## Abstract

The coherent reception of intensity modulated signal has been recently widely investigated, in which the signal is recovered by the envelop detection. High linewidth tolerance is achieved with such scheme. However, strong optical carrier exists during the transmission, which degrades the optical power efficiency. In this paper, an efficient modulation scheme for discrete multi-tone (DMT) signal is proposed based on the Mach-Zehnder modulator (MZM). Different from the traditional intensity modulation, the proposed method employs both intensity and phase domain. Thus, the optical carrier power can be greatly reduced by adjusting the bias of MZM around the null point. By employing coherent detection and digital carrier regeneration (DCR), the carrier suppressed DMT signal can be recovered using envelop detection. No carrier frequency or phase estimation is required. Numerical investigations are made to demonstrate the feasibility, in which significant improvements are found for the proposed DCR method, showing great tolerance against laser linewidth and carrier power reduction. Finally, a 124-Gb/s transmission of polarization-division multiplexed DMT (PDM-DMT) signal is demonstrated over 100-km SSMF, with only −8 dB optical carrier to signal power ratio (CSPR).

© 2016 Optical Society of America

## 1. Introduction

Discrete multi-tone (DMT) technology is one of the orthogonal frequency-division multiplexing (OFDM) based modulation format [1, 2], which employs the Hermitian symmetry to generate real-valued output signal. Since DMT technology transmits data only using the intensity domain and does not use the phase domain, people typically employ the intensity-modulation and direct-detection (IM/DD) structure with a pair of single-electrode modulator and photo detector. Due to the high spectral efficiency and cost effectiveness, DMT technology has been widely employed in digital subscriber lines (DSLs). Recently, DMT has also been widely investigated in short/medium reach optical communications, such as optical access/metro networks and data centers (DC) [3–7].

However, DMT technology suffers from severe power fading, which is induced by the chromatic dispersion (CD). 40% capacity degradation is observed for the 100-Gb/s DMT IM-DD system when increasing the distance from 10 km to 40 km [8, 9]. In order to further extend the reach for high-capacity and intensity-modulation based optical system, coherent detection has been recently widely studied, in which the intensity signal is recovered by envelop detection [10–13]. Such structure is highly preferred in the metro/access networks, due to several features: 1) the low-cost feature, DFB and even VCSEL can be used in the coherent detection, due to the robustness against laser phase noise; 2) higher receiver sensitivity, no Erbium-doped optical fiber amplifier (EDFA) is required, which simplifies the receiver structure; 3) longer reach, digital CD compensation can be made before the recovery of intensity signal; 4) polarization-division multiplexing (PDM) can further double the spectral efficiency (SE).

Employing the Mach-Zehnder modulator (MZM), the intensity modulation of DMT signal is, however, still not efficient enough. Strong optical carrier exists in the generated output, resulting in very high optical carrier-to-signal power ratio (CSPR). In this paper, we propose an intensity/phase modulation scheme for DMT signal, using the MZM. Different from traditional intensity modulation, the proposed scheme employs both intensity and phase domain (0 and π). Consequently, the modulation efficiency is greatly improved if setting the bias around the null point. With the proposed digital carrier regeneration (DCR) method, DMT signal can be restored with weak optical power, which is regenerated in the digital domain. Numerical investigations are made to demonstrate the performance of proposed scheme, showing great tolerance against laser phase noise. And, a fading-free 124-Gb/s transmission of PDM-DMT signal is experimentally demonstrated over a 100-km standard single mode fiber (SSMF), with only −8 dB CSPR.

## 2. Principal of digital carrier regeneration

Figure 1(a) shows the schematic diagram of proposed DCR method. The DCR method typically takes four steps:

- 1) extract the carrier ${c}_{n}$ from the received signal ${r}_{n}$ by a narrow digital low pass filter (LPF), where ${c}_{n}$ is expressed as
and ${r}_{n}$ is expressed as

$${r}_{n}=\left({d}_{n}+A\right)\cdot \mathrm{exp}\left[j{\omega}_{IF}n+j\Delta \theta (n)\right],$$in which $A$ is the constant amplitude of the residual carrier, ${d}_{n}$ is the real-valued base-band DMT signal, ${\omega}_{IF}$ is the intermediate angular frequency, and $\Delta \theta (n)$ is the phase difference between received optical carrier and LO at the n

_{th}sampling point; - 2) amplify the extracted carrier by a factor of $\alpha $, obtaining an amplified digital carrier $\alpha \cdot {c}_{n}$;
- 4) calculate the modulus of the amplified digital carrier, and then subtract this modulus from the modulus of the combined signal

The key advantage of proposed DCR method comes from the digital amplification of the carrier. Since the bias of MZM is set around the null point, the carrier is not “strong” enough to recover the DMT signal. Distortions will be caused by the envelop detection [11, 12], due to the phase modulation (0 or π). Figures 1(b) and 1(c) shows the recovered DMT signal for α = 4 and α = 20, respectively. Severe distortion is observed at the bottom of the recovered DMT signal, when the amplification coefficient is small (α = 4). With increased amplification coefficient (α = 20), the recovered carrier becomes strong enough to guarantee successful signal recovery, as shown in Fig. 1(c). The red line in figure represents the amplified digital carrier. Here, Eq. (4) is valid only when the amplification factor is large enough. However, there is a threshold for the amplification factor, the performance would not be improved by enlarging the amplification factor larger than the threshold. The rests of DSP is similar with the envelop detection based method [10–12]. The advantages of proposed method include: 1) robust against the laser phase noise, thus allowing the use of low-cost DFB or DBR lasers; 2) improved performance from the efficient amplitude and phase modulation; 3) polarization division multiplexing is further supported, doubling the spectrum efficiency.

## 3. Numerical investigations

In order to verify the robustness of proposed method against laser phase noise, simulation is carried out at 100-Gb/s net data rate, with 16-QAM bit loading on each subcarrier. Figure 2 shows the OSNR performance at three different linewidth: 100 kHz, 1 MHz and 10 MHz, respectively. For 100-Gb/s net data rate and 16-QAM bit loading, the required OSNR for a forward error correction (FEC) limit of 3.8x10^{−3} (corresponding to 7% overhead) is about 17.5 dB, for both 100-kHz and 1-MHz linewidth. While, the required OSNR is about 18 dB for the 10-MHz linewidth, resulting in about 0.5-dB OSNR penalty, compared to 100-KHz and 1-MHz linewidth. Thus, the proposed DCR method is thought to support the low-cost DFB lasers, where the linewidth of DFB laser typically ranges from 4MHz to 10MHz, and high modulation order, such as 16-QAM.

The carrier to signal power ratio is another important factor that impacts the overall performance. Figure 3 shows the curve of CSPR versus Q-factor, also under the linewidth of 100 kHz, 1 MHz and 10MHz, respectively. The OSNR is kept at 20 dB, and the bandwidth of LPF is set to 250MHz. From the figure, we see that the extracted carrier will be affected by the filtered noise, especially when CSPR is very low. However, higher CSPR will limit the actual signal power. Tradeoff should be made, and the optimum CSPR is observed at around −10 dB. On the other hand, linewidth slightly impacts the system performance. It is found that the Q-factors are almost the same for linewidth less than 1MHz. But, about 0.6-dB Q-factor penalty is observed when linewidth increases to 10 MHz, at corresponding optimum CSPRs.

We also carried out a simulation comparison to distinguish the proposed DCR method from the widely investigated RF pilot-assisted method [14–16]. For fair comparison, the same 100-Gb/s net data rate is assumed with 16-QAM bit loading on each subcarrier. In the measurements, the OSNR is set to 20 dB to ensure error-free performance (after decoding) using 7% FEC overhead. We measure the curves of CSPR versus Q-factor at linewidth from 100 kHz to 10MHz, as shown in Figs. 4(a)-4(c). A 0.6 dB Q-factor penalty is found for the RF pilot-assisted method at corresponding optimum CSPRs when linewidth is 100 kHz, which is compared to DCR method. The Q-factor penalty increases to 1.2 dB, when linewidth becomes 1 MHz. Further increasing the linewidth to 10 MHz, the Q-factor of RF pilot-assisted method will drop rapidly below the FEC limit. The result shows great advantage of DCR method against laser linewidth. We noted that the optimum CSPR of RF pilot-assisted method slightly differs from the DCR method, which is mainly due to the difference in their tolerance against linewidth.

## 4. Experimental setup

Figure 5 shows the experimental setup for the 124-Gb/s transmission of PDM-DMT signal over 100-km SSMF, based on the proposed DCR method. An external cavity laser (ECL) with 100-kHz linewidth operating at 1550-nm is used as the optical carrier. The DMT signal is generated by an arbitrary waveform generator (AWG) operating at 60-GSa/s. The FFT size is 512, in which the first 5 subcarriers are reserved as the guard band. The rest of subcarriers are bit loaded according to the probing response, resulting in a 124-Gb/s raw bit rate. Hermitian symmetry is then used to produce real-valued output. The length of cyclic prefix (CP) is 1/32 of the FFT size. In each frame, 16 training symbols are followed by 300 payload symbols. The FEC code takes 7% overhead, corresponding to a BER threshold of 3.8x10^{−3}. The DMT signal is modulated via the MZM. The CSPR is changed by adjusting the bias of MZM. Figure 5(a) shows the generated optical spectrum by the proposed method. To emulate polarization multiplexing, a polarizationbeam splitter (PBS) is used with one branch delayed by a symbol period.

The transmission link consists of a single span of 100-km SSMF, and an Erbium-doped optical fiber amplifier (EDFA) to compensate the link loss. The received signal is mixed with a tunable ECL in a polarization diversity 90° hybrid, followed by four balanced detectors. The four RF outputs, corresponding to the two I/Q components in both polarizations, are then fed to a Tektronix real-time scope, acquired at 50-GSa/s sample rate, and processed in a MATLAB program. Figure 5(b) shows the recieved DMT signal with −9 dB CSPR. The offline DSP includes: chromatic dispersion compensation (CDC), digital carrier regeneration (DCR), symbol synchronization, fast Fourier transform (FFT), channel estimation (CE) by 2x2 MIMO, and symbol decision.

## 5. Results and discussion

Figure 6 shows the bit loading profile according to the probing response. For simplicity, only 4-QAM and 16-QAM are used, resulting in a raw data rate of 124 Gb/s. Figure 7 shows the channel response over 100-km SSMF, with or without digital CDC. Due to the coherent detection, both in-phase quadrature information of signal can be detected, and the CD induced fading can be avoided by digital CD compensation, showing the feasibility of proposed method for longer than 100-km transmission reach.

Figure 8 shows the curve of CSPR versus BER at back to back measurement. The raw data rate is 124 Gb/s. The CSPR is adjusted by changing the bias of MZM. Note that the optimum CSPR for the proposed DCR method is found to be around −8 dB, which is much lower than conventional intensity modulation of DMT signal using MZM [10]. Figure 9 shows the OSNR versus BER performance with the optimum CSPR at back-to-back measurement. The required OSNR for the 7% FEC limit of 3.8x10^{−3} is about 18 dB. Figure 10 shows the curve of launch power versus BER for the transmission of 124-Gb/s PDM-DMT signal over 100-km SSMF. The optimum BER (1.6x10^{−4}) is achieved at a launch power of about 1 dBm. The inset of Fig. 10 shows the corresponding recovered 4-QAM and 16-QAM constellations.

## 5. Conclusion

A novel modulation method for DMT signal has been proposed based on the MZM. The proposed method employed both intensity and phase domain of MZM. Low CSPR was achieved by setting the bias of MZM around the null point. By employing coherent detection and proposed DCR method, the optical carrier information could be restored digitally. And, without any frequency offset or phase compensation, the DMT signal could be recovered with the restored digital carrier. Numerical investigations have shown great advantages of DCR method against laser linewidth. In the experiment, we also demonstrated fading-free 124-Gb/s transmission of PDM-DMT signal over 100-km SSMF, which shows potential of the proposed method for metro/access networks.

## Acknowledgment

This work is supported by the National Natural Science Foundation of China (Grant No. 61505154).

## References and links

**1. **W. Shieh and C. Athaudage, “Coherent optical orthogonal frequency division multiplexing,” Electron. Lett. **42**(10), 587–589 (2006). [CrossRef]

**2. **T. Takahara, T. Tanaka, M. Nishihara, Y. Kai, L. Li, Z. Tao, and J. C. Rasmussen, “Discrete multi-tone for 100 Gb/s optical access networks,” in Proc. OFC’14 (2014), paper M2I.1. [CrossRef]

**3. **X. Chen, A. Li, D. Che, Q. Hu, Y. Wang, J. He, and W. Shieh, “High-speed fading-free direct detection for double-sideband OFDM signal via block-wise phase switching,” in Proc. OFC’13 (2013), paper PDP5B.7.

**4. **R. Hu, Q. Yang, M. Luo, S. H. Yu, Z. Zhang, and J. B. Xu, “Two orthogonal carriers assisted 82-Gb/s dual-band DDO-OFDM transmission over 320-km SSMF,” in Proc. OFC’15 (2015), paper TH2A.25.

**5. **J. Tang, R. Giddings, and X. Jin, “Real-time optical OFDM transceivers for PON applications,” in Proc. OFC’11 (2011), paper OTuK3. [CrossRef]

**6. **C. H. Yeh, C. W. Chow, H. Y. Chen, and Y. F. Wu, “10-Gbps OFDM upstream rate by using RSOA-ONU with seeding-light for 75 km long-reach PON access,” in Proc. OFC’12 (2012), paper JTh2A.65. [CrossRef]

**7. **Y. Kai, M. Nishihara, T. Tanaka, T. Takahara, L. Li, Z. Tao, B. Liu, J. C. Rasmussen, and T. Drenski, “Experimental comparison of pulse amplitude modulation (PAM) and discrete multi-tone (DMT) for short-reach 400-Gbps data communication,” in Proc. ECOC’13 (2013), paper Th.1.F.3.

**8. **T. Tanaka, M. Nishihara, T. Takahara, W. Yan, L. Li, Z. Tao, M. Matsuda, K. Takabayashi, and J. C. Rasmussen, “Experimental demonstration of 448-Gbps+ DMT transmission over 30-km SMF,” in Proc. OFC’14 (2014), paper M2I.5. [CrossRef]

**9. **W. Yan, T. Tanaka, B. Liu, M. Nishihara, L. Li, T. Takahara, Z. Tao, J. C. Rasmussen, and T. Drenski, “100 Gb/s optical IM-DD transmission with 10G-class devices enabled by 65 GSamples/s CMOS DAC core,” in Proc. OFC’13 (2013), paper OM3H.1. [CrossRef]

**10. **R. Hu, Q. Yang, M. Luo, J. Li, X. Xiao, and C. Li, “Coherent OFDM-PON using intensity modulation and heterodyne detection,” in Proc. ACP’14 (2014), paper AW4E.1. [CrossRef]

**11. **C. J. Xie, S. Spiga, P. Dong, P. Winzer, A. Gnauck, C. Gréus, M. Ortsiefer, C. Neumeyr, M. Müller, and M. C. Amann, “Generation and transmission of 100-Gb/s PDM 4-PAM using directly modulated VCSELs and coherent detection,” in Proc. OFC’14 (2014), paper Th3K.2. [CrossRef]

**12. **R. Hu, Q. Yang, M. Luo, X. Xiao, X. Xiao, H. Li, and W. Shieh, “A cost-effective 2.5 Gb/s/λbi-directional coherent UDWDM-PON with computationally-efficient DSP,” in Proc. ECOC’14 (2014), paper. Th.2.6.4. [CrossRef]

**13. **Q. Hu, D. Che, Y. Wang, A. Li, J. Fang, and W. Shieh, “Beyond amplitude-only detection for digital coherent system using directly modulated laser,” Opt. Lett. **40**(12), 2762–2765 (2015). [CrossRef] [PubMed]

**14. **S. L. Jansen, I. Morita, N. Takeda, and H. Tanaka, “Pre-emphasis and RF-pilot tone phase noise compensation for coherent OFDM transmission systems,” in Proc. IEEE/LEOS Summer Topical Meetings (IEEE, 2007), paper MA1.2. [CrossRef]

**15. **S. Randel, S. Adhikari, and S. L. Jansen, “Analysis of RF-pilot-based phase noise compensation for coherent optical OFDM systems,” IEEE Photonics Technol. Lett. **22**(17), 1288–1290 (2010). [CrossRef]

**16. **X. Zhou, X. L. Yang, R. Li, and K. P. Long, “Efficient joint carrier frequency offset and phase noise compensation scheme for high-speed coherent optical OFDM systems,” J. Lightwave Technol. **31**(11), 1755–1761 (2013). [CrossRef]