All-optical broadcast technology based on aluminum-doped highly nonlinear fiber

Abstract. Making use of the four-wave mixing, a realization of all-optical 1×6 broadcast technology based on aluminum-doped highly nonlinear fiber (AL-HNLF) is performed. In the system, a set of experiments at channel intervals of 50, 100, and 200 GHz are conducted, and clear eye diagrams as well as low error performance are obtained with an input optical signal of a continuous wave and a 10-Gb/s return-to-zero on-off keying signal, which mostly benefits from the AL-HNLF used in this system. In detail, resulting from the high stimulated Brillouin scattering threshold of AL-HNLF, more power can be launched into the fiber. Additionally, similar performances of different channel spaces demonstrate the adjustability for this technology. With these distinguishing features, this technology might satisfy the basic expectation of utility.

All-optical broadcast technology based on aluminum-doped highly nonlinear fiber 1 Introduction Since optical fibers possess advantages such as transmission bandwidth, low transmission loss, anti-interference ability, and other characteristics, optical communication 1,2 technology has gained in popularity in the communications infrastructure. Optic-electric-optic (O/E/O) conversion is a result of its development. However, this existing (O/E/O) conversion is at the expense of a complex structure, high cost, and the highest conversion efficiency is limited to an "electronic bottleneck." Therefore, all-optical communication would be a more attractive strategy for communication modes. To be exact, the method of broadcast whose structure is simple as well as giving an advantage in terms of wavelength multiplexing tends to be more popular. 3,4 The dominant implementations of all-optical broadcast are as follows: cross-gain modulation, 5-10 cross-phase modulation, [11][12][13] and four-wave mixing (FWM) for semiconductor optical amplifier (SOA) 14 and FWM for highly nonlinear fiber (HNLF). 15 Since FWM has the capability of signal processing transparently in data format 16 and can be used in both phase and intensity modulations, it attracts people's interest. Furthermore, compared with FWM for SOA, FWM for HNLF has a fast response, flat broadband, and low noise. However, stimulated Brillouin scattering (SBS) 17,18 limits the amount of power that can be launched into a fiber and the maximum nonlinearity that can be achieved from HNLF as well. Recently, a novel type of HNLF called aluminumdoped HNFL (AL-HNLF) has been of note for its SBS threshold, which is much higher than that of a conventional HNLF. 19 With a high SBS threshold, a more efficient nonlinear process can be obtained without SBS suppression. Interestingly, the experiments using AL-HNLF do result in a high performance. The bit error rates (BERs) of most data signals reach 10 −9 , which indicates most signals are obtained with error-free performance. The BERs for the worst signals reach 10 −3 , but their performance can still be improved by error correction.
In this paper, three similar experiments with channel intervals of 50, 100, and 200 GHz are conducted based on AL-HNLF and the results demonstrate a good performance of the broadcast technology based on AL-HNLF. Thus, it turns out to be adjustable and flexible for this technology in practical applications.

Principle of Broadcast
A 10-Gb∕s return-to-zero signal (at the frequency of f s ) with a pump light (at the frequency of f cw ) is injected into the HNLF. As a result of the nonlinear effect in HNLF, two new signals that carry the broadcast information are generated and their frequencies are f 1 and f 2 , respectively. The frequency relationships are given by (1) When the power of the two new signals is high enough to be a pump, another two new signals are produced at the frequencies of (3) Similarly, as long as the new signals meet the phase match condition and have enough power, higher-order signals will be generated. In other words, the information is copied in different wavelengths. Let c denote the velocity of the light in a vacuum. The FWM process in terms of wavelength could be derived by substituting f ¼ ðc∕λÞ into Eqs. (1)-(4). Let us take Eq. (1) for example. The corresponding wavelength relationship is given by *Address all correspondence to: Ju Wang, E-mail: wangju@tju.edu.cn In our experiment, the slight wavelength differences among λ cw , λ s , λ 1 result in the approximate equation ðλ 1 ∕λ cw Þ ≈ 1. Hence, Eq. (6) becomes The relationship in Eq. (7) works for Eqs. (2)-(4) and also works for the higher-order signals that are generated in the FWM process. In our experiments, we apply the approximate equations into the whole data analysis.
As shown in Fig. 1, the system components are the laser diode (LD), polarization controller (PC), intensity modulator (IM), erbium-doped fiber amplifier (EDFA), optical bandpass filter (OBF), 3-dB optical coupler (OC), AL-HNLF [length: 130 m, nonlinear coefficient: 7.4 W −1 km −1 , SBS threshold for 130 m: 1000 mW, attenuation: 15 dB∕km, dispersion slope: 0.011ps∕ðnm 2 kmÞ, effective area: 13.5μm 2 ], tunable band-pass filter (TBF), oscilloscope (OSC), optical spectrum analyzer (OSA), and BER tester. LD1 generates a continuous wave (CW) at a wavelength of 1556 nm (λ s ), modulated by a 10-Gb∕s pseudorandom bit sequence (2 31 -1) electrical data signal. LD2 produces another CW (at the wavelength of λ cw ); for any CW, no additional SBS suppression stages are used in this system. We adjust the PC if necessary before they are coupled into the OC. At the input of EDFA, the average power is 1.1 dBm, and the coupled signal is amplified up to 34 dBm after the EDFA. Then the amplified signal passes through the 3-nm OBF to block amplified spontaneous emission noise, and is finally launched into AL-HNLF with an average input power of 33 dBm. Different wavelengths are filtered through the TBF, which are evaluated in the OSA. The corresponding eye diagrams are shown in the OSC. Meanwhile, a BER tester is applied to analyze the performances of broadcast signals.

Results
The performances for the configurations of a 50-GHz channel interval and 1556.4-nm wavelength of λ cw generated by LD2 are described as follows. Figure 2 depicts the optical spectra at the input and output of the AL-HNLF. As shown in Fig. 2(b), six new signals are detected. The wavelengths are 1555.6, 1556.8, 1555.2, Fig. 1 Experimental system for aluminum-doped highly nonlinear fiber (AL-HNLF) based on broadcast of a 10-Gb∕s return-to-zero signal.  1557.2, 1554.8, and 1557.6 nm, which are identical to the analysis above. Note that due to the device limit, before being fed into the AL-HNLF, the undesired frequency component could not be completely filtered out. Hence, as shown in Fig. 2(a), two remaining frequency components are found at λ 1 and λ 2 , which are caused by the nonlinear process in the EDFA. Figures 3 and 4 show the optical spectra and eye diagrams at different wavelengths of the new generated signals. With a decrease of power density from λ 1 to λ 6 , the performance of the broadcast signals are deteriorated, especially at λ 5 (1554.8 nm) and λ 6 (1557.6 nm). Furthermore, the "0" level performs better than the "1" level. In brief, the clear eye diagrams from λ 1 to λ 4 demonstrate a good performance for the broadcast signals.   The BERs of broadcast signals are shown in Fig. 5. Broadcast signals at λ 1 (1555.6 nm), λ 2 (1556.8 nm), λ 3 (1555.2 nm), and λ 4 (1557.2 nm) run error free with low power, while data signals at λ 5 (1554.8 nm) and λ 6 (1557.6 nm) require a relatively high power to offer an errorfree (10 −9 ) approximation. However, an error floor (10 −3 ) exists for λ 5 (1554.8 nm) and λ 6 (1557.6 nm), which indicates that the performance of λ 5 (1554.8 nm) and λ 6 (1557.6 nm) could be improved with an error-correcting code stage.
The performances for the configurations of 100-GHz channel interval and 1556.8-nm wavelength of λ cw generated by LD2 are described as follows. Figure 6 represents the optical spectra at the input and output of the AL-HNLF. As shown in Fig. 6(b) Figures 7 and 8 show the optical spectra and eye diagrams at different wavelengths for the new generated signals. The analysis is similar to 50 GHz. The BERs of the broadcast signals are shown in Fig. 9. The case is similar to 50 GHz.
The performances for the configurations of a 200-GHz channel interval and 1557.6-nm wavelength of λ cw generated by LD2 are described as follows. Figure 10 describes the optical spectra at the input and output of the AL-HNLF. As shown in Fig. 10       depict the optical spectra and eye diagrams at different wavelengths for the new generated signals. The analysis is similar to 50 GHz. The BERs of broadcast signals are shown in Fig. 13. The case is similar to 50 GHz.

Conclusion
In this paper, three different channel intervals of 50, 100, and 200 GHz broadcast experiments have been implemented, respectively. The clear eye diagrams as well as the low error performance demonstrate the broadcast technology based on AL-HNLF without any SBS suppression stage. Although data signals at λ 1 and λ 6 are not as good as the others, they can still be improved by reducing coupling losses and losses in the AL-HNLF or increasing the magnification of the EDFA. Additionally, though these experiments are conducted at 10 Gb∕s and since FWM possesses the transparent characteristics of fast response and signal processing in data format, similar results can be extracted at other rates. Consequently, these three experiments show that the wavelength is tunable, which indicates a strong application for broadcast technology.
Jinlong Yu received his MS and PhD degrees from Tianjin University, China, in 1994 and 1997, respectively. In 1997, he joined Tianjin University. He became a full professor and group leader in the Optical Fiber Communication Laboratory of Tianjin University. He researches high-speed optical signal processing and functional elements for radio over fiber and RoF applications.
Ju Wang received her BS and MS degrees from Tianjin University in 2008 and 2010, respectively, and her PhD degree from Tianjin University in 2013. Her doctoral work principally concerned optical signal processing, especially all-optical regeneration in highly nonlinear fiber. In 2012, she joined the High-Speed Optical Communications Group at the Technical University of Denmark as a visiting PhD student. Since September 2012, she has been working with the Optical Fiber Communication Laboratory of Tianjin University. Her research interests include all-optical signal processing and optoelectrical oscillators.
Wenrui Wang is an associate professor at Tianjin University and his research interests include optical fiber communication and photonic microwaves.
Shi Jia is a PhD student at Tianjin University.
Qiong Wu is a graduate student at Tianjin University whose current research interests include optical application and sensors.
Lixia Dou is a graduate student at Tianjin University whose current research interests include optical application and sensors.
Biao Huang is a graduate student at Tianjin University whose current research interests include optical application and sensors.