A simple experimental scheme for M-QAM optical signals generation

Abstract

A simple scheme to generate optical quadrature amplitude modulation (QAM) signals is proposed based on different types of delay interferometers (DIs). The simulated results show that 16QAM, 64QAM and 256 QAM optical signals can be generated by 2×2, 3×3 and 4×4 DI, respectively, and the outputs of the proposed scheme are similar to those of the conventional schemes. The operation principle is discussed and the transmission properties of the square 16QAM as well as 64QAM signals are analyzed and compared with common approach.

Keywords

advanced modulation formats / multi-level modulation formats / quadrature amplitude modulation / delay interferometers (DIs)

1 1 Introduction

To reduce the influence of noise, nonlinearity, dispersion, and other factors in a high-speed ultra-long haul (ULH) optical fiber communication system, as well as enhance the optical signal quality, many technologies have been adopted. For example, replacing erbium-doped fiber amplifier (EDFA) with Raman amplifier, optimizing system management of dispersion and power, setting an adjustable dispersion compensator at the end of the link, compensating the polarization mode dispersion (PMD) and high-order dispersion, paving a special fiber (such as LEAF, DSF), introducing forward error correction (FEC) and the modulation format (RZ, CSRZ, DPSK, etc.) technologies and so on 1–623456. The performance that profits from various modulation formats is particularly prominent in a ULH system 7–158910111213141515. It had been validated again in a multi-channel transmission platform of 10 Gbps RZ and NRZ in this paper. The platform is based on dispersion compensation chirped fiber Bragg gratings (CFBGs) that compensate over 2500 km-G.652 fiber.

2 2 RZ signal

If the duty cycle of RZ is 0.5 (half of NRZ), by Fourier transform relations, the corresponding width of the spectrum will have been doubled, as shown in

Fig0 NRZ and RZ signal.(a) Pulses; (b) spectrums

Full size|PPT slide

Fig. 1 16. In the figure, for the unipolar code, if the bit rate of the single channel is B, the required bandwidth of NRZ is not less than 2B (2B is the smallest transmission bandwidth required by Nyquist principles when the filter roll-off coefficient is 1 17); and that of RZ is not less than 4B. Namely, for 10 Gbps systems, at least 0.32 nm (in C-band) channel bandwidth is required. The widening of the spectrum of RZ increases the group dispersion and decreases the spectral efficiency of the system. However, the dispersion tolerance of a 10 Gbps system is great, and the dispersion compensation is not practically difficult. The smallest channel spacing is 0.4 nm (50 GHz) as recommended in the ITU-T, and 10 Gbps RZ does not reduce spectral efficiency in practice. On the other hand, RZ has better nonlinearity tolerance and clock extraction condition, and the power control and clock synchronization are easier. Thus, it can be expected that transmission performance of 10 Gbps RZ would be better in ULH than in NRZ.

RZ code can be generated through code converters (for example, SHF1551 RZ converter) or two-stage modulation method. The latter was adopted in this paper. The two stage modulation methods are classed as the half clock method and the full clock method. In the latter, the pulse modulator is driven by a full clock, in which the modulator work is at the same state as the data modulator (modulation equipment as

Fig0 RZ transmitter and principle.(a) RZ transmitter; (b) principle

Full size|PPT slide

Fig. 2(a)); In the former, the pulse modulator is driven by a half clock, in which the DC bias of the modulator is at the highest point of the transmission curve, drive voltage is 2V_π (Fig. 2(b)). The latter one was chosen in the paper.

3 3 Dispersion compensation module-linear CFBG

The grating mentioned in this paper is a linear chirped fiber-optic Bragg grating that was researched and manufactured by the Institute of Lightwave Technology of Beijing Jiaotong University 18. Lots of advanced techniques were adopted such as hydrogen carrier, the apodizer and the packaging technique. As a result, the grating has perfect performance: the temperature coefficient is between 0.2 and 0.5 pm/°C; the polarization module dispersion is very small, PMD < 2 ps; the reflectivity spectrum is flat, the power spectrum ripple is less than 1 dB (

Fig0 Reflectance spectrum, delay and delay ripple of CFBG

Full size|PPT slide

Fig. 3); the -3 dB bandwidth is about 0.35 nm, enough for the transmission of the 10 Gbps system (including RZ code and CSRZ code); the group delay ripple is not larger than |± 15| ps; the group delay is large, about 2600 ps/nm (the G.652 optical fiber length that can be compensated by CFBG is about 160 km); the insertion loss is low, about 4–5 dBm (the loss is equal to that of 7–8 km DCF, and the DCF of this length can compensate only about 50 km G.652 optical fiber).

4 4 Experiment system

The structure of the system is shown in

Fig0 10 Gbps experimental setup based on CFBG

Full size|PPT slide

Fig. 4. This 10 Gbps platform already had been set up, which is about 2500 km in length, and has 16 wavelength channels. The channel space is 0.8 nm (100 GHz), according to the wavelength standard (in C band) of the ITU-T suggestion. The total 2560 km G.652 optical fiber has 27 segments, 22 segments of 100 km, another 5 segments from 60 km to 90 km. This setting is for easier dispersion management.

The structure of the transmitter includes the tunable laser and two LiNbO₃-MZ modulators, transmitting 9.953 Gbps pseudo-random bit frame (PRBS = 10²³ - 1). The signal-to-noise ratio (SNR) of the RZ is 30.5. The output eye and spectrum (Note: for ease of comparison, two actual results measured of the spectrum were put in the same picture box) of the transmitter are shown in

Fig0 RZ's eye-pattern & RZ and NRZ's spectrum. (a) RZ's eye-pattern; (b) RZ and NRZ's spectrum

Full size|PPT slide

Fig. 5.

In dispersion management, the dispersion of each channel is compensated by 16 CFBGs. Because the group delay of the CFBG (about 160 km G.652 optical fiber, as shown above) is more than 100 km (the amplification spacing), while less than 200 km (2 times amplification spacing), some segments of 60–90 km was needed fiber to adjust the remains of the total dispersion, making the largest absolute value of the remainder of the total dispersion as small as possible (in this paper it is less than 120 km) at any EDFA. At the same dispersion compensation location, the gratings (16 roots) are connected together in series.

In loss management, the input power of the fiber is 5 dBm. At each segment end, the power can be re-boosted to 5 dBm through a light amplifier (EDFA). According to the above, not all segments are connected to the grating. As a result, the loss of each segment is different; the loss of some segments without CFBG is about 21 dBm, and the loss of some segments with CFBG is about 26 dBm. Therefore, the gain of different EDFAs is also different, but all the input power of every segment is certain (5 dBm). The EDFA is the product of our institute 19, the largest gain in small signal is about 30 dB, the maximum output power is 15 dBm, the noise figure is 4–5 dB, and the gain is flat between 1540 nm and 1560 nm. The EDFA location is adjusted with the noise figure, and the small noise figure is set in front.

The receiver is the APD detector. When various biases are placed in the best state, the back to back sensitivity is -20 dBm.

5 5 Test results and analysis

The eye patterns and bit-error rate (BER) curves of NRZ and RZ had been tested at 2000 km and 2500 km of the transmission system (

Fig0 Eye-patterns at different transmission distance. (a) 2081 km-RZ; (b) 2560 km-RZ; (c) 2081 km-NRZ; (d) 2560 km-NRZ

Full size|PPT slide

Figs. 6 and

Fig0 Bit-error ratio curves. (a) RZ; (b) NRZ

Full size|PPT slide

7, taking CH9 for example, the center wavelength is 1557.365 nm, and channel bandwidth is 0.28 nm at 3000 km, slightly narrow to RZ. Other channels are similar to the following results). As shown in the eye patterns (Fig. 6), the two signals have similar shape at 2000 km. The power penalty can also prove this: NRZ's power penalty is about 2 dB and RZ's is about 3 dB (as Fig. 7, bit-rate is 9.953 Gbps, PRBS = 10²³ - 1). At 2500 km, RZ's eye patterns do not seem to have changed (Fig. 6(b)), while that of the NRZ's have obvious deterioration (Fig. 6(d)). At the same time, the NRZ's power penalty is already very large, and the RZ's power penalty is about 5 dB. The RZ's transmission result is much better than NRZ.

In the 10 Gbps system, there are two factors influencing the optical pulse transmission performance: chrom-dispersion and nonlinearity. Two modulation format pulses have the same input power to fiber, and the RZ's transmission performance is better, so it is indicated that RZ's nonlinearity tolerance is bigger than NRZ. Based on the dispersion tolerance, the RZ should have poor performance. However, the opposite result shows that the RZ's nonlinearity tolerance plays a more important role.

References

[1] Gnauck A H, Winzer P J, Konczykowska A, Jorge F, Dupuy J Y, Riet M, Charlet G, Zhu B, Peckham D W. Generation and transmission of 21.4-Gbaud PDM 64-QAM using a novel high-power DAC driving a single I/Q modulator. Journal of Lightwave Technology, 2012, 30 (4): 532-536

[2] Lu G W, Sakamoto T, Kawanishi T. Rectangular QPSK for generation of optical eight-ary phase-shift keying. Optics Express, 2011, 19 (19): 18479-18485

[3] Bakhtiari Z, Wang J, Wu X X, Yang J Y, Nuccio S R, Hellwarth R W, Willner A E. Demonstration of 10-40-Gbaud baud-rate-tunable optical generation of 16-QAM from a QPSK signal using a variable DGD element. In: 2011 Conference on Lasers and Electro-Optics (CLEO), CThX5

[4] Seimetz M, Noelle M, Patzak E. Optical systems with high-order DPSK and star QAM modulation based on interferometric direct detection. Journal of Lightwave Technology, 2007, 25(6): 1515-1529

[5] Kobayashi1 T, Sano A, Masuda H, Ishihara K, Yoshida E, Miyamoto Y, Yamazaki H, Yamada T. 160-Gb/s polarization-multiplexed 16-QAM long-haul transmission over 3123 km using digital coherent receiver with digital PLL based frequency offset compensator. In: 2010 Conference on OFC/NFOEC. 2010, OTuD1

[6] Seimetz M. Performance of coherent optical square 16-QAM-systems based on IQ-transmitters and homodyne receivers with digital phase estimation. In: Optical Fiber Communication Conference, 2006 and the 2006 National Fiber Optic Engineers Conference. 2006, 10

[7] Gnauck A H, Winzer P J, Chandrasekhar S, Liu X, Zhu B, Peckham D W. 10 × 224-Gb/s WDM transmission of 28-Gbaud PDM16-QAM on a 50-GHz grid over 1200 km of fiber. In: 2010 Conference on OFC/NFOEC. 2010, PDPB8

[8] Yu J J. Zhou X, Gupta S, Huang YK, Huang M F. A novel scheme to generate 112.8-Gb/s PM-RZ-64QAM optical signal. IEEE Photonics Technology Letters, 2010, 22(2): 115-117

[9] Yu J J, Zhou X, Huang Y K, Gupta S, Huang M F, Wang T. 112.8-Gb/s PM-RZ-64QAM optical signal generation and transmission on a 12.5 GHz WDM grid. In: 2010 Conference on OFC/NFOEC. 2010, OThM1

[10] Nakazawa M, Okamoto S, Omiya T, Kasai K, Yoshida M. 256 QAM (64 Gbit/s) Coherent Optical Transmission over 160 km with an Optical Bandwidth of 5.4 GHz. IEEE Photonics Techonology Letters, 2010: 185-187

[11] Yamazaki H, Yamada T, Goh T, Sakamaki Y, Kaneko A. 64QAM modulator with a hybrid configuration of silica PLCs and LiNbO3 phase modulators for 100-Gb/s applications. In: 35th European Conference on Optical Communication (ECOC), 2009, 1-4

RIGHTS & PERMISSIONS

2014 Higher Education Press and Springer-Verlag Berlin Heidelberg