Open Access
Add to BibSonomy
Abstract
This paper describes the characterization of a novel directly modulated multi-section laser with a master-slave configuration. Amplitude and phase noise measurements show relative intensity noise values of around −150 dB/Hz and a 3-dB linewidth of around 3 MHz. The laser’s suitability for optical access networks, enabled by the chirp reduction from the external injection locking, is shown by demonstrating unamplified 30 Gbit/s C-band transmission over 25 km and 50 km of single mode fiber using PAM4, as well as 30 Gbit/s PAM4 and PAM8 amplified transmission over 75 km.
Published by Optica Publishing Group under the terms of the Creative Commons Attribution 4.0 License. Further distribution of this work must maintain attribution to the author(s) and the published article's title, journal citation, and DOI.
1. Introduction
The constant increase in data consumption driven by popular services such as video call applications and high definition video services is driving the requirement for higher transmission rates in passive optical networks (PON). This requirement has led to the creation of a taskforce to standardize the use of 25 Gbit/s and 50 Gbit/s per wavelength PONs [1] and multiple demonstrations of the required transmission rates over 10’s of km have been published in recent years using both directly modulated lasers (DML) [2–7], and externally modulated lasers (EML) [8–14]. The use of DMLs offers the advantages of a simpler transmitter architecture and higher output power, but it imposes a significant penalty for the transmission over fiber due to its nonlinear response and complex interaction between laser chirp and fiber dispersion, especially as the networks move to 50 Gbit/s per wavelength [15]. The focus on high speed transmission has thus been centered in the O-band, where fiber dispersion is minimal and longer distances can easily be achieved [3–7,12,13], however the additional fiber loss in the O-band can decrease the power budget. The use of EMLs also alleviates the chirp issue, enabling transmission over greater distances even in the C-band, but at the cost of additional transmitter complexity and insertion loss leading to a reduced power budget [8]. Integrated DMLs with a master-slave configuration [16,17] can solve these problems by using two separate sections: a master laser (ML) operating in continuous wave and a slave laser (SL) which is directly modulated with the driving signal. By injecting the output of the ML into the SL, the chirp can be significantly reduced, enabling improved transmission performance. In comparison with a standard DML, this configuration represents an increase in complexity. However, when comparing it to integrated EMLs with a semiconductor optical amplifier (SOA) solutions [13], the proposed master-slave configuration offers a wider wavelength tunability range at considerably lower complexity and cost due to the simple regrowth-free fabrication process required [16,18]. As the cost of photonic integrated circuits (PIC) decreases, the use of optical injection locking, in conjunction with an SOA, can become a more competitive solution for a low chirp, high output power transmitter for use in short-reach systems such as optical access networks. This can remove the need for discrete optical amplifiers and facilitate C-band time/wavelength division multiplexed (TWDM) systems operating over tens of km’s. Recently, such a master-slave DML PIC was shown to transmit 10 Gbit/s on multiple wavelengths aligning to the NG-PON2 standard [17]. In this work, a customized package has been developed for the PIC and used to demonstrate 30 Gbit/s pulse amplitude modulation (PAM) transmission over up to 75 km of single mode fiber (SMF). The combination of reduced chirp due to the master-slave configuration and a simplified Volterra nonlinear equalizer (VNLE), enables performance below the hard-decision forward error correction (HD-FEC) threshold with much lower laser bandwidth (BW) and equalization complexity than previous demonstrations of DML transmission over the C-band [2]. This demonstration proves the potential of such lasers for next generation, longer range, high-speed PONs.
2. Integrated master-slave directly modulated laser
The device was fabricated using AlGaInAs material with 5 strained quantum wells uniformly grown in the active region on an n-doped In-P substrate. The PIC is composed of four main sections including an ML composed of a gain section (MG) between two reflector sections (MR1 and MR2), a variable optical attenuator (VOA), an SL composed of a gain section (SG) between two reflector sections (SR1 and SR2), and an SOA. A 2-$\mu$m ridge waveguide formed on zinc doped p-type semiconductor material (AlGaInAs) was employed to couple light between adjacent sections and anti-reflective coating was placed on the front and back facets. The reflector sections of the ML and SL were created using high-order surface grating structures defined using electron-beam lithography with a slot depth of 1.35 $\mu$m and a slot width of 1.15 $\mu$m. The length of each of the sections of the PIC can be found in Table 1. The VOA separates both laser sections and controls the power injected from the ML into the SL. The length of the slave gain section was kept short to reduce the photon lifetime and increase the laser BW. The SOA, used to boost the output power, has an angled waveguide to reduce reflections. A ruggedized packaging system (Fig. 1(a)) that allows PIC characterization and testing in the laboratory was designed. The solution is based on a typical metal butterfly package that accommodates a single 10G RF input for the gain section of the SL, several DC inputs to control the bias currents for the other sections and a thermo-electric cooler (TEC) for temperature control. In order to further reduce back-reflections, the package also includes an angled incident optical fiber.
Fig. 1. (a) Picture of the PIC and the package used in this work. (b) LIV plot of the free-running slave laser with the SOA biased at 30 mA. (c) Normalized modulation response of the device with and without optical injection measured using a probe station. (d) Optical spectra of the PIC tuned to a wide range of wavelengths. (e) Modulation response of the packaged device at the operating point used for this work. (f) RIN measurement. (g) Phase noise measurement.
Table 1. Length, bias current at the operating point and typical power consumption of each of the sections of the PIC.
The SL has a typical threshold current of 10 mA under free-running operation (Fig. 1(b)), and of around 12 mA when the ML gain section is biased at 30 mA. The nonlinear response shown by the free-running laser is due to the saturation of the SOA section at the defined bias current of 30 mA for the SOA. Figure 1(c) shows the modulation response of the PIC when the SL is free-running and with injection from the ML, measured with the PIC on a probe station. An increase in the 3-dB BW can be seen when using optical injection. The output frequency from the laser can be tuned over a range of 100’s of GHz by varying the different bias currents while keeping a side mode suppression ratio of around 50 dB (Fig. 1(d)), making the device suitable for TWDM PON applications. The operating bias values used for this work can be found in Table 1, together with typical values of power consumption per section (these values may vary depending on the operating point of the PIC). It can be noted that the power consumption of the PIC is in the order of the 100’s of mW, so the total consumption from the device is dominated by the TEC.
With the SL injected by the ML, the packaged device shows a 3 dB BW of around 8 GHz (Fig. 1(e)). The RIN from the laser was measured using the setup detailed in [19]. The obtained RIN values up to a frequency of 20 GHz are shown in Fig. 1(f). The laser relaxation oscillation at around 6 GHz can be seen, and values below −140 dB/Hz were obtained over the whole frequency range. The linewidth was measured by capturing the spectrum of the laser in continuous-wave operation with a high resolution (20 MHz) optical spectrum analyzer (Fig. 1(g)). A 3-dB linewidth of around 3 MHz was calculated by measuring the 40-dB linewidth of the signal to be 300 MHz.
3. Transmission experiment
The performance of the DML for 30 Gbit/s PAM4 transmission over distances relevant for PONs was tested with two different experimental setups, as presented in Fig. 2: Unamplified transmission over 25 km and 50 km of SMF, and amplified transmission over 75 km of SMF. An erbium doped fiber amplifier (EDFA) was required at the transmitter to reach 75 km transmission due to optical fiber attenuation. A pseudorandom multilevel PAM4 sequence $2^{15}$ symbols long was used in all cases. The 15 GBaud multilevel signal was generated and filtered offline with a 10 GHz root-raised cosine (RRC) filter using MATLAB, and sent to an arbitrary waveform generator (AWG) operating at 90 GSa/s. The output of the AWG, was sent to a 16 dB gain linear RF amplifier before driving the SL of the DML (biased at 37 mA). The amplitude swing of the driving current was optimized for each experiment in order to obtain optimal performance at the receiver. A peak-to-peak current of 35 mA was set for the unamplified experiments, while it was reduced to 25 mA for the amplified transmission. The reduced drive current produces a more linear output at the expense of reducing the extinction ratio of the signal and hence degrading the signal to noise ratio (SNR). This trade-off was found to be beneficial when the amplified system was employed, as the use of the EDFA overcomes the degradation in SNR caused by increased link loss over 75 km. It should be noted that the use of a booster EDFA in this way results in additional fiber nonlinear effects as the optical launch power is increased. The resulting optical signal with an output power of 7 dBm from the DML was, for the unamplified experiment, transmitted over 25/50 km of SMF, while for the amplified one it was attenuated to −6 dBm before entering the EDFA which boosted the power to +22 dBm. An optical band pass filter (OBPF) was placed after the EDFA to reduce out of band amplified spontaneous emission (ASE), resulting in a launch power of +14 dBm into 75 km of SMF. At the receiver, the transmitted optical signal entered a VOA which was used to sweep the received optical power (ROP) at the 20 GHz photodiode (PD). A 23 dB gain RF amplifier boosted the obtained photocurrent, which was captured with the real time oscilloscope (RTO) at a sampling rate of 100 GSa/s. The captured signal was matched filtered offline with an RRC filter and resampled to 1 sample per symbol for digital equalization.
In order to compare the relative impacts of chromatic dispersion, noise and nonlinearity on the performance of the 75 km transmission experiment, a 10 GBaud PAM8 (30 Gbit/s) signal of $2^{15}$ symbols long was generated and filtered offline (10 GHz RRC filter), and transmitted over 75 km of SMF. The increased number of levels of the PAM8 signal makes it more sensitive to nonlinear distortions coming from the complex laser dynamics, while the lower baud rate reduces the penalty coming from fiber dispersion. The same experimental setup and conditions as those associated with the 75 km PAM4 transmission described above were used; with the exception of the peak-to-peak value of the driving current driving the DML. This value was set to 30 mA due to the increased SNR requirements of the PAM8 signal.
4. Simulation
In order to study the influence of injection locking in the performance of the system, a simulation was run to replicate the experimental setup. The simulation includes the bandwidth limitation of the different components, the electrical noise in the signal driving the laser, the laser rate equations with external injection [20], fiber propagation by numerically solving the nonlinear Schrödinger Equation and direct detection with a PD including shot noise, thermal noise and dark current. It should be noted that the purpose of the laser rate equations in this work is to provide an insight on the influence of injection locking on the system performance and not to exactly replicate the dynamics and operation of the multisection integrated device used for the experiment.
A pseudorandom multilevel sequence $2^{17}$ symbols long was used to generate a 15 GBaud PAM4 signal, which was transmitted over 75 km of fiber with a launch power of +14 dBm with and without optical injection.
5. Results
The performance of the experimental system was studied, for transmission distances of 0 km, 25 km, 50 km and 75 km, in terms of the bit error ratio (BER) obtained by counting errors over several captured sequences. In order to overcome the nonlinear distortions coming from the complex dynamics of the DML, a simplified second order VNLE was used. The number of linear taps was set to 21, the number after which the performance was found to stabilize. In order to achieve optimal performance at a minimal computational complexity, the number of nonlinear taps required was obtained by analyzing the performance of the PAM4 experiment at a ROP of −6 dBm. As shown in Fig. 3(a), the performance reaches a floor at 8 nonlinear taps, after which the performance gain becomes marginal. Given this optimal tap length, the computational complexity was further reduced by limiting the number of beating terms of the nonlinear multipliers [21]. Figure 3(b) shows the BER obtained for each transmission scenario with a VNLE as the number of beating terms is increased from 0 (linear equalization) to its maximum value (the full 8 taps of the VNLE). It can be observed how the performance improvements after 3 beating terms are marginal in all cases, so a value of 3 was used (resulting in a complexity reduction of the nonlinear operations in terms of the number of real multiplications of over 55% with respect to the full VNLE). A similar analysis of the required nonlinear taps and beating terms was performed for the amplified 75 km PAM8 transmission, resulting in optimal number of 6 nonlinear taps and 3 beating terms (complexity reduction of over 42%).
Fig. 3. (a) BER vs. nonlinear taps after different transmission distances for PAM4 transmission at ROP −6 dBm. (b) BER vs beating terms after different transmission distances for PAM4 transmission with 8 nonlinear taps at ROP −6 dBm.
As shown in Fig. 4(a), BERs below the HD-FEC limit $3.8{\times }10^{-3}$ are achieved at ROPs as low as −8.5 dBm for back-to-back (B2B) PAM4 transmission, while ROPs of around −7.5 dBm, −6.7 dBm and −5.5 dBm are required after PAM4 transmission distances of 25 km, 50 km and 75 km respectively. An error floor can be seen for the B2B and 25 km transmission at BERs around $10^{-4}$, while the 50 km transmission results are shown up to a power of −5 dBm due to power limitations of the unamplified system. A power penalty of 2 dB at the HD-FEC limit can be seen for the 75 km PAM8 transmission with respect to the 75 km PAM4 results, suggesting that noise and nonlinear distortions are a more critical impairment in the system than chromatic dispersion, even after 75 km of SMF.
Fig. 4. (a) BER vs. ROP after different transmission distances for PAM4 (solid lines) and PAM8 (dashed lines) transmission experiments. (b) BER vs. ROP for transmission simulation with optical injection (solid lines) and free-running slave laser (dashed lines).
The trend of the 75 km PAM4 transmission results suggests that it may outperform the B2B experiment at high ROPs. This may be explained considering two different factors: First, the OBPF, introduced to remove amplified spontaneous emission (ASE) coming from the EDFA, also removes out of band ASE produced by the master and slave sections of the DML. Second, the lower driving current into the DML produces a more linear response with lower extinction ratio, which degrades the performance at low ROPs, when the system is in the noise-limited region, but improves results at high powers where the nonlinear distortions become the main limiting impairment.
Figure 5 shows the received eye diagrams for some of the experiments performed. Figure 5(a-d) show the PAM4 eye diagrams for different configurations at a ROP of −6 dBm, while Fig. 5(e-f) show the PAM8 eye diagrams at a ROP of −2 dBm. Figure 5(a-b) show the eye diagram on the PAM4 back-to-back case with a 21-taps linear equalizer and with the simplified VNLE respectively. A strong eye-skew coming from the nonlinear laser dynamic behavior can be seen when the linear equalizer was used, which was properly compensated by the nonlinear equalization implemented. Figure 5(c-d) show the eye diagrams after 75 km transmission with linear equalization and VNLE respectively. A clear eye-opening degradation can be observed due to the ASE noise added by the EDFA and additional fiber transmission, and eye-skew can be seen for the linear equalizer case. Figure 5(e-f) show the eye diagrams for PAM8 transmission using a linear equalizer and the VNLE respectively. It can be seen how the nonlinear distortions from the system strongly degrade the upper levels when only linear equalization is applied, while the VNLE is capable of recovering the full signal and achieving performance below the HD-FEC limit at a ROP of −2 dBm.
Fig. 5. Experimental eye diagrams at the receiver for: (a) PAM4 B2B, feedforward equalizer (FFE), (b) PAM4 B2B, VNLE, (c) PAM4 75 km, FFE, (d) PAM4 75 km, VNLE, (e) PAM8 75 km, FFE, (f) PAM8 75 km, VNLE. To account for the SNR requirements of each modulation format, PAM4 eye diagrams are shown at a ROP of −6 dBm, while PAM8 eye diagrams are shown at a ROP of −2 dBm.
As previously mentioned, the simulation platform was developed in order to examine the impact of optical injection on the tested transmission scenarios. Figure 4(b) shows the BER vs. ROP results obtained by simulation for the B2B and 75 km transmission cases after applying a simplified VNLE. Dashed lines show the performance of a free-running slave laser and solid lines show the performance with injection locking. A performance improvement can be seen for the free-running case for the B2B transmission due to a higher extinction ratio, which is attributed to the increased modulation response without injection (we note that the external injection increases the relaxation oscillation frequency of the laser, which results in a greater overall modulation bandwidth but also produces a slight decrease in response around the free-running relaxation oscillation frequency in the 2 GHz to 10 GHz range, as illustrated in Fig. 6(a)). The strong distortions induced by the interaction between frequency chirp and fiber dispersion result in significant performance degradation after 75 km of fiber without external injection, while in the injected case good performance is achieved over the relatively long transmission distance as a result of the reduced chirp in the injection locked case. This difference in modulation-induced chirp is illustrated in Fig. 6(b), where the frequency chirp is shown for the optical injection case (blue) and the free-running case (red), showing a significant decrease in chirp when injection locking is used. Figure 7 shows eye diagrams generated using the system simulation described. Figure 7(a-b) present the analog PAM4 eye diagrams at the transmitter for the optical injection and free-running case respectively. A more complex dynamic behavior and a noticeable eye-skew can be seen for the latter, and poor eye-opening due to intersymbol interference (ISI) coming from limited device bandwidth is observed in both cases. Figure 7(c-d) show the B2B eye diagram at a ROP of −6 dBm after VNLE for the injected and free-running case respectively, where the increased extinction ratio of the free-running case results in a better eye diagram once the ISI and nonlinear distortions induced by the laser are compensated by the VNLE. Figure 7(e-f) show the equalized eye diagrams after 75 km transmission at a ROP of −6 dBm for the injected and free-running case respectively. A clear degradation in eye-opening can be seen in the free-running case due to complex chirp-induced distortions that the VNLE was not capable to compensate for.
Fig. 6. (a) Modulation response of the simulated laser using optical injection and free-running slave laser. (b) Simulated frequency chirp as a function of time for a 15 GBaud PAM4 signal using optical injection and free-running slave laser.
Fig. 7. Simulated eye diagrams: (a) PAM4 with optical injection, transmitted analog signal, (b) PAM4 free-running, transmitted analog signal, (c) PAM4 B2B with optical injection, VNLE, (d) PAM4 B2B free-running, VNLE, (e) PAM4 75 km with optical injection, VNLE, (f) PAM4 75 km free-running, VNLE. The equalized eye diagrams are shown at a ROP of −6 dBm.
6. Conclusion
We have shown the characterization results of a multisection laser that can be used for directly modulated short reach systems, with RIN levels of around −150 dB/Hz and an optical linewidth of 3 MHz. We have successfully demonstrated an unamplified 30 Gbit/s PAM4 transmission over up to 50 km of SMF in the C-band using a packaged, wavelength tunable, DML PIC employing optical injection. BER values below the HD-FEC threshold were obtained for ROPs as low as −6.5 dBm by using a low complexity VNLE operating at 1 sample per symbol. EDFA amplified transmission over 75 km of SMF was performed, obtaining below-FEC performance at −5.5 dBm and −3.5 dBm when using 15 GBaud PAM4 and 10 GBaud PAM8 respectively. The simplified VNLE employed in this work is shown to be capable of greatly improving the received signal quality, even after strong degradation from eye-skew effects and power-dependent nonlinearities. We have demonstrated via simulation that the aforementioned results were enabled by the external optical injection technology, which has proven to be suitable for high-speed transmitters capable of operating over longer distances. Improvements in PIC fabrication and packaging that increase the device BW to around 15 GHz, should enable 50 Gbit/s transmission over tens of km’s of SMF. These results are especially relevant in the context of the trends in access network technology, as they fulfill the data requirements imposed by the current standards without the added complexity of external modulators and optical amplifiers, while allowing potential implementation of low cost WDM systems.
Funding
Irish Research Council (EBPPG/2018/53); Enterprise Ireland (DT/2019/0014A); Science Foundation Ireland (12/RC/2276-P2, 13/RC/2077-P2, 18/SIRG/5579).
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. “IEEE 802.3ca-2020 - IEEE Standard for Ethernet Amendment 9: Physical Layer Specifications and Management Parameters for 25 Gb/s and 50 Gb/s Passive Optical Networks,”.
2. H. Xin, K. Zhang, L. Li, H. He, and W. Hu, “50 Gbps PAM-4 Over Up to 80-km Transmission With C-Band DML Enabled by Post-Equalizer,” IEEE Photonics Technol. Lett. 32(11), 643–646 (2020). [CrossRef]
3. R.-Y. Chen, C.-H. Dong, Y.-H. Chang, Y.-J. Chen, W. Lin, C.-C. Wei, and Y.-J. Chiu, “Equalizer-free 25Gb/s NRZ and 50Gb/s PAM-4 Transmission over 50-km SMF using Directly Modulated 1300nm DFB Laser,” in 2019 24th OptoElectronics and Communications Conference (OECC) and 2019 International Conference on Photonics in Switching and Computing (PSC), (IEEE, Fukuoka, Japan, 2019), pp. 1–3.
4. L. Xue, L. Yi, W. Hu, R. Lin, and J. Chen, “Optics-Simplified DSP for 50 Gb/s PON Downstream Transmission using 10 Gb/s Optical Devices,” J. Lightwave Technol. 38(3), 583–589 (2020). [CrossRef]
5. P. P. Baveja, M. Li, H. Zhang, Y. Wang, Y. Liang, H.-H. Liao, I. L. Ho, and J. Zheng, “10 km CWDM Transmission using 56 Gb/s PAM-4 Directly Modulated Lasers for 200G/400G Transceivers,” in 2017 European Conference on Optical Communication (ECOC), (IEEE, Gothenburg, 2017), pp. 1–3.
6. N.-P. Diamantopoulos, W. Kobayashi, H. Nishi, K. Takeda, T. Kakitsuka, and S. Matsuo, “Amplifierless PAM-4/PAM-8 transmissions in O-band using a directly modulated laser for optical data-center interconnects,” Opt. Lett. 44(1), 9 (2019). [CrossRef]
7. W. Wang, P. Zhao, Z. Zhang, H. Li, D. Zang, N. Zhu, and Y. Lu, “First Demonstration of 112 Gb/s PAM-4 Amplifier-free Transmission over a Record Reach of 40 km Using 1.3 um Directly Modulated Laser,” in Optical Fiber Communication Conference Postdeadline Papers, (OSA, San Diego, California, 2018), p. Th4B.8.
8. X. Tang, S. Liu, X. Xu, J. Qi, M. Guo, J. Zhou, and Y. Qiao, “50-Gb/s PAM4 over 50-km Single Mode Fiber Transmission Using Efficient Equalization Technique,” in Optical Fiber Communication Conference (OFC) 2019, (OSA, San Diego, California, 2019), p. W2A.45.
9. J. Chen, A. Tan, Z. Li, T. Xu, L. Liang, W. Chen, Y. Li, and Y. Song, “50-km C-Band Transmission of 50-Gb/s PAM4 Using 10-G EML and Complexity-Reduced Adaptive Equalization,” IEEE Photonics J. 11(1), 1–10 (2019). [CrossRef]
10. Q. Hu, K. Schuh, M. Chagnon, F. Buchali, S. T. Le, and H. Bülow, “50 Gb/s PAM-4 Transmission Over 80-km SSMF Without Dispersion Compensation,” in 2018 European Conference on Optical Communication (ECOC), (2018), pp. 1–3.
11. N. Eiselt, J. Wei, H. Griesser, A. Dochhan, M. H. Eiselt, J.-P. Elbers, J. J. V. Olmos, and I. T. Monroy, “Evaluation of Real-Time 8 × 56.25 Gb/s (400G) PAM-4 for Inter-Data Center Application Over 80 km of SSMF at 1550 nm,” J. Lightwave Technol. 35(4), 955–962 (2017). [CrossRef]
12. K. Zhong, X. Zhou, J. Huo, H. Zhang, J. Yuan, Y. Yang, C. Yu, A. P. T. Lau, and C. Lu, “Amplifier-Less Transmission of Single Channel 112Gbit/s PAM4 Signal Over 40km Using 25G EML and APD at O band,” in 2017 European Conference on Optical Communication (ECOC), (IEEE, Gothenburg, 2017), pp. 1–3.
13. S. Kanazawa, T. Shindo, M. Chen, N. Fujiwara, M. Nada, T. Yoshimatsu, A. Kanda, Y. Nakanishi, F. Nakajima, K. Sano, Y. Ishikawa, K. Mizuno, and H. Matsuzaki, “High Output Power SOA Assisted Extended Reach EADFB Laser (AXEL) TOSA for 400-Gbit/s 40-km Fiber-Amplifier-Less Transmission,” J. Lightwave Technol. 39(4), 1089–1095 (2021). [CrossRef]
14. J. H. Lee, S. H. Chang, J. Y. Huh, S.-K. Kang, K. Kim, and J. K. Lee, “EML based real-time 112 Gbit/s (2 × 56.25 Gbit/s) PAM-4 signal transmission in C-band over 80 km SSMF for inter DCI applications,” Opt. Fiber Technol. 45, 141–145 (2018). [CrossRef]
15. Y. Fu, D. Kong, M. Bi, H. Xin, S. Jia, K. Zhang, W. Hu, and H. Hu, “Computationally efficient 104 Gb/s PWL-Volterra equalized 2D-TCM-PAM8 in dispersion unmanaged DML-DD system,” Opt. Express 28(5), 7070 (2020). [CrossRef]
16. P. M. Anandarajah, S. Latkowski, C. Browning, R. Zhou, J. O’Carroll, R. Phelan, B. Kelly, J. O’Gorman, and L. P. Barry, “Integrated Two-Section Discrete Mode Laser,” IEEE Photonics J. 4(6), 2085–2094 (2012). [CrossRef]
17. A. Sharma, A. Kaszubowska-Anandarajah, M. J. Wallace, G. Jain, F. Smyth, J. Braddell, and P. Anandarajah, “Wavelength Tunable Directly Modulated Laser for TWDM Applications,” in European Conference on Optical Communication, ECOC2021, (Bordeaux, France, 2021), p. We3D.1.
18. Q. Lu, W.-H. Guo, D. Byrne, and J. F. Donegan, “Design of Slotted Single-Mode Lasers Suitable for Photonic Integration,” IEEE Photon. Technol. Lett. 22(11), 787–789 (2010). [CrossRef]
19. V. J. Urick, P. S. Devgan, J. D. McKinney, and J. L. Dexter, “Laser Noise and Its Impact on the Performance of Intensity-Modulation with Direct-Detection Analog Photonic Links,” Tech. rep., Defense Technical Information Center, Fort Belvoir, VA (2007).
20. L. Zhang, R. Dou, and J. Chen, “Characteristics of the injection-locked master-slave lasers,” Appl. Opt. 47(14), 2648 (2008). [CrossRef]
21. N.-P. Diamantopoulos, H. Nishi, W. Kobayashi, K. Takeda, T. Kakitsuka, and S. Matsuo, “On the Complexity Reduction of the Second-Order Volterra Nonlinear Equalizer for IM/DD Systems,” J. Lightwave Technol. 37(4), 1214–1224 (2019). [CrossRef]
