dB, and 3.5 dB, respectively. As a result, according to the simulated results, the OSNR penalties measured in the above experiment are in line with expectations. At a certain fiber transmission distance, the wider the linewidth of the laser, the greater the OSNR penalty. In the SHD system, the requirement of laser linewidth can be much relaxed due to its phase noise canceling properties. It is necessary to figure out a trade-off between laser linewidth and OSNR penalty in single-carrier high-capacity short-reach SHD transmission system.
Fig. 4. The simulated results of OSNR penalty of 80-Gbaud DP-64QAM at a BER of 1.25e-2 versus CD with various laser linewidths. LW: linewidth; CDC: CD compensation.
Table 1. Performance comparison of 80-Gbaud DP-64QAM using different lasers
The total channel impairments of 40-km SMF is greater than that of 25 km SMF. The MIMO equalizer with 61 taps is insufficient to compensate for signal distortions caused by power attenuation, noise, and CD of 40-km SMF. Under the circumstances, we investigate the effect of the number of MIMO equalizer taps on pre-FEC BER performance with and without CD compensation. Figure 5(a) displays the result for 40-km SMF with 1 MHz linewidth DFB laser. In the absence of CD compensation, the BER decreases with the increase of the number of taps. Whereas the BER is only slightly decreased with CD compensation. The cumulative CD of 40 km SMF is 680ps/nm, and the spectral width of 80-Gbaud DP-64QAM signal is about 0.64 nm, and thus an 80-Gbaud symbol with a pulse width of 12.5ps would be expanded to about 435ps (35 symbols). Here, the signal sequence is resampled to 2 samples per symbol before MIMO equalization. It is theoretically estimated that about 70 MIMO equalizer taps are required to compensate for CD in the case of 80-Gbaud DP-64QAM transmitting 40 km SMF. In fact, in addition to CD, signal will also suffer from other impairments such as I-Q impairments, ASE noise of EDFA, etc. Therefore, more equalizer taps are required in actual experiment to compensate for entire linear impairments including CD and the others. In our experiment, when the number of taps is larger than 81, the BER performance with or without CD compensation is almost the same, as shown in Fig. 5(a). In other words, 81 taps are required to compensate for all impairments in the experiment, of which approximately 70 taps are used to compensate for CD. Figure 5(b) shows the BER performance versus the number of taps for 25 km SMF with 2.6 MHz linewidth uncooled DFB laser. The cumulative CD of 25 km SMF is 425ps/nm, which will expand an 80-Gbaud pulse to about 272ps (22 symbols). Similarly, in the case of 80-Gbaud DP-64QAM transmitting 25 km SMF, the theoretical number of taps used to compensate for CD is roughly estimated to be 44. Without frequency-domain CD compensation algorithm, the minimum number of taps is 57 in the experiment to compensate for linear impairments, of which about 44 taps are needed for CD mitigation. Therefore, when the MIMO equalizer uses 61 taps, the CD compensation algorithm is needed for 40 km SMF with 1 MHz linewidth DFB laser, but not for 25 km SMF with 2.6 MHz linewidth uncooled DFB laser.
Fig. 5. The measured BER performance of 80-Gbaud DP-64QAM versus the number of MIMO equalizer taps with and without CD compensation for (a) 40 km SMF with 1 MHz linewidth DFB laser; (b) 25 km SMF with 2.6 MHz linewidth uncooled DFB laser. LW: linewidth; CDC: CD compensation.
Furthermore, the performance of SHD transmission with 100-Gbaud DP-32QAM and 120-Gbaud DP-16QAM using a DFB laser with a linewidth of 1 MHz is characterized. In the case of 100-Gbaud 32QAM, SHD transmission over 2 km, 10 km, 25 km, and 40 km is experimentally achieved, as shown in Fig. 6(a) for the pre-FEC BER performance versus OSNR. Note that 61 taps are still used in the MIMO equalizer. The CD compensation algorithm is adopted in 25 km and 40 km SMF cases, but not in 2 km and 10 km SMF cases. At a BER of 1.25e-2, the corresponding OSNR for BTB, 2 km, 10 km, 25 km, and 40 km cases are 29.5 dB, 30 dB, 30.6 dB, 32 dB, and 33 dB, respectively, which are all lower than that of 80-Gbaud 64QAM in the identical configuration. The OSNR penalties between these four transmission distances and BTB case are about 0.5 dB, 1.1 dB, 2.5 dB, and 3.5 dB, respectively. In addition, the pre-FEC BER performance versus the number of MIMO equalizer taps for 100-Gbaud DP-32QAM SHD transmission over 40 km SMF with 1 MHz linewidth DFB laser is characterized, as shown in Fig. 6(b). The 100-Gbaud DP-32QAM corresponds to a symbol period of 10ps, which is lower than that of 80-Gbaud DP-64QAM (12.5ps). The spectral width of 100-Gbaud DP-32QAM signal is about 0.8 nm, thus a 100-Gbaud symbol would be expanded to about 544ps (∼55 symbols). It is theoretically estimated that 110 taps are required to compensate for CD in the case of 100-Gbaud DP-32QAM transmitting 40 km SMF. In the experiment, 37 MIMO equalizer taps are sufficient when CD compensation is implemented, while without CD compensation algorithm, the number of taps of the equalizer needs to reach 125 to obtain BER performance approaching to the case with CD compensation. Within the 125 taps, about 110 taps are used to compensation for CD impairments, which is much larger than that of 80-Gbaud DP-64QAM at same configuration. As a result, the impacts of CD on 100-Gbaud DP-32QAM signal is greater than that of 80-Gbaud DP-64QAM with same configuration.
Fig. 6. The performance of 800-Gb/s SHD transmission with 100-Gbaud DP-32QAM. (a) The measured BER performance versus OSNR for 2 km, 10 km, 25 km, and 40 km SMF transmission. (b) The measured BER performance versus the number of MIMO equalizer taps with and without CD compensation for 40 km SMF with 1 MHz linewidth DFB laser. LW: linewidth; CDC: CD compensation.
In the case of 120-Gbaud DP-16QAM, the pre-FEC BER performance for SHD transmission over 2 km, 10 km, 25 km, and 40 km is shown in Fig. 7(a). Uniformly, MIMO equalizer used 61 taps, and the CD compensation algorithm is adopted in 25 km and 40 km SMF cases. At a BER of 1.25e-2, the corresponding OSNR for BTB, 2 km, 10 km, 25 km, and 40 km cases are 26.3 dB, 26.6 dB, 27 dB, 27.8 dB, and 28.4 dB, respectively, which are lowest among the three QAM formats in the identical configuration. Compared with the BTB case, the OSNR penalties for these four transmission distances are 0.3 dB, 0.7 dB, 1.5 dB, and 2.1 dB, respectively. The required OSNR and OSNR penalties of the three types of QAM formats at a BER of 1.25e-2 are summarized in Table 2. Among the three QAM signals, 120-Gbaud DP-16QAM possesses the lowest OSNR requirement, while 80-Gbaud DP-64QAM has the highest OSNR requirement. In addition, assuming a low latency FEC BER threshold of 4e-3, the BER values of 120-Gbaud DP-16QAM can reach 4e-3 threshold. The corresponding OSNR for BTB, 2 km, 10 km, 25 km, and 40 km cases are 29 dB, 29.4 dB, 30 dB, 31 dB, and 32 dB, respectively. Thereby the OSNR penalties between these four transmission distances and BTB case are about 0.4 dB, 1 dB, 2 dB, and 3 dB at a low latency FEC BER threshold of 4e-3, respectively. On the other hand, 120-Gbaud DP-16QAM signal corresponds to a lowest symbol period of 8.3ps among the three types of QAM signals. The spectral width of 120-Gbaud DP-16QAM signal is about 0.96 nm, and thus a 120-Gbaud symbol would be expanded to about 653ps (∼79 symbols). The theoretical number of MIMO equalizer taps used to compensate for CD of 40 km SMF is roughly estimated to be 158. Figure 7(b) depicts the measured pre-FEC BER performance versus the number of taps for 120-Gbaud DP-16QAM SHD transmission over 40 km SMF with 1 MHz linewidth DFB laser. Only 41 taps are needed in case with CD compensation. When the BER performance with or without CD compensation is almost the same, the number of taps without CD compensation is larger than 177, which is largest among the three QAM signals. 177 taps are required to compensate for linear impairments in the experiment, of which approximately 158 taps are used for CD mitigation. The experimental results indicate that the CD tolerance of 120-Gbaud DP-16QAM signal is the largest. The CD tolerance of the three types of QAM formats is summarized in Table 3.
Fig. 7. The performance of 800-Gb/s SHD transmission with 120-Gbaud DP-16QAM. (a) The measured BER performance versus OSNR for 2 km, 10 km, 25 km, and 40 km SMF transmission. (b) The measured BER performance versus the number of MIMO equalizer taps with and without CD compensation for 40 km SMF with 1 MHz linewidth DFB laser. LW: linewidth; CDC: CD compensation.
Table 2. OSNR comparison of the three QAM signals using 1 MHz DFB laser at a BER of 1.25e-2
Table 3. CD tolerance of the three QAM signals in 800-Gb/s SHD system with 1 MHz DFB laser
Moreover, considering the short transmission distance such as 2 km, the fiber-induced CD is slight, even negligible. In this case where the CD is tiny, we investigate the BER performance versus OSNR for these three QAM signals with different MIMO equalizer taps, as shown in Fig. 8. The results of only one of the two polarizations are shown due to the similar performance between them. The complexity of the MIMO equalizer is positively correlated with the number of taps, the more the number of taps, the higher the complexity of the equalizer. It can be found that when the number of taps of 80-Gbaud DP-64QAM, 100-Gbaud DP-32QAM and 120-Gbaud DP-16QAM is larger than 29, 29, and 37, respectively, the increased number of taps only slightly improves the BER performance. Therefore, the optimal number of MIMO equalizer taps of the three QAM signals are 29, 29, and 37 for SHD transmission over 2 km SMF, respectively. The optimal number of taps for 16QAM is slightly larger than that of the other two QAM signals. This can be interpreted that the combined effects of transceiver I-Q impairments, limited electrical bandwidth and channel impairments for 120-Gbaud DP-16QAM are slightly higher than that of the other two QAM signals. Firstly, for high bit rate coherent systems, transceiver IQ-impairments due to component imperfection or misalignment, such as I-Q imbalance and I-Q skew, may lead to significant performance degradation. Although prior manual calibration can remove some of the impairments, these impairments cannot be completely eliminated and can still appear due to limited accuracy and change of environment. The real-valued 4 × 4 MIMO equalizer possesses the ability of I-Q impairments mitigation [28]. Laser phase noise will limit the I-Q impairments compensation capability of the equalizer, which can be solved in the SHD system due to its characteristics of phase noise elimination. Higher baud rate signals with lower symbol period are more sensitive to I-Q skew. Secondly, higher electrical bandwidth is required for higher baud rate signals. The inter-symbol interference (ISI) caused by electrical bandwidth limitation has a greater impact on higher baud rate signals. The IQM with 35 GHz bandwidth is the lowest bandwidth device in our experimental system. The post equalizer can compensate for ISI caused by bandwidth limitations to a certain extent, and more taps are needed for higher baud rate signals [29]. Thirdly, channel impairments such as ASE noise of EDFA and tiny CD of 2 km SMF are also inevitable. For 2 km SMF, the number of taps used to compensate CD for the three QAM signals are roughly estimated to be 4, 6, and 8, respectively. However, due to the inevitable other impairments mentioned above, more equalizer taps are required to obtain favorable performance. In addition, 120-Gbaud DP-16QAM signals are slightly more sensitivity to some impairments, so the optimal number of taps for 16QAM is slightly more. Nevertheless, using 29 equalizer taps for 120-Gbaud DP-16QAM, the BER values can still reach 4e-3 threshold, and the OSNR at a BER of 1.25e-2 is 28 dB, which is the lowest.
Fig. 8. The measured BER performance versus OSNR with different MIMO equalizer taps for (a) 80-Gbaud DP-64QAM; (b) 100-Gbaud DP-32QAM; (c) 120-Gbaud DP-16QAM.
We successfully realized single-carrier 800-Gb/s SHD transmission using three types of QAM formats, i.e., 80-Gbaud DP-64QAM, 100-Gbaud DP-32QAM and 120-Gbaud DP-16QAM. Higher-order QAM formats carry more bits, meaning the lower baud rates for higher-order QAM formats with the same bit rate of 800 G. Among the three signals, 80-Gbaud DP-64QAM signal has the highest requirement of OSNR and highest pre-FEC BER floor. The performance of 100-Gbaud DP-32QAM is in the middle level. For 120-Gbaud DP-16QAM, both the requirement of OSNR and the pre-FEC BER floor are lowest, achieving the BER values of 4e-3 threshold. In addition, in case of short-reach SHD transmission over 2 km SMF where the CD is tiny, the minimum number of taps for 64QAM and 32QAM is the same, while it for 16QAM is slightly more. This suggests that the complexity of the MIMO equalizer for 120-Gbaud DP-16QAM is slightly higher than that of the other two QAM signals. On the other hand, when the bit rate reaches 800-Gb/s, the CD should be considered for tens of kilometers of SMF transmission such as 25 km and 40 km. A lower baud rate corresponds to a larger symbol period, and thus the impact of CD-introduced pulse broadening delay is smaller for lower baud rate signals. For 800-Gb/s SHD transmission over 25 km and 40 km, 120-Gbaud DP-16QAM signals require CD compensation algorithm or more equalizer taps. The employment of either the CD compensation algorithm or more equalizer taps would increase the complexity of DSP. It is an important topic for practical application to reduce the complexity of CD compensation through advanced methods.
4. Conclusion
In summary, we have experimentally demonstrated single-carrier net 800-Gb/s SHD transmission of 80-Gbaud DP-64QAM, 100-Gbaud DP-32QAM, and 120-Gbaud DP-16QAM with a 2.6 MHz linewidth uncooled DFB laser and a 1 MHz DFB laser. Short-reach SHD transmission applications with different distances including 2 km, 10 km, 25 km, 40 km have been realized. A low-cost uncooled DFB laser with a linewidth of up to 2.6 MHz has been successfully used for 80-Gbaud DP-64QAM SHD transmission over 25 km SMF. The detailed experimental evaluation of net 800Gb/s SHD transmission performance of each configuration is carried out. Compared to the SHD system using an ECL with a linewidth of 26KHz, a 1 MHz DFB laser and a 2.6 MHz uncooled DFB laser do not introduce significant penalty in BTB case, indicating that the SHD system allows the use of uncooled DFB lasers with low-cost and wide linewidths. Among the three different QAM formats, 80-Gbaud DP-64QAM obtains the highest OSNR requirement and the highest BER floor, but the best tolerance of CD. The performance of 100-Gbaud DP-32QAM is in the middle level. The 120-Gbaud DP-16QAM, on the contrary, achieves the lowest OSNR requirement and the lowest BER floor but the worst tolerance of CD. The BER value of 120-Gbaud DP-16QAM can reach 4 × 10−3 (assuming a low-latency FEC threshold). The detailed experimental investigation lays the foundation for practical application of SHD in different short-reach scenarios. For optical network applications with requirements of high OSNR, longer distance, or low bandwidth, 64QAM SHD scheme is more suitable at the same bit rate, while 16QAM SHD scheme is more suitable for network applications with requirements of low OSNR, shorter distance, or large bandwidth. For some moderate applications, 32QAM SHD scheme may be the best choice. Look ahead, higher order QAM formats may be selected for higher-capacity SHD system, the major challenges are high-performance equalizers, highly linear devices, and high OSNR.
Funding
National Natural Science Foundation of China (62125503, 62261160388,62101198); Natural Science Foundation of Hubei Province (2021CFB011); the Key R&D Program of Hubei Province of China (2020BAB001, 2021BAA024); the Shenzhen Science and Technology Innovation Program (JCYJ20200109114018750); the Innovation Project of Optics Valley Laboratory (OVL2021BG004); the Cooperation Project between Hisense Broadband and Huazhong University of Science and Technology.
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. K. Zhong, X. Zhou, Y. Wang, T. Gui, Y. Yang, J. Yuan, L. Wang, W. Chen, H. Zhang, J. Man, L. Zeng, C. Yu, A. P. T. Lau, and C. Lu, “Recent advances in short reach systems,” Proceedings of Optical Fiber Communications Conference (OFC) (2017), paper Tu2D.7. [CrossRef]
2. X. Pang, O. Ozolins, L. Zhang, A. Udalcovs, R. Lin, R. Schatz, U. Westergren, S. Xiao, W. Hu, G. Jacobsen, S. Popov, and J. Chen, “Beyond 200 Gbps per lane intensity modulation direct detection (IM/DD) transmissions for optical interconnects: Challenges and recent developments,” Proceedings of Optical Fiber Communications Conference (OFC) (2019), paper W4I.7.
3. T. Wettlin, S. Calabrò, T. Rahman, J. Wei, N. Stojanovic, and S. Pachnicke, “DSP for high-speed short-reach IM/DD systems using PAM,” J. Lightwave Technol. 38(24), 6771–6778 (2020). [CrossRef]
4. K. Matsuda, R. Matsumoto, H. Miura, K. Onohara, and N. Suzuk, “Hardware-efficient signal processing technologies for coherent PON systems,” J. Lightwave Technol. 37(6), 1614–1620 (2019). [CrossRef]
5. X. Zhou, R. Urata, and H. Liu, “Beyond 1 Tb/s Intra-Data Center Interconnect Technology: IM-DD OR Coherent?” J. Lightwave Technol. 38(2), 475–484 (2020). [CrossRef]
6. G. N. Liu, L. Zhang, T. Zuo, and Q. Zhang, “IM/DD transmission techniques for emerging 5 G fronthaul, DCI, and metro applications,” J. Lightwave Technol. 36(2), 560–567 (2018). [CrossRef]
7. J. Cheng, C. Xie, M. Tang, and S. Fu, “A low-complexity adaptive equalizer for digital coherent short-reach optical transmission systems,” Proceedings of Optical Fiber Communications Conference (OFC) (2019), paper M3H.2.
8. M. Morsy-Osman, M. Sowailem, E. El-Fiky, T. Goodwill, T. Hoang, S. Lessard, and D. V. Plant, “DSP-free ‘coherent-lite’ transceiver for next generation single wavelength optical intra-datacenter interconnects,” Opt. Express 26(7), 8890–8903 (2018). [CrossRef]
9. Y. Chen, X. Wang, Y. Zeng, J. Chen, W. Li, Y. Zhang, M. Zhang, C. Zhao, D. Liu, and M. Tang, “All-optical polarization split of the signal and LO for a bi-directional self-homodyne coherent system,” Opt. Lett. 46(12), 2819–2822 (2021). [CrossRef]
10. W. Li, M. Zhang, Y. Chen, C. Zhao, L. Huo, T. Tong, H. Du, and M. Tang, “Digital in-service relative time delay estimation for SDM self-homodyne coherent systems,” Opt. Express 29(24), 39079–39095 (2021). [CrossRef]
11. X. Zhou, Y. Gao, J. Huo, and W. Shieh, “Theoretical Analysis of Phase Noise Induced by Laser Linewidth and Mismatch Length in Self-Homodyne Coherent Systems,” J. Lightwave Technol. 39(5), 1312–1321 (2021). [CrossRef]
12. M. Y. S. Sowailem, E. El-Fiky, M. Morsy-Osman, Q. Zhuge, T. M. Hoang, S. Paquet, C. Paquet, I. Woods, O. Liboiron-Ladouceur, and D. V. Plant, “Self-homodyne system for next generation intra-datacenter optical interconnects,” Opt. Express 25(22), 27834–27844 (2017). [CrossRef]
13. T. Gui, J. Cao, X. Chen, K. Zheng, S. Yuan, X. Fang, Y. Lei, Q. Zhan, D. Wang, Q. Sui, Z. Li, and L. Li, “Real-time Single-Carrier 800Gb/s DP-64QAM Demonstration using Bi-Directional Self-homodyne Coherent Transceivers with 200 krad/s Endless Active Polarization Controller,” Proceedings of Optoelectronics and Communications Conference (OECC) (2021), paper T5A.5.
14. L. Wang, Y. Chen, X. Wang, C. Zhao, J. Chen, and M. Tang, “Simplest DSP in Self-Homodyne Coherent Transmission Using Bidirectional Adaptive Polarization Controller,” Proceedings of Optical Fiber Communications Conference (OFC) (2021), paper M5G.3.
15. X. Liang, J. D. Downie, J. E. Hurley, H. Su, D. Butler, S. Johnson, and W. Hurley, “Study of Self-Homodyne Coherent System Using Multicore Fiber for Data Center Links,” IEEE Photonics J. 14(4), 7234306 (2022). [CrossRef]
16. B. J. Puttnam, R. S. Luís, J. Sakaguchi, W Klaus, J. Mendinueta, Y. Awaji, N. Wada, and E. Agrell, “Pb/s, Homogeneous, Single-mode, Multi-Core Fiber Systems,” Proceedings of European Conference and Exhibition on Optical Communications (ECOC) (2016), paper 1–3.
17. M. Yang, C. Cai, Y. Liang, L. Shen, Y. Zhu, H. Zhang, C. Yao, Y. Shao, L. Zhang, C. Yan, L. Yang, R. Wang, J. Chu, and J. Wang, “19-Core SDM Self-Homodyne Coherent Transmission using Ultrafast Laser Inscribed Fan-In/Fan-Out Devices,” Proceedings of Optical Fiber Communications Conference (OFC) (2023), paper Tu3I.7.
18. H. Huang, H. Chen, Y. Huang, Q. Zhang, J. Chen, N. K. Fontaine, M. Mazur, R. Ryf, and Y. Song, “MIMO Processing with Linear Beat Interference Cancellation for Space Division Multiplexing Self-Homodyne Coherent Transmission,” J. Lightwave Technol. 40(13), 4136–4149 (2022). [CrossRef]
19. Z. Qu, S. Fu, M. Zhang, M. Tang, P. Shum, and D. Liu, “Analytical Investigation on Self-Homodyne Coherent System Based on Few-Mode Fiber,” IEEE Photonics Technol. Lett. 26(1), 74–77 (2014). [CrossRef]
20. Y. Chen, J. Li, P. Zhu, H. Peng, J. Zhu, Y. Tian, Z. Wu, D. Ge, P. Zhou, K. Liu, Y. Xu, J. Chen, Y. He, and Z. Chen, “Cost effective wavelength reused MDM system for bidirectional mobile fronthaul,” Opt. Express 24(20), 22413–22422 (2016). [CrossRef]
21. Y. Zeng, X. Wang, W. Li, L. Wang, and M. Tang, “Integrated communication and polarization sensing in self-homodyne coherent systems,” Opt. Lett. 47(18), 4684–4687 (2022). [CrossRef]
22. T. Gui, X. Wang, M. Tang, Y. Yu, Y. Lu, and L. Li, “Real-Time Demonstration of Homodyne Coherent Bidirectional Transmission for Next-Generation Data Center Interconnects,” J. Lightwave Technol. 39(4), 1231–1238 (2021). [CrossRef]
23. L. Wang, Y. Zeng, T. Yang, C. Xin, H. Du, X. Wang, and M. Tang, “First Real-time MIMO-free 800Gb/s DP-64QAM Demonstration Using Bi-Directional Self-Homodyne Coherent Transceivers,” Proceedings of European Conference and Exhibition on Optical Communications (ECOC) (2021), paper 1–3.
24. R. Zhang, K. Kuzmin, Y. Chen, and W. I. Way, “800 G/λ Self-Homodyne Coherent Links with Simplified DSP for Next-Generation Intra-Data Centers,” J. Lightwave Technol. 41(4), 1216–1222 (2023). [CrossRef]
25. Y. Tang, K. Ho, and W. Shieh, “Coherent Optical OFDM Transmitter Design Employing Predistortion,” IEEE Photonics Technol. Lett. 20(11), 954–956 (2008). [CrossRef]
26. K. Kikuchi, “Characterization of semiconductor-laser phase noise and estimation of bit-error rate performance with low-speed offline digital coherent receivers,” Opt. Express 20(5), 5291–5302 (2012). [CrossRef]
27. T. N. Huynh, L. Nguyen, and L. P. Barry, “Phase noise characterization of SGDBR lasers using phase modulation detection method with delayed self-heterodyne measurements,” J. Lightwave Technol. 31(8), 1300–1308 (2013). [CrossRef]
28. A. Li, Y. Zhu, W. Peng, Y. Cui, and Y. Bai, “103-GBd PDM-16QAM Coherent Detection Highly Tolerant to Transmitter IQ Impairments Enabled by Real-Valued 4 × 4 MIMO Equalizer,” Proceedings of European Conference and Exhibition on Optical Communications (ECOC) (2019), paper 1–3.
29. M. A. Mestre, F. Jorge, H. Mardoyan, J. Estarán, F. Blache, P. Angelini, A. Konczykowska, M. Riet, V. Nodjiadjim, J-Y. Dupuy, and S. Bigo, “100-Gbaud PAM-4 Intensity-Modulation Direct-Detection Transceiver for Datacenter Interconnect,” Proceedings of European Conference and Exhibition on Optical Communications (ECOC) (2016), paper 1–3.
