While optical OFDM has been demonstrated for superior transmission performance, its analogue waveform in the time domain challenges many conventional all-optical wavelength converters (AOWC) that are needed for future flexible optical networks. There only exist a few reports on AOWC of OFDM signals, which are mainly based on the low-efficient four-wave mixing. In this paper, we propose an AOWC for OFDM signals by using two-mode injection-locking in a low-cost Fabry–Pérot laser. The control signal and the probe signal at a milliwatt power level are combined and injected into the FP laser. By a proper control, they can be injection-locked to two longitudinal modes in the FP laser and subsequently, the transmission of the probe signal is conditioned by the control signal. We conduct an experimental study on various aspects of this AOWC. Despite a vendor-specified electrical-to-optical (E/O) modulation bandwidth of 2.5 GHz, we find that the optical-to-optical (O/O) modulation bandwidth of AOWC is free from this limit and can be much wider. We examine the linear transfer curve of the AOWC by simply using the OFDM waveforms as the stimulus. The performance tolerance to the wavelength detuning and injected power ratio is also measured. The proposed AOWC can provide a linear transfer function from the control signal to the probe signal to support the random-fluctuated OFDM waveform. We also investigate the maximum capacity of the AOWC by using the adaptive bit-loading OFDM. Finally, we measure the power penalty after the AOWC at two different bit rates to show the tradeoff between the penalty and capacity.
© 2016 Optical Society of America
Optical OFDM signals have been used in various optical transmission experiments. They can achieve superior transmission performance with relatively lower computational complexity, compared with conventional single-carrier transmissions . Unlike conventional digital signals in single-carrier modulations, optical OFDM waveforms in the time domain are analogue and fluctuate like random noise. This unique feature demands the linear conversion to all-optical wavelength converters (AOWC) that are envisioned to facilitate wavelength re-allocation and avoid wavelength contention in future flexible optical networks based on wavelength-division-multiplexed (WDM) optical transmissions . Despite a large number of existing techniques on AOWC , there are only a few reports on OFDM signals. They are based on four-wave mixing (FWM) either in a highly nonlinear fiber (HNLF) [3–6], a semiconductor optical amplifier (SOA) [7,8], or in a silicon waveguide . Another approach uses sum frequency generation (SFG) and difference frequency generation (DFG) nonlinear processes in a periodically-poled lithium-niobate (PPLN) waveguide . However, these nonlinear processes tend to be low-efficient and generally require additional optical amplifiers to boost the signals for enough conversion efficiency. On the other hand, AOWC can use two-mode injection locking (TMIL) in a Fabry–Pérot (FP) laser , whose main advantage is a much higher power-efficiency compared with FWM. The control signal and the probe signal at milliwatt power levels are combined and injected into the FP laser. By a proper configuration, they can be injection-locked to two longitudinal modes in the FP laser and amplified resonantly, and then, the transmission of the probe signal is conditioned by the control signal. TMIL can be used for different applications, including AOWC of conventional single-carrier signals, signal regeneration, return-to-zero (RZ) to non-return-to-zero (NRZ) format conversion, and single to multi-wavelength broadcasting [11–14]. We have recently applied it for all-optical wavelength conversion of OFDM signals for the first time and proved its feasibility .
In this paper, we conduct an extensive investigation of AOWC of OFDM signals using TMIL in an FP laser. TMIL is an all-optical signal process and free from the transport and parasitic effects that limit the electrical-to-optical (E/O) modulation bandwidth [16, 17], and therefore, we can measure a much larger optical-to-optical (O/O) modulation bandwidth of AOWC despite using a low-end FP laser with a vendor-specified 2.5-GHz electrical modulation bandwidth. The analogue waveform of OFDM signals requires a linear transfer curve and we examine the linearity of the AOWC by simply using the OFDM waveforms as the stimulus. TMIL is inherently susceptible to several parameters related to injection locking and consequently, the performance tolerance to the wavelength detuning and injected power ratio is measured. We also investigate the capacity of the AOWC by using the adaptive bit-loading OFDM. Finally, we measure the power penalties after the AOWC at two different bit rates, which show a tradeoff between the penalty and the capacity.
2. Experimental setup of AOWC using TMIL
Figure 1(a) depicts the experimental setup for AOWC of OFDM signals using two-mode injection-locking in the FP laser. The experiment is based on intensity-modulation/direct-detection (IM/DD) optical OFDM signals. A continuous-wave (CW) light emitted from one external cavity laser (ECL) is used as the control signal, and modulated by a 10-GHz Mach-Zehnder (MZ) modulator which is driven by an arbitrary waveform generator operated at the sampling rates from 10 GS/s to 25 GS/s with an analog bandwidth of 14 GHz. The OFDM signals with different order of quadrature amplitude modulation (QAM) for individual OFDM subcarriers are generated by the off-line Matlab program and then loaded into the AWG. Another ECL is employed as the probe signal. The control signal and probe signal are coupled by a 3-dB coupler, passed through an optical circulator and injected into an FP laser. Inside the FP laser, the OFDM signal carried by the control signal is copied onto the probe signal via TMIL. Since TMIL is sensitive to the injected power, wavelength and polarization, both the control and probe laser sources are wavelength and power tunable, and equipped with polarization controllers (PCs). The FP laser employed in this experiment is a low-end and low-cost butterfly packaged multiple-quantum-well (MQW) one with temperature control at 25 °C. The vendor-specified electrical modulation bandwidth is 2.5 GHz, the threshold current is less than 9 mA, and the slope efficiency is 0.11 mW/mA. After AOWC, the output signal is passed through a variable optical attenuator (VOA), an erbium-doped fiber amplifier (EDFA), a tunable optical filter (TOF) with a bandwidth of 0.5 nm, and input into a PIN photodiode with 10-GHz bandwidth. Note that the TOF is needed to filter out the control signal and remove the out-of-band amplified spontaneous emission (ASE) noise of the EDFA, and the VOA and EDFA are used to vary the received optical power before being launched into the PIN photodiode for the BER measurements. Note that the EDFA may be skipped for the low bit rates signals that have a lower receiver sensitivity, but the received optical power is always measured after the TOF. The received RF signal is digitalized by a real-time digital oscilloscope (DPO) and up-loaded for the receiver signal processing and performance evaluation. The DPO's sampling rate is intentionally set at 50 GS/s to avoid the aliasing noise due to the large bandwidth of the photodiode.
The block diagram of Fig. 1(b) shows the details of the digital signal processing for OFDM modulation at the transmitter and OFDM demodulation at the receiver which are generated by Matlab programs. In order to make the output of inverse Fast Fourier transform (IFFT) real-valued, we use the Hermitian symmetry before sending the mapped constellation to the IFFT at the transmitter side . The signal is slightly clipped with a clipping ratio of 15.05dB to control the peak-to average-power ratio (PAPR) of OFDM signal and the cyclic prefix ratio (CP) is 1/8. For simplicity, we use the raw bit rate in this paper, which only excludes the overhead of the CP. 200 OFDM symbols, including 5.1 × 104 information symbols, are used throughout this work to estimate the BER. A Matlab off-line digital signal processing (DSP) program which reverses the main functions of the transmitter DSP is employed to recover the received OFDM signal. The frequency responses of various optical and electronic components are estimated and compensated in the off-line OFDM receiver.
3. Experimental results and discussions
The mechanism of TMIL has been discussed in theory and experiment, and many advantages have been demonstrated for AOWC of on-off keying (OOK) signals [11, 12]. In this paper, we experimentally investigate the AOWC using TMIL in the FP laser with an emphasis on the analogue feature that is required for OFDM signals. Although we have not measured every aspect, some advantages of TMIL in the FP laser are also applicable to our proposed AOWC, such as wide wavelength range among many modes [11, 12]. In addition, our work is mainly for IM/DD as the mechanism resembles cross gain modulation (XGM). Despite that there may also exist cross phase modulation, we can make XGM dominant by controlling the conditions of TMIL. Further, it is possible, but not trivial, to apply the proposed AOWC to complex OFDM signals by using an advanced optical signal processing structure that can sequentially separate, process, and combine the in-phase and quadrature components .
a) Output optical spectra of the FP laser with or without TMIL
Figure 2(a) illustrates the optical spectrum of the free running FP laser, which exhibits a broadband and multiple longitudinal modes from 1525 nm to 1565 nm with an approximate free-spectrum range (FSR) of 1.328 nm. The central output wavelength is 1547 nm and the output power is 3.93 dBm with a 30-mA bias current (3.3 times of its threshold current), which is used throughout this work. The output wavelength of the FP modes can be fine tuned by adjusting the operating temperature via the temperature controller inside the FP laser to accommodate an arbitrary input wavelength within the injection-locking range of the input signal.
We choose two FP longitudinal modes at 1545.82 nm (control signal) and 1543.20 nm (probe signal) for AOWC. Note that our selection is just an example and other selection of the modes could provide similar results as demonstrated in those similar publications [11, 12]. Initially, the FP-LD is injection-locked by the probe signal with an input power of −1.4 dBm. The control signal with an input power of 0.5 dBm is then coupled into the FP laser for TMIL. Due to TMIL, both the control signal and probe signal are resonantly amplified, and the side-mode suppression ratio (SMSR) is more than 45 dB as shown in Fig. 2(b). Figure 2(c) shows the output spectrum after filtering out of the control signal. The control signal is suppressed more than 43 dB, which is large enough to avoid the interference between the probe signal and the control signal.
b) Frequency responses of AOWC using TMIL
Conventionally, the FP laser is directly modulated by varying the bias current (not shown in Fig. 1(a)), which is named as E/O modulation in this paper. When the FP laser is used for AOWC as in Fig. 1(a), the probe signal is modulated by the control signal via TMIL, which is named as O/O modulation in this paper. The small E/O bandwidth of the FP laser does not equal to the O/O bandwidth of the proposed AOWC, which is similar to optical cross-gain modulation (XGM) discussed in [16, 17]. Optical modulation and electrical modulation can directly modulate the carrier density in the same way, but optical modulation circumvents the carrier transport effects and the circuit parasitic which result in an additional low-frequency roll-off in the modulation response, and consequently, the bandwidth of optical-to-optical modulation can be increased over that of the electrical modulation. In addition, according to the analysis in , the O/O modulation bandwidth and the relaxation oscillation frequency are strongly coupled, and the high-speed capability of TMIL mainly depends on the small carrier density variations and high photon densities.
In order to evaluate the O/O modulation bandwidth of the AOWC, we test the frequency response when the AWG is operated at 25 GS/s, which means the measured bandwidth can reach 12.5 GHz. This O/O bandwidth measured by the subcarrier frequency response using channel estimation in the receiver is shown in Fig. 3, where the frequency responses of the transmitter and receiver, shown in the two dashed blocks in Fig. 1(a), have been calibrated out. For comparison, we also show the E/O bandwidth of the FP laser, corresponding to the direct modulation of the FP laser. The poor performance of the E/O modulation is mainly a manifestation of the poor electrical packaging that is traded for a lower cost. Typically, the poor electrical packaging can result in a larger parasitic capacitance which limits the injection current into the active region and leads to the reduction of the high frequency differential efficiency. It seriously affects the E/O bandwidth of the FP laser, but has negligible effect on O/O bandwidth of the AOWC. In fact, we can observe an O/O bandwidth slightly more than 10 GHz, much larger the E/O bandwidth. Although the magnitude drops about 6 dB at 7 GHz, it is quite flat between 7 GHz and 11 GHz. Therefore, the comparison in Fig. 3 proves that the O/O bandwidth of the proposed AOWC is not limited by the low E/O modulation bandwidth of the FP laser, which at least relaxes the requirements of the complicated and expensive electrical packaging.
c) Transfer function of the AOWC using TMIL
In the proposed AOWC, the power level of the probe signal is conditioned by that of the control signal. Therefore, we can plot the transfer function, i.e., the power level of the control signal versus that of the probe signal. Since the OFDM waveforms are inherently analogue, it is important to guarantee a linear transfer function of the proposed AOWC. Firstly, we observe the typical output waveform of the AOWC and its corresponding histogram, shown in Figs. 4(a) and 4(b). The waveform is quite symmetrical around zero and the histogram tightly follows the Gaussian distribution. In general, the histogram of the OFDM waveform follows the Gaussian distribution and Fig. 4(b) clearly shows that this property is preserved after AOWC. This shows that the AOWC provides a linear transfer function without introducing waveform distortion. Note that the amplitude measured by the scope is proportional to the optical power in both Figs. 4 and 5.
Secondly, we directly measure the transfer function by simply using the OFDM waveform as the stimulus. Although this is different to the conventional stimulus of a ramp signal, it is more close to real operation. Figure 5 shows the scattering plots using the OFDM waveforms before and after the AOWC, respectively. The symbol rates in Figs. 5(a) and 5(b) are the same, but the bit rates are different by using different order of modulation. The transfer curves are mostly linear within the range of −10 mV to 10 mV, and the negative slope is simply due to that we use the inverted response of the AOWC using TMIL . The signal swing of the output is also of importance, which corresponds to the extinction ratio in the AOWC of the OOK signal . Since the OFDM signal is analogue, the larger signal swing can be explained as a larger optical modulation index and a better performance of the AOWC. The transfer curve of the higher bit rate tends to be more spreading and further study is under plan to understand the rich information from the scattering plot. Nevertheless, Fig. 5 provides another angle to control the TMIL and helps to explain the degradation of the AOWC.
d) Tolerance to the imperfection of TMIL
The performance of the proposed AOWC relies on the TMIL, which typically requires accurate wavelength control and a large injection power ratio. Since there are two injected signals to the FP laser, the optimal conditions of TMIL for AOWC require a proper control of four parameters, i.e., the respective wavelengths and power levels of the probe and control signal. In addition, in this measurement, we use an OFDM-16-QAM signal.
Firstly, we study the performance sensitivity to wavelength detuning. The reference wavelengths of control signal and probe signal are aligned to two free-running longitudinal modes of the FP laser. The power levels of the injected control signal at wavelength λc and probe signal at wavelength λp are 0.8 and 0 dBm, respectively. Figure 6(a) shows the result. Depending on the wavelength detuning of the probe signal ∆λp, the output signal at the output can be inverted or non-inverted. The former is used in Fig. 6(a). Similar to the discussion of extinction ratio in , for the probe signal, there is an optimal wavelength detuning of 0.116 nm, which corresponds to the largest signal swing of the output signal. Further, there is an optimal value of ∆λc for the minimum BER value for each choice of ∆λp, and the optimal values of ∆λc depends on ∆λp, which is similar to the experimental result in . As shown in Fig. 6(a), the optimal wavelength detuning of the control signal is 0.084nm.
Secondly, the system performance versus injected power is studied when both of the control signal and probe signal are at the optimal wavelength detuning. As shown in Fig. 6(b), the lowest BER depends on both the injected power levels of the probe and control signal. Overall, the larger power level of the control signal outputs the lower BER floors because, according to the detailed numerical analysis in , both the mode shift and the gain saturation caused by the injected control signal can increase the extinction ratio of the converted signal. For the injected probe signal, a larger power is needed for the stable injection locking but not too large to reduce the quality factor of the resonator. When the power of the injected control signal is 0.8 dBm, the best performance of the system with the BER less than 10−4 can be obtained when the injected power of the probe signal is varied from −1 dBm to 4 dBm.
e) Capacity of the AOWC using TMIL
The O/O modulation bandwidth of the proposed AOWC is beyond 10 GHz. Since the proposed AOWC is modulation-format-independent, the capacity of AOWC can be varied by using the signal with different spectrum efficiency. Meanwhile, the required BER before feed-forward error correction (FEC) can be quite different. Therefore, we can measure the BER versus for capacity of the AOWC as a guideline for different BER requirements. For the OFDM signals, the FFT length is 512 and 254 subcarriers are used for data transmission. We investigate the capacity of the AOWC by using the adaptive bit-loading OFDM with a fixed optical power of −0.5 dBm into the photodiode. The bit loading algorithm is mainly based on the SNR gap approximation for m-QAM-based bit loading and Levin–Campello adaptive bit-loading algorithm [21, 22]. The AWG is operated at 20 GS/s and the DPO at 50 GS/s. At the different capacities, the injection conditions may be fine-tuned for the best performance. Therefore, the power level of the injected control signal varies from −0.5 to 1.1 dBm, and the probe signal, from −2 to 0.2 dBm, respectively. Figure 7(a) shows that the BER monotonously increases with the capacity. Apparently, there is a tradeoff between the BER performance and capacity of the wavelength converter. Error-free (within one OFDM block of 5.1 × 104 information symbol) AOWC may be obtained when the capacity is lower than 17.64 Gb/s. When the capacity is increased to 40.798 Gb/s, the BER is the FEC limit of 2.3 × 10−3. As examples, Figs. 7(b)-7(d) show the subcarrier SNR distribution, adaptive bit-loading of the subcarriers, and the subcarrier BER distribution of the OFDM signal at 33.194 Gb/s, respectively. The slightly worse performance close to dc is due to the lower cut-off bandwidth of the ac-coupling circuitries in this experiment. Figure 7(e) shows the successful AOWC of OFDM-8/16/32/64-QAM signals. The constellation points become slightly more divergent after wavelength conversion, and it becomes apparent with the higher-order QAM. The additional noise is partly due to the nonlinear processing of the AOWC.
f) Power penalty after AOWC
Finally, we measure the power penalty after the AOWC, as shown in Figs. 8(a) and 8(b). The bit rates are 10 Gb/s with OFDM-8-QAM (AWG at 10 GS/s) and 17.118 Gb/s (AWG at 20 GS/s) with OFDM-4-QAM, respectively. Since the bit rates are low and the receiver sensitivity is sufficiently low, we do not use the EDFA at the receiver side for this measurement and the received optical power is measured after the TOF. At 10 Gb/s, the power penalty is 1.5 dB at a BER of 10−4 or less than 1.8 dB at a FEC BER limit of 2.3 × 10−3. At 17.118 Gb/s, the power penalty becomes 2.3 dB or 2.6 dB, correspondingly. The increased power penalty of 0.8 dB is mainly because the increased modulation bandwidth is close to the bandwidth of the AOWC. The measurement in Fig. 8 clearly shows the trade-off between the performance and capacity of the proposed AOWC, i.e., we may meet the requirement of a small power penalty for AOWC by using a lower speed signal. Note that the control signal in our experiment is almost chirp-free and ASE-free. If it transmits over optical fiber links with fiber dispersion and optical amplifiers, the proposed AOWC may filter out part of the ASE noise due to the frequency selective nature of the cavity, and reduce the amplitude fluctuation and frequency chirp, as discussed in [12,23]. Subsequently, the proposed AOWC may have negative power penalties.
In this paper, we have experimentally investigated AOWC of OFDM signals using TMIL in a low-cost FP Laser. After injection locking, the SMSR is more than 45 dB. Despite an E/O bandwidth of 2.5 GHz, the O/O bandwidth of AOWC can be larger than 10 GHz. The proposed AOWC can provide a linear transfer function from the control signal to the probe signal to support the random-fluctuated OFDM waveform and therefore, we have measured the linearity by using the OFDM waveforms as the stimulus. We have also measured the system performance tolerance versus injected power and wavelength detuning. By using the adaptive bit-loading OFDM with higher spectral efficiency, the capacity of the AOWC can reach 40.798 Gb/s at the FEC BER limit. Finally, we have measured the power penalty after the AOWC. For 10-Gb/s OFDM-8-QAM, the power penalties are 1.5 dB at a BER of 10−4 and less than 1.8 dB at the FEC BER limit of 2.3 × 10−3, respectively. For 17.118-Gb/s OFDM-4-QAM, they are increased to 2.3 dB and less than 2.6 dB, respectively.
This work was supported in part by National High Technology Research and Development Program of China (863 Program) (2015AA015501) and NSFC (No. 61420106011, No. 61405024, No. 61101095).
References and links
1. W. Shieh and I. Djordjevic, OFDM for Optical Communications, (Academic Press, 2009).
2. J. M. H. Elmirghani and H. T. Mouftah, “All-optical wavelength conversion: technologies and applications in DWDM networks,” IEEE Commun. Mag. 38(3), 86–92 (2000). [CrossRef]
3. S. You, C. Li, Q. Yang, M. Luo, Y. Qiu, X. Xiao, and S. Yu, “Seamless sub-band wavelength conversion of Tb/s-Class CO-OFDM superchannels,” IEEE Photonics Technol. Lett. 26(8), 801–804 (2014). [CrossRef]
4. X. Tang, J. Yu, Z. Dong, W. Jian, and G.-K. Chang, “Polarization insensitive and subchannel-preserved wavelength conversion of 432-Gb/s polarization multiplexed O-OFDM QPSK signals using co-polarization pumps,” in Optical Fiber Communication /National Fiber Optic Engineers Conference (OFC/NFOEC, 2011), paper OThC8. [CrossRef]
5. Z. Dong, J. Yu, H.-C. Chien, L. Chen, and G.-K. Chang, “Wavelength conversion for 1.2Tb/s optical OFDM super-channel based on four-wave mixing in HNLF with digital coherent detection,” in European Conference and Exhibition on Optical Communication (ECOC, 2011), paper Th.11.LeSaleve.5. [CrossRef]
6. C. Li, M. Luo, Z. He, H. Li, J. Xu, S. You, Q. Yang, and S. Yu, “Phase noise canceled polarization-insensitive all-optical wavelength conversion of 557-Gb/s PDM-OFDM signal using coherent dual-pump,” J. Lightwave Technol. 33(13), 2848–2854 (2015). [CrossRef]
7. J. Lu, Z. Dong, and Z. Cao, “All-optical wavelength conversion based on parallel dual-pump four-wave mixing in semiconductor optical amplifier for OFDM optical signal,” Guangzi Xuebao 38(11), 2858–2862 (2009).
8. R. Schnabel, U. Hilbk, T. Hermes, P. Meissner, C. Helmolt, K. Magari, F. Raub, W. Pieper, F. J. Westphal, R. Ludwig, L. Kuller, and H. G. Weber, “Polarization insensitive frequency conversion of a 10-channel OFDM signal using four-wave-mixing in a semiconductor laser amplifier,” IEEE Photonics Technol. Lett. 6(1), 56–58 (1994). [CrossRef]
9. C. Li, C. Gui, X. Xiao, Q. Yang, S. Yu, and J. Wang, “On-chip all-optical wavelength conversion of multicarrier, multilevel modulation (OFDM m-QAM) signals using a silicon waveguide,” Opt. Lett. 39(15), 4583–4586 (2014). [CrossRef] [PubMed]
10. X. Wu, W.-R. Peng, V. Arbab, J. Wang, and A. Willner, “Tunable optical wavelength conversion of OFDM signal using a periodically-poled lithium niobate waveguide,” Opt. Express 17(11), 9177–9182 (2009). [CrossRef] [PubMed]
11. J. Horner and E. Patzak, “Large signal analysis of all-optical wavelength conversion using two-mode injection-locking in semiconductor lasers,” IEEE J. Quantum Electron. 33(4), 596–608 (1997). [CrossRef]
12. H. K. Tsang, L. Y. Chan, S. P. Yam, and C. Shu, “Experimental characterization of dual-wavelength injection-locking of a Fabry–Perot laser diode,” Opt. Commun. 156(4–6), 321–326 (1998). [CrossRef]
13. T. Nikas, A. Bogris, and D. Syvridis, “Two-mode injection-locked FP laser receiver: a regenerator for long-distance stable fiber delivery of radio-frequency standards,” Opt. Lett. 40(6), 886–889 (2015). [CrossRef] [PubMed]
14. C. W. Chow, C. S. Wong, and H. K. Tsang, “All-optical modulation format conversion and multicasting using injection-locked laser diodes,” J. Lightwave Technol. 22(11), 2386–2392 (2004). [CrossRef]
15. X. Yi, J. Dang, J. Zhang, N. Jiang, and K. Qiu, “All-optical wavelength conversion of OFDM signals using two-mode injection-locking in Fabry–Pérot lasers,” in Conference on Lasers and Electro-Optics (CLEO) (OSA, 2016), paper JTh2A.124. [CrossRef]
16. X. Jin, T. Keating, and S. L. Chuang, “Theory and experiment of high-speed cross-gain modulation in semiconductor lasers,” IEEE J. Quantum Electron. 36(12), 1485–1493 (2000). [CrossRef]
17. T. Keating, X. Jin, S. L. Chuang, and K. Hess, “Temperature dependence of electrical and optical modulation responses of quantum-well lasers,” IEEE J. Quantum Electron. 35(10), 1526–1534 (1999). [CrossRef]
19. X. Yi, R. Yu, J. Kurumida, and S. J. B. Yoo, “A theoretical and experimental study on modulation-format-independent wavelength conversion,” J. Lightwave Technol. 28(4), 587–595 (2010). [CrossRef]
20. Y. Luo, F. Li, J. Liu, L. Gan, C. Lu, and P. K. A. Wai, “10-Gb/s all-optical VPN in WDM-PON using injection-locked Fabry–Pérot laser diodes,” IEEE Photonics Technol. Lett. 26(22), 2299–2302 (2014). [CrossRef]
21. A. Garcia-Armada, “SNR gap approximation for M-PSK-based bit loading,” IEEE Trans. Wirel. Commun. 5(1), 57–60 (2006). [CrossRef]
22. T.-N. Duong, N. Genay, M. Ouzzif, J. L. Masson, B. Charbonnier, P. Chanclou, and J. C. Simon, “Adaptive loading algorithm implemented in AMOOFDM for NG-PON system integrating cost-effective and low-bandwidth optical devices,” IEEE Photonics Technol. Lett. 21(12), 790–792 (2009). [CrossRef]
23. P. Seddighian and L. R. Chen, “Noise transfer characterization in wavelength conversion based on injection locking in a Fabry–Perot laser,” J. Lightwave Technol. 28(7), 1087–1093 (2010). [CrossRef]