September 1, 2013 / Vol. 38, No. 17 / OPTICS LETTERS
3355
Photonic microwave amplification for radio-over-fiber links using period-one nonlinear dynamics of semiconductor lasers Yu-Han Hung1 and Sheng-Kwang Hwang1,2,* 1 2
Department of Photonics, National Cheng Kung University, Tainan, Taiwan
Advanced Optoelectronic Technology Center, National Cheng Kung University, Tainan, Taiwan *Corresponding author:
[email protected] Received June 3, 2013; revised July 21, 2013; accepted August 6, 2013; posted August 9, 2013 (Doc. ID 191618); published August 26, 2013
For radio-over-fiber links, microwave-modulated optical carriers with high optical modulation depth are preferred because high optical modulation depth allows generation of high microwave power after photodetection, leading to high detection sensitivity, long transmission distance, and large link gain. This study investigates the period-one nonlinear dynamics of semiconductor lasers for optical modulation depth improvement to achieve photonic microwave amplification through modulation sideband enhancement. In our scheme, only typical semiconductor lasers are required as the amplification unit. The amplification is achieved for a broad microwave range, from less than 25 GHz to more than 60 GHz, and for a wide gain range, from less than 10 dB to more than 30 dB. The microwave phase quality is mainly preserved while the microwave power is largely amplified, improving the signal-to-noise ratio up to at least 25 dB. The bit-error ratio at 1.25 Gbits∕s is better than 10−9 , and a sensitivity improvement of up to at least 15 dB is feasible. © 2013 Optical Society of America OCIS codes: (060.4510) Optical communications; (140.3520) Lasers, injection-locked; (140.5960) Semiconductor lasers; (190.3100) Instabilities and chaos; (350.4010) Microwaves. http://dx.doi.org/10.1364/OL.38.003355
Microwave photonics has attracted considerable attention [1] particularly because of the strong demand in distributing microwaves over long distances through fibers for antenna remoting applications. For these radioover-fiber (RoF) links, the microwave-modulated optical carriers should preferably possess certain features, such as high microwave frequency, low microwave phase noise, optical single-sideband modulation, and high optical modulation depth. The last feature is preferred because it allows generation of high microwave power after photodetection, leading to high detection sensitivity and long transmission distance [2,3]. Various schemes, including passive filtering [2–5] and Brillouin scattering [6,7], have been proposed to increase the optical modulation depth of low-efficiency direct or external modulation commonly adopted for RoF. These schemes mainly rely on considerable power attenuation of the optical carriers and therefore require optical amplifiers for loss compensation to achieve microwave amplification optically. Under continuous-wave optical injection, period-one (P1) nonlinear dynamics can be excited through undamping the relaxation resonance of semiconductor lasers. The optical spectrum exhibits the regeneration of the injection and equally separated oscillation sidebands from the regeneration. Owing to the redshifted cavity resonance enhancement, the lower oscillation sideband is typically much stronger than the upper one. These unique characteristics of the P1 dynamics have attracted increasing research interest for applications in optical signal processing [8–10] and photonic microwave generation [11–18] and transmission [19–21]. Recently, by taking advantage of the intensity asymmetry between the oscillation sidebands, Hung et al. demonstrated that optical double-sideband modulation signals typically generated through direct or external modulation in RoF links 0146-9592/13/173355-04$15.00/0
can be converted into optical single-sideband modulation signals to mitigate the microwave power fading effect [22]. To achieve photonic microwave amplification, the present study investigates the P1 dynamics for optical modulation depth improvement by applying the resonance enhancement of the lower oscillation sideband. This scheme relies on modulation sideband enhancement instead of optical carrier attenuation. The experimental setup of this study is presented in Fig. 1. A typical single-mode distributed-feedback semiconductor laser (Furukawa FRL15DCW5-A81) functions as the proposed amplification unit. Under a bias current of 100 mA and a stabilized temperature of 20°C, the freerunning laser oscillates at 193.38 THz with a power of 17 mW measured at the output of its fiber pigtail. A tunable laser (Anritsu Tunics Plus) generates input optical carriers. To excite the P1 dynamics, the frequency of the input optical carriers is detuned by f i from the free-running frequency of the injected laser. In addition, the power of the input optical carriers is varied by using a power adjuster consisting of an attenuator and/or an
Fig. 1. Experimental setup for the present study. LD, laser diode; TL, tunable laser; PA, power adjuster; PC, polarization controller; C, circulator; EM, external modulator; PG, pattern generator; M, mixer; OSA, optical spectrum analyzer; MSA, microwave spectrum analyzer; PD, photodiode; ET, error tester; LPF, low-pass filter. © 2013 Optical Society of America
3356
OPTICS LETTERS / Vol. 38, No. 17 / September 1, 2013
amplifier and is measured at the output port of the circulator connected to the injected laser. To indicate the injection strength received by the injected laser, an injection ratio ξi , defined as the square root of the power ratio between the input optical carriers and the injected laser at free-running, is used. A polarization controller aligns the polarization of the input optical carriers with that of the injected laser. An external modulator (EOspace AX-AV5-40) superimposes microwaves at frequency f m from a microwave source (Agilent E8257D) on the input optical carriers. Phase-shift keying data from a pattern generator (Anritsu MP2101A) are added on the microwaves through a mixer. The spectral features of the injected laser output are displayed on an optical spectrum analyzer (Advantest Q8384), and also on a microwave spectrum analyzer (Agilent 8564EC) following a photodiode (Discovery Semiconductors DSC20H). The photodetected signal is downconverted to the baseband by mixing it with a local microwave source at f m for the bit-error ratio (BER) analysis performed using an error tester (Anritsu MP2101A). An input optical carrier without microwave modulation at ξi ; f i 1.1; 21 GHz excites the P1 dynamics of the injected laser, as shown in Fig. 2(a). While the input optical carrier regenerates [23], oscillation sidebands sharply appear, which are equally separated from the regeneration by f 0 35 GHz. Since the optical input reduces the necessary gain for the injected laser, the laser cavity resonance redshifts through the antiguidance effect [24]. Accordingly, the lower oscillation sideband is resonantly enhanced as opposed to the upper one, and thus its power level is about 2 dB weaker than that of the regeneration and 19 dB stronger than that of the upper one. By applying this oscillation sideband enhancement, the optical modulation depth of a microwave-modulated (MM) optical input can be increased for photonic microwave amplification. To characterize the enhancement in the lower modulation sideband, as is shortly demonstrated, and to analyze the improvement in optical modulation depth, a sideband-to-carrier ratio (SCR), defined as the power ratio of the lower modulation sideband to the optical carrier, is used in this study [2]. By slightly modulating the input optical carrier at
Fig. 2. (a) Optical spectra of P1 dynamics (black curve), MM input (red curve), and MM output (blue curve) at ξi ; f i 1.1; 21 GHz. For the sake of clear visibility, curves are separated from each other. The X axis is relative to the free-running frequency of the injected laser. (b) Microwave spectra of (a), centering at 35 GHz with a 30 kHz resolution. When measuring the microwave linewidth, the highest resolution of 1 Hz is used.
f m 35 GHz, as shown in Fig. 2(a), an MM input is generated with an input SCR of −35 dB, which indicates an optical modulation depth of about 3.6%. Under the same (ξi , f i ), the injected laser emits an MM output at f m with the lower modulation sideband greatly enhanced by about 29 dB, the upper one moderately amplified by about 8 dB, and the optical carrier slightly reduced by about 4 dB. This results in an output SCR of approximately −2 dB and thus a considerable improvement of the optical modulation depth. Accordingly, as shown in Fig. 2(b), a significant amplification of the microwave power, about 27 dB, is achieved under the same received optical power at the photodiode. To further demonstrate that the microwave amplification is attributable mainly to the enhancement of the lower modulation sideband, Fig. 3(a) presents the power of each spectral component in terms of the input SCR under the same (ξi , f i ) considered in Fig. 2(a). Regardless of the input SCR, each spectral component of the MM output varies slightly around a nominal power level, which is largely determined by the respective spectral component of the P1 dynamics at this (ξi , f i ). Accordingly, when comparing the output with the input, the power level of the optical carrier decreases slightly by about 4 dB, whereas that of the lower modulation sideband enhances substantially, by 11 to 35 dB, to a power level close to that of the optical carrier. Although the upper modulation sideband is amplified at low input SCR values, it is about 22 dB weaker than the lower one. Therefore, as presented in Fig. 3(b), an improved output SCR around −1 dB is obtained, and an amplified output microwave power around −28 dBm is achieved, which is largely determined by the beating between the lower modulation sideband and the optical carrier. For comparison, the input microwave power is also shown and, as expected, increases linearly with the input SCR. As a result, the microwave gain reduces linearly with the input SCR, or equivalently the input microwave power, under the same (ξi , f i ), as shown in Fig. 4(a). To obtain a different microwave gain for an MM input with a fixed SCR and f m , the injected laser can be operated at different (ξi , f i ) for a different output SCR while maintaining f 0 . For example, at ξi ; f i 1.5; 15 GHz where f 0 35 GHz, the P1 dynamics exhibit a lower
Fig. 3. (a) Power of optical carriers (squares), lower modulation sidebands (triangles), and upper modulation sidebands (circles) for MM inputs (open symbols) and MM outputs (filled symbols) as a function of input SCR. (b) Output SCR and microwave power for MM outputs (filled symbols), and microwave power for MM inputs (open symbols) in terms of input SCR. All inputs are kept at ξi 1.1, f i 21 GHz, and f m 35 GHz.
September 1, 2013 / Vol. 38, No. 17 / OPTICS LETTERS
Fig. 4. (a) Microwave gain and (b) phase noise variance ratio in terms of input SCR (circles) for ξi ; f i ; f m 1.1; 21 GHz; 35 GHz and in terms of f m (triangles) under a fixed input SCR of −35 dB. The output SCR as a function of f m (squares) is also shown in (a).
oscillation sideband that is 10 dB weaker than the regeneration but is 14 dB stronger than the upper one. At this (ξi , f i ), an MM output with a −10 dB SCR is generated for the same MM input considered in Fig. 2(a), leading to a gain of 22 dB. The gain can be adjusted linearly from 22 to 28 dB by tuning the output SCR from −10 to −0.4 dB at different (ξi , f i ). To achieve a specific microwave gain for an MM input with a fixed SCR at different f m values, different (ξi , f i ) can be adopted to simultaneously obtain the required f 0 f m and output SCR. Figure 4(a) shows an amplification of about 29 dB for f m 25 to 35 GHz, which results from the obtained output SCR of around −0.8 dB. The lowest achievable f m is limited by the possible (ξi , f i ) leading to the P1 dynamics with f 0 f m before entering into other states, a problem that can be solved by biasing the laser at a different level or by using a different laser. The highest achievable f m , however, is restricted mainly by the bandwidth of the electronics used in this study. Figure 4(a) also shows a similar output SCR of around −0.8 dB for f m > 35 GHz, which demonstrates the feasibility of microwave amplification to a similar level at 29 dB for f m up to at least 60 GHz. The microwave spectral purity of the P1 dynamics is poor because of the laser intrinsic noise [11,12,17,18], which typically leads to a 3 dB microwave linewidth of the order of megahertz. This raises the concern of whether the phase quality of the microwaves would deteriorate after amplification through the proposed P1 dynamics scheme. As demonstrated in Fig. 2(b), the 3 dB microwave linewidth is maintained at the same value of 1 Hz after amplification. This suggests that the phase of the noisy P1 dynamics is locked to the highly correlated MM input, thus preserving the microwave phase quality after amplification. To further demonstrate such preservation, as an example, the single-sideband phase noise at the 2 MHz frequency offset in Fig. 2(b) is estimated and is found to be reduced from about −65 to about −90 dBc∕Hz after amplification. Since the phase noise is estimated as the ratio of the power at a nonzero frequency offset to that at the zero frequency offset, this phase noise reduction of 25 dB results mainly from the power amplification of 27 dB at f m 35 GHz. Similar observations are found for other frequency offsets in Fig. 2(b). This suggests that negligible extra
3357
phase noise is introduced when the microwave power is amplified. The microwave phase quality for each operating condition considered in Fig. 4(a) is also similarly studied. To quantify the phase noise over a broad range, the phase noise variance is estimated by integrating the single-sideband phase noise from the frequency offset of 100 kHz to 2 MHz [17]. The ratio of the phase noise variance between the output and the input is presented in Fig. 4(b). A reduction in the phase noise variance ranging from 7 to 27 dB is observed after amplification. By closely comparing Figs. 4(a) and 4(b), the extent of each phase noise variance reduction is found to result mainly from the level of each corresponding microwave amplification, similar to the aforementioned observation. This demonstrates that negligible extra phase noise is introduced during the amplification process as well for the operating conditions considered in Fig. 4(a). As a result, the microwave phase quality is mostly preserved, whereas the microwave power increases drastically, thus considerably improving the signal-to-noise ratio of the microwaves after amplification. The improvement in the signal-to-noise ratio is observed for the microwaves as well as the data they carry. Figure 5 shows the spectra of the downconverted data with a bit rate of 1.25 Gbits∕s under ξi ; f i ; f m 1.1; 21 GHz; 35 GHz. The data are amplified by about 7 and 10 dB for input SCRs of −15 and −20 dB, respectively. The amplification level of the data is close to that of the microwaves presented in Fig. 4(a). Considering the studied data rate, which is limited by the bandwidth of the electronics, this observation implies that the gain bandwidth around the microwaves is at least of the order of gigahertz. This suggests that the proposed system is applicable for RoF links with very high bit rates. A BER analysis is performed to further characterize the signal quality after amplification. As shown in Fig. 6(a), a BER better than 10−9 is achieved with a sensitivity improvement of about 4 and 5.5 dB for an input SCR of −15 and −20 dB, respectively. These values of sensitivity improvement are approximately half of the values of their respective amplification shown in Fig. 5, because the optical power received by the photodiode is proportional to the square root of the electrical power it generates. A sensitivity improvement of up to at least 15 dB for low input SCR values is therefore feasible for the operating conditions considered in this study, such as those
Fig. 5. Spectra of downconverted data for input SCR equal to (a) −15 dB and (b) −20 dB. Both operating conditions are kept at ξi ; f i ; f m 1.1; 21 GHz; 35 GHz with a bit rate of 1.25 Gbits∕s and a bit sequence of 231 − 1.
3358
OPTICS LETTERS / Vol. 38, No. 17 / September 1, 2013
Fig. 6. (a) BER in terms of received optical power for MM inputs (open symbols) and MM outputs (filled symbols) at input SCR of −15 dB (circles) and −20 dB (squares). (b), (c) Eye diagrams for open and filled squares in (a), respectively, at BER 10−9 . All inputs are kept at ξi 1.1, f i 21 GHz, and f m 35 GHz with a bit rate of 1.25 Gbits∕s and a bit sequence of 231 − 1.
shown in Fig. 4(a), which cannot be demonstrated experimentally because of the power limitation of the electronics used in this study. Representative eye diagrams are shown in Figs. 6(b) and 6(c) to demonstrate the signal stability. Similar time jitters suggest that the phase quality of the data is mainly preserved after amplification. This study investigates the P1 nonlinear dynamics of semiconductor lasers for photonic microwave amplification through optical modulation depth improvement. Since the underlying mechanism relies on modulation sideband enhancement instead of optical carrier attenuation, only a typical semiconductor laser is required as the amplification unit. The amplification is achieved for a broad microwave range, from less than 25 GHz to more than 60 GHz, and for a wide gain range, from less than 10 dB to more than 30 dB. The microwave phase quality, including linewidth and phase noise, is mainly preserved while the microwave power is largely amplified, improving the signal-to-noise ratio up to at least 25 dB. A BER better than 10−9 at 1.25 Gbits∕s with a sensitivity improvement of up to at least 15 dB is therefore feasible. This work is supported by the National Science Council of Taiwan under contract NSC99-2112-M-006013-MY3. References 1. J. P. Yao, J. Lightwave Technol. 27, 314 (2009).
2. C. Lim, M. Attygalle, A. Nirmalathas, D. Novak, and R. Waterhouse, IEEE Trans. Microwave Theory Tech. 54, 2181 (2006). 3. M. Attygalle, C. Lim, and A. Nirmalathas, J. Lightwave Technol. 24, 1703 (2006). 4. Y. Dong, Z. Li, X. Tian, Q. Wang, H. He, C. Liu, Y. Wang, W. Hu, and T. H. Cheng, IEEE Photon. Technol. Lett. 19, 1236 (2007). 5. A. Perentos, F. Cuesta-Soto, M. Rodrigo, A. Canciamilla, B. Vidal, L. Pierno, A. Griol, N. S. Losilla, L. Bellieres, F. LopezRoyo, A. Melloni, and S. Iezekiel, Opt. Express 20, 25478 (2012). 6. K. J. Williams and R. D. Esman, Electron. Lett. 30, 1965 (1994). 7. A. Loayssa, D. Benito, and M. J. Garde, Opt. Lett. 25, 197 (2000). 8. S. C. Chan, S. K. Hwang, and J. M. Liu, Opt. Lett. 31, 2254 (2006). 9. S. K. Hwang, H. F. Chen, and C. Y. Lin, Opt. Lett. 34, 812 (2009). 10. C. H. Chu, S. L. Lin, S. C. Chan, and S. K. Hwang, IEEE J. Quantum Electron. 48, 1389 (2012). 11. T. B. Simpson and F. Doft, IEEE Photon. Technol. Lett. 11, 1476 (1999). 12. S. C. Chan and J. M. Liu, IEEE J. Sel. Top. Quantum Electron. 10, 1025 (2004). 13. M. Pochet, N. A. Naderi, Y. Li, V. Kovanis, and L. F. Lester, IEEE Photon. Technol. Lett. 22, 763 (2010). 14. X. Q. Qi and J. M. Liu, IEEE J. Sel. Top. Quantum Electron. 17, 1198 (2011). 15. Y. S. Yuan and F. Y. Lin, IEEE Photon. J. 3, 644 (2011). 16. A. Quirce and A. Valle, Opt. Express 20, 13390 (2012). 17. J. P. Zhuang and S. C. Chan, Opt. Lett. 38, 344 (2013). 18. T. B. Simpson, J. M. Liu, M. AlMulla, N. G. Usechak, and V. Kovanis, IEEE J. Sel. Top. Quantum Electron. 19, 1500807 (2013). 19. A. Kaszubowska, L. P. Barry, and P. Anandarajah, IEEE Photon. Technol. Lett.14, 1599 (2002). 20. S. C. Chan, S. K. Hwang, and J. M. Liu, Opt. Express 15, 14921 (2007). 21. C. Cui, X. Fu, and S. C. Chan, Opt. Lett. 34, 3821 (2009). 22. Y. H. Hung, C. H. Chu, and S. K. Hwang, Opt. Lett. 38, 1482 (2013). 23. S. Donati and S. K. Hwang, Prog. Quantum Electron. 36, 293 (2012). 24. S. C. Chan, IEEE J. Quantum Electron. 46, 421 (2010).