Open Access
Add to BibSonomy
Abstract
We experimentally demonstrate the transmission of multi-order quadrature amplitude modulation (QAM) orthogonal frequency division multiplexing (OFDM) millimeter wave signals at D-band. Meanwhile, the system nonlinearity analysis is also given, which is originated from the unsatisfactory optoelectronic devices, multi-order QAM and the high peak-to-average power ratio (PAPR) of OFDM signal. To alleviate the system nonlinearity, Volterra nonlinearity compensation (VNC) is adopted. Probabilistic shaping (PS) has been regarded as an effective approach to ensure the system robustness. By using the technologies of probabilistic shaping and Volterra nonlinearity compensation, 57.21-Gbit/s 4096-QAM OFDM signal at 117 GHz can be delivered over 13.42-m wireless distance in our experiment, achieving the normalized general mutual information (NGMI) threshold of 0.83 with 25% soft-decision forward-error correction (SD-FEC) overhead. In addition, we simulated the D-band millimeter wave simulation system in VPI software. The NGMI performance between conventional and discrete Fourier transform-spread (DFT-S) 16-QAM OFDM has been compared in simulation at the same optical signal-to-noise ratio (OSNR). The case of conventional 16-QAM OFDM has better performance. To the best of our knowledge, this is the first demonstration of transmission of multi-order QAM OFDM millimeter wave signal at D-band using PS and VNC methods.
© 2023 Optica Publishing Group under the terms of the Optica Open Access Publishing Agreement
1. Introduction
Recently, the demand for large capacity has been stimulated due to the advent of the cloud computing, 4K / 8K video, virtual reality and other services. However, the available frequency resources for 4G are seriously inadequate, only providing a few hundred Mbps for wireless communications. Therefore, in order to enhance mobile broadband and provide sufficient frequency resources, D-band millimeter wave (MMW) (110GHz-170GHz) is an excellent choice [1,2]. However, the lack of electronic devices with large bandwidths will be a major obstacle to realize high speed transmission at D-band. To handle this issue, high-order quadrature amplitude modulation (QAM) can improve spectral efficiency compared with low order modulation formats. Moreover, photonics-assisted millimeter-wave technology is an attractive technique to generate high speed MMW signals, regardless of the bottleneck bandwidth of electronic devices. Recently, high-order QAM based radio over fiber (ROF) system has been widely studied [3–11].
However, for ultra-high order QAM D-band millimeter wave transmission, the optical signal-to-noise ratio (OSNR) requirement increases, the transmission distance is restricted and strong nonlinearity is induced when the number of constellation points increases. It is well known to us, as the order of the QAM signal increases, the transmission distance and transmission rate of the signal will decrease exponentially. And in Ref. [1], the order of the QAM signal is just 64. Therefore, it is interesting to investigate how to achieve high order QAM signal transmission in D-band ROF system.
As we know, probabilistic shaping (PS) technology, as a new technology, can provide an additional shaping gain by changing the probability of constellation points to achieve a Gaussian-like constellation distribution. So, when the transmitted power is fixed, the Euclidean distance of constellation points after PS increases to improve the noise robustness [1,3,12–19]. For example, in Ref. [1], the bit error rate (BER) of 1-Tb/s PS-64-QAM signal is lower than 4 × 10−2 when it is transmitted over 3.1-m wireless distance in D-band millimeter wave system. In general, nonlinearity in the ROF system is caused by mixers, power amplifiers, electrooptical components, photoelectric conversion as well as the fiber. Especially, orthogonal frequency division multiplexing (OFDM) peak-to-average power ratio (PAPR) effect further increases the nonlinearity problem. Instead of traditional OFDM, discrete Fourier transformspread (DFT-S) is a technology which can reduce the PAPR of OFDM effectively [20,21]. In Ref. [21], the BER of 256-QAM OFDM signal is lower than the threshold of 3.8 × 10−3 with DFT-S technology. And, Volterra nonlinearity compensation (VNC) can be used to mitigate the nonlinearity [22–25]. In Ref. [23], by employing VNC, the BER of 100-Gb/s/λ PAM-4 is lower than 1 × 10−2 after 50-km fiber transmission at the power budget of 29.3 dB. Two nonlinear compensation methods are applied in D-band millimeter wave systems, and it is found that the Volterra nonlinear compensation method can optimize the performance of the system.
In this paper, we experimentally demonstrate 10-Gbaud ultra-high order QAM OFDM signal transmission over 13.42-m wireless distance at D-band. In order to improve the noise robustness and mitigate nonlinearity, PS and VNC have been utilized in our system. To our knowledge, this is the first time to demonstrate 57.21-Gbit/s 4096-QAM OFDM signal transmission over 13.42-m wireless distance at D-band.
2. Analysis of system nonlinearity
Figure 1 shows the generation process of millimeter wave signals via a square-law detector. As shown in Fig. 1, a beam of continuous light-wave (CW1) is coupled with CW2 carrying signals by a coupler. Then the coupled signal is beat via a square-law detector to generate millimeter wave signals. Meanwhile, signal-to-signal beating interference (SSBI) is produced at baseband. Assume that the bandwidth of OFDM signal is B. If the frequency space between CW1 and CW2 is equal or greater than B, the SSBI will not be overlapped with transmitted OFDM signal. Otherwise, the SSBI will have an impact on OFDM signal, leading the nonlinearity. In our experiment, the frequency space between CW1 and CW2 is 117 GHz, and the bandwidth of OFDM signal is 10-Gbaud. So, the SSBI will not impact our generated millimeter wave signal in our experiment system. The optical power of CW1(local source), CW2(signal light) and the coupled signal are defined as ${P_{CW}}$, ${P_{signal}}$ and C, respectively, and ${P_{mm}}$ represents the power of generated millimeter wave signal. The power for optical millimeter signal can be expressed as follows:
Due to the insertion loss of the device, $\mathrm{\delta }$ is a constant greater than zero and less than 1.
Because the wireless transmission loss of millimeter wave signal at D-band is relatively large, it is necessary to make sure that the millimeter wave signal at the transmitter side has sufficient power. It can be seen from Eq. (1) that when the optical power of CW2 is equal to the half-coupled signal, namely, ${P_{signal}} = \frac{C}{2}$ or ${P_{signal}} = {P_{CW}}$, the maximum power of millimeter-wave signal can be obtained. However, due to the high PAPR of millimeter wave OFDM signal, if the signal has great power, it will easily suffer saturation effects caused by electro-optical components. This happens in our system because the power of the signal is very high, which is equal to that of the local oscillator (LO). In order to overcome this nonlinear saturation effect, some advanced DSP such as PS and Volterra are indispensably implemented.
3. Analysis of simulation
Figure 2 (a) depicts the 117GHz millimeter wave simulation system by using commercial software VPI. Because of the limitation of the simulation environment, the millimeter wave generated by the square rate detection is not transmitted through the antenna in the wireless environment. Figures 2 (b)-(d) show the optical spectrum after I/Q modulator, optical spectrum after optical coupler, electrical spectrum after PD of the signal.
Fig. 2. (a) The simulation system of 117 GHz millimeter wave. (b) Optical spectrum of signal. (c) Optical spectrum of coupled signal. (d) The electrical spectrum of 117 GHz millimeter wave signal after PD.
It can be seen from Fig. 2 (d), no SSBI will influence our millimeter wave signal, which is in accord with the above analysis. To compare the performance of traditional OFDM and DFTS OFDM for PS-16-QAM, the OSNR has been changed to get their NGMI performance. As shown in Fig. 3, with the increase of OSNR, the NGMI performance of traditional OFDM is better than DFT-S OFDM because DFT-S brings additional problems, such as spreading noise to the entire DFT-S OFDM subcarrier and reducing the optical signal-to-noise ratio.
4. Experiment setup
Figure 4 shows the advanced DSP steps. Firstly, enough pseudo-random sequences are generated from offline MATLAB, and then implemented by probabilistic shaping. The bit per symbol of 256-QAM, 1024-QAM, 2048-QAM and 4096-QAM are 7.2856, 8.0586, 9.0592 and 10.0585 after PS, respectively. Then serial PS signal is converted to parallel signal in frequency-domain. Next, the parallel signal with a length of 1024 can be transformed into time-domain signal by traditional inverse Discrete Fourier transform (IDFT) method and the last 64 points which used as cyclic prefixes are inserted between DFT-S OFDM symbols to reduce ISI. Then, the parallel signals in time-domain are converted into series by parallel to series conversion. The training sequences are added in front of the OFDM signal for synchronization and channel estimation. A frame consists of two training sequences, the first is OOK signal used for synchronization and the second is QPSK signal for channel estimation, and 12 OFDM symbols. Then, 512 zeroes are inserted in front of a frame of signal for over-sampling. The 12 OFDM symbols are the transmitted high-order QAM signals. Figure 5 shows the experimental setup. Firstly, the real part and the imaginary part of signal are generated from off-line MATLAB and loaded into the arbitrary waveform generator (AWG) as I/Q two-branch signals. Then the two paths of continuous waves generated by the AWG are amplified by a parallel amplifier with a gain of 25dB, which are modulated by an I/Q modulator. ECL1 carrying the OFDM data is modulated by I/Q modulator with a bandwidth of 30 GHz. The output signal from I/Q modulator is amplified by PM–EDFA. ECL2 is coupled with optical modulated signal to increase the transmitted power. It is necessary to adjust the power of ECL2 to ensure that the power of MMW signals generated by UTC-PD is enough. the optical spectrum after the coupler is shown in Fig. 6. Subsequently, the coupled signal is amplified through another PM-EDFA, and the polarization state is optimized via the cascaded polarization controller (PC). The output signal from the PC is fed into a tunable optical attenuator to adjust the power of the optical signal into the UTC-PD. Based on the square detection law of UTC-PD, 117GHz millimeter wave signal can be generated.
Fig. 5. The experimental setup of the millimeter wave system at D-band. ECL: external cavity laser, I/Q MOD: I/Q modulator, EA: electrical amplifier, PM-EDFA: polarization-maintaining Erbium-doped fiber amplifier, PC: polarization controller, PBS: polarization beam splitter, PMOC: polarization-maintaining optical coupler, UTC-PD: Uni-Traveling-Carrier photodiode, HA: horn antenna, OSC: oscilloscope, AWG: arbitrary waveform generator, LNA: low noise amplifier, VOA: variable optical attenuator.
The wireless MMW signal is transmitted over 13.42-m wireless distance through the D-band horn antenna with a gain of 25dBi. At the receiver side, it is received through the same antenna as the one at the transmitter side. The received millimeter-wave signal is amplified by a low noise amplifier with a gain of 30dB. The millimeter-wave signal and 112GHz RF source will be mixed in the mixer, and 5GHz IF signal is down-converted. Then the 5GHz RF signal is amplified through the electrical amplifier before being captured by an oscilloscope with a bandwidth of 30GHz and a sample rate of 80GSa/s.
The RF signal received by the oscilloscope is finally processed by offline DSP. The specific process is shown in Fig. 4. Firstly, the 5GHz RF signal is down converted to the baseband, and then the I/Q signals are balanced by employing Schmidt orthography algorithm. Then 10Gbaud signal is obtained through oversampling. The first training sequence is used to detect the synchronization head, and then complex-valued VNC method is performed to compensate for the nonlinearity. After the serial-to-parallel conversion, the 1024-point FFT is used to convert the signal to QAM signal in the frequency-domain. As discussed previously, the second training sequence is used to estimate the channel response of each channel. Finally, the NGMI decision is finished.
5. Experiment results
NGMI is used to measure the system performance and the NGMI threshold with 25% soft decision forward-error-correction (SD-FEC) overhead is 0.83 [26]. It can be observed from Fig. 7 that at the baud rate of 10-Gbaud, with the increase of ROP, the conventional 16-QAM OFDM has good NGMI performance compared with DFT-S 16-QAM OFDM. Therefore, the case of conventional OFDM will be used in the next step.
The performances of different multi-order PS-QAM signals are compared. Figure 8 shows the measured NGMI versus the input optical signal power into the UTC-PD for different QAM modulation cases when the transmission speed is 10-Gbaud. Figure 8 shows the measured NGMI versus the input optical signal power into the UTC-PD for different QAM modulation cases when the transmission speed is 10-Gbaud. As shown in Fig. 8, the NGMI reduces with the increase of the order of QAM, and it is evident that when the ROP reaches 9-dBm, the optimal NGMI still can’t satisfy the 0.83 NGMI threshold due to long wireless transmission distance. So, with poor NGMI performance, the constellation for 256-QAM is basically not clear, and the NGMI performance is continually getting worse for ultra-high order modulation 4096QAM.
As shown in Fig. 9, when the baud rate decreases from 10-Gbaud to 5-Gbaud, the NGMI performance has an obvious improvement, for the case of 256-QAM, the NGMI increases from 0.577 to 0.783, which is still below the 0.83 NGMI threshold. To further improve the NGMI performance, the effect of VNC applied at the receiver side will be discussed.
The NGMI performance versus different Volterra taps at different ROP values in the case of 1024-QAM is tested. As shown in Fig. 10, with the increase of Volterra taps, NGMI performance enhances and at 335 taps, the optimal NGMI performance can be obtained. When the length of the taps continues to increase, the VNC converges gradually and over-fitting effect is caused, so the NGMI performance will degrade. Besides, power saturation will be caused in a large ROP, while a low SNR will also be induced by a small ROP. It can be seen that there is not much difference in NGMI performance at the case of 9-dBm and 10-dBm, although 10dBm is the optimal receiving power. At the case of 10-dBm, the NGMI of 10-Gbaud PS-1024QAM is 0.9321, which surpasses the 0.83 NGMI threshold.
With the optimal Volterra taps, the NGMI performance versus ROP is explored at different QAM modulation formats at the transmission speed of 10-Gbaud. As shown in Fig. 11, with the increase of ROP, the NGMI performance is improved, in particular, the optimal NGMI performance can be achieved at the point of 9-dBm ROP. It is evident that the NGMI of all modulated signals can satisfy the 0.83 NGMI threshold. The situation of lower baud rate is further investigated as shown in Fig. 12. With the decrease of baud rate, the NGMI performance is enhanced.
Fig. 12. NGMI versus Baud rate of different PS-QAM at 10 dBm ROP with VNC; the constellation of (a) 256QAM; (b) 1024QAM; (c) 2048QAM; (d) 4096QAM.
When ROP is fixed at 10dBm and the data rate is 5Gbaud, the corresponding constellation diagrams of PS-256-QAM, PS-1024-QAM, PS-2048-QAM and PS-4096-QAM are shown in Fig. 12 (a), (b), (c) and (d), respectively.
The NGMI threshold, FEC overhead and Net transmission rate for various modulation formats are shown in Table 1. VNC technology has been used for all modulation schemes. PS4096-QAM is taken for example to calculate the net transmission rate. When 25% SD-FEC threshold is used, the code rate c is 0.8, and the net rate R can be calculated as: 10.0585-12 × (10.8) = 7.6585. Therefore, the net transmission rate is 10 × 7.6585 × 12 × 980/(512 + 1088 × 14) = 57.21 Gb/s.
Table 1. Net transmission rate of the different modulation formats
6. Conclusion
We experimentally demonstrate a transmission of 57.21-Gbit/s 4096-QAM OFDM signal at Dband over 13.42-m wireless distance using probabilistic shaping and Volterra nonlinearity compensation techniques. We theoretically analyzed the nonlinearity of the system, which is mainly resulted from the unsatisfactory electrical and photoelectrical devices, multi-order QAM and the high PAPR of OFDM. With the help of complex-valued Volterra nonlinearity compensation approach, at the point of 10-dBm ROP, NGMI of PS-4096-QAM can satisfy the 0.83 NGMI threshold.
Funding
National Natural Science Foundation of China (61527801, 61805043, 61835002, 61675048, 61720106015, 61935005).
Disclosures
The authors declare no conflicts of interest.
Data availability
The raw/processed data can be obtained upon reasonable request.
References
1. X. Li, J. Yu, L. Zhao, K. Wang, C. Wang, M. Zhao, W. Zhou, and J. Xiao, “1-Tb/s Millimeter-Wave Signal Wireless Delivery at D-Band,” J. Lightwave Technol. 37(1), 196–204 (2019). [CrossRef]
2. W. Zhou, L. Zhao, J. Zhang, K. Wang, J. Yu, Y. Chen, S. Shen, R. Shiu, and G. K. Chang, “135-GHz D-band 60Gbps PAM-8 wireless transmission employing a joint DNN equalizer with BP and CMMA,” J. Lightwave Technol. 9, (2020). Early Access.
3. L. Zhao, Y. Zhang, and W. Zhou, “Probabilistically shaped 64QAM OFDM signal transmission in a heterodyne coherent detection system,” Opt. Commun. 434, 175–179 (2019). [CrossRef]
4. K. Takiguchi and N. Nishio, “Flexible data rate THz-wave communication using Nyquist pulses and opticaldomain reception signal processing,” in Proc. Opt. Fiber Commun. Conf., 2020, Art. no. M4I.5.
5. S-R. Moon, M. Sung, J. K. Lee, and S-H Cho, “Cost-Effective Photonics-Based THz Wireless Transmission Using PAM-N Signals in the 0.3 THz Band,” J. Lightwave Technol. 39(2), 357–362 (2021). [CrossRef]
6. X. Li, J. Yu, and J. Xiao, “Demonstration of Ultra-Capacity Wireless Signal Delivery at W-Band,” J. Lightwave Technol. 34(1), 180–187 (2016). [CrossRef]
7. L. Zhang, M. Qiao, S. Wang, Z. Lu, L. Zhang, X. Pang, X. Zhang, and X. Yu., “Nonlinearity-aware optoelectronic terahertz discrete multitone signal transmission with a zerobias diode,” Opt. Lett. 45(18), 5045–5048 (2020). [CrossRef]
8. T. Harter, C. Füllner, J. N. Kemal, S. Ummethala, M. Brosi, E. Bründermann, W. Freude, S. Randel, and C. Koos, “110-m THz wireless transmission at 100 Gbit/s using a Kramers–Kronig Schottky barrier diode receiver,” in Proc. Eur. Conf. Opt.Commun., 2018, pp. 1–3.
9. S. Jia, L. Zhang, S. Wang, W. Li, M. Qiao, Z. Lu, N. M. Idrees, X. Pang, Hao Hu, X. Zhang, L. K. Oxenløwe, and X. Yu, “2 × 300 Gbit/s Line Rate PS-64QAM-OFDM THz Photonic-Wireless Transmission,” J. Lightwave Technol. 38(17), 4715–4721 (2020). [CrossRef]
10. M. Chen, L. Wang, D. Xi, L. Zhang, H. Zhou, and Q. Chen, “Comparison of Different Precoding Techniques for Unbalanced Impairments Compensation in Short-Reach DMT Transmission Systems,” J. Lightwave Technol. 38(22), 6202–6213 (2020). [CrossRef]
11. R-K. Shiu, Y-W. Chen, P-C. Peng, S-J. Su, G-M. Shao, J. Chiu, J-W. Li, and G-K. Chang, “A Simplified Radio-Over-Fiber System for Over 100-km Long-Reach n-QAM Transmission,” IEEE Photonics J. 12(3), 1–8 (2020). [CrossRef]
12. K. Wu, J. He, Z. Zhou, J. He, and J. Shi, “Probabilistic Amplitude Shaping for a 64-QAM OFDM W-Band RoF System,” IEEE Photonics Technol. Lett. 31(13), 1076–1079 (2019). [CrossRef]
13. K. Wang, X. Li, M. Kong, P. Gou, W. Zhou, and J. Yu, “Probabilistically shaped 16QAM signal transmission in a photonics-aided wireless Terahertz-wave system,” Proc. OFC 2018, San Diego, CA, M4J.7.
14. A. Ghazisaeidi, “Advanced C + L-band transoceanic transmission systems based on probabilistically shaped PDM-64QAM,” J. Lightwave Technol. 35(7), 1291–1299 (2017). [CrossRef]
15. J. Cho, X. Chen, S. Chandrasekhar, G. Raybon, R. Dar, L. Schmalen, E. Burrows, A. Adamiecki, S. Corteselli, Y. Pan, D. Correa, B. McKay, S. Zsigmond, P. J. Winzer, and S. Grubb, “Trans-Atlantic field trial using high spectral efficiency probabilistically shaped 64-QAM and single-carrier real-time 250-Gb/s 16-QAM,” J. Lightwave Technol. 36(1), 103–113 (2018). [CrossRef]
16. F. Buchali, F. Steiner, G. Bocherer, L. Schmalen, P. Schulte, and W. Idler, “Rate adaptation and reach increase by probabilistically shaped 64- QAM: An experimental demonstration,” J. Lightwave Technol. 34(7), 1599–1609 (2016). [CrossRef]
17. C. Pan and F. R. Kschischang, “Probabilistic 16-QAM shaping in WDM systems,” J. Lightwave Technol. 34(18), 4285–4292 (2016). [CrossRef]
18. Y. Zhu, A. Li, W. Peng, C. Kan, Z. Li, S. Chowdhury, Y. Cui, and Y. Bai, “Spectrally-efficient single-carrier 400 G transmission enabled by probabilistic shaping,” Proc. OFC 2017, Los Angeles, CA, M3C.1.
19. Z. Wang, Y. Xiao, S. Wang, Y. Yan, B. Wang, Y. Chen, Z. Zhou, J. He, and L. Yang, “Probabilistic shaping based constellation encryption for physical layer security in OFDM RoF system,” Opt. Express 29(12), 17890–17901 (2021). [CrossRef]
20. F. Li, X. Li, J. Zhang, J. Yu, and Z. Cao, “Fiber-Wireless-Fiber Link for DFT-Spread OFDM Signal Transmission at W-Band,” IEEE Photonics Technol. Lett. 27(12), 1273–1276 (2015). [CrossRef]
21. F. Li, J. Yu, Y. Fang, Z. Dong, X. Li, and L. Chen, “Demonstration of DFT-spread 256QAMOFDM signal transmission with cost-effective directly modulated laser,” Opt. Express 22(7), 8742–8748 (2014). [CrossRef]
22. L. Zhao, R. Shiu, W. Zhou, R. Zhang, S. Shen, Y. Li, J. Yu, and G. K. Chang, “Nonlinear Compensation In W-band MM-wave Communication System With Heterodyne Coherent Detection,” Opt. Fiber Technol. 54, 102099 (2020). [CrossRef]
23. J. Zhang, J. Yu, K. Wang, W. Zhou, X. Xiao, J. Xiao, L. Zhao, X. Pan, B. Liu, and X. Xin, “200-Gb/s/λ PDMPAM-4 PON with 29-dB Power Budget based on Heterodyne Coherent Detection,” Optical Fiber Communication Conference (OFC), San Diego, CA, USA, 2019, Th3F.1.
24. L. Chen, J. Xiao, and J. Yu, “Application of Volterra Nonlinear Compensation in 75-GHz mm-Wave FiberWireless System,” IEEE Photonics J. 9(1), 1–7 (2017). [CrossRef]
25. W. Yan, B. Liu, L. Li, Z. Tao, T. Takahara, and J. C. Rasmussen, “Nonlinear Distortion and DSP-based Compensation in Metro and Access Networks using Discrete Multi-tone,” in European Conference and Exhibition on Optical Communication. Optical Society of America, Amsterdam, 2012: Mo.1.B.2.
26. J. Cho, L. Schmalen, and P. J. Winzer, “Normalized Generalized Mutual Information as a Forward Error Correction Threshold for Probabilistically Shaped QAM,” 2017 European Conference on Optical Communication (ECOC), Gothenburg, 2017, pp. 1–3.
