The generation of differential-phase-shift keying (DPSK) signals is demonstrated using a directly modulated passive feedback laser at 10.709-Gb/s, 14-Gb/s and 16-Gb/s. The quality of the DPSK signals is assessed using both noncoherent detection for a bit rate of 10.709-Gb/s and coherent detection with digital signal processing involving a look-up table pattern-dependent distortion compensator. Transmission over a passive link consisting of 100 km of single mode fiber at a bit rate of 10.709-Gb/s is achieved with a received optical power of −45 dBm at a bit-error-ratio of 3.8 × 10−3 and a 49 dB loss margin.
© 2012 OSA
Differential-phase-shift keying (DPSK) is attractive for long-haul, regional, metropolitan and access networks, as its optical signal-to-noise ratio sensitivity and robustness to fiber nonlinearities support long amplified and unamplified transmission distances, and its receiver sensitivity supports large splitting factors . DPSK signals can be recovered using either noncoherent detection (Mach-Zehnder delay interferometer (DI) with a delay equal to the bit period) or digital signal processing (DSP) enabled coherent detection. Coherent technology and DSP has emerged as a feasible solution to accommodate the required receiver sensitivity, wavelength selectivity and reach in optical networks [1, 2]. In particular, coherent detection of DPSK signals has been shown to extend the unrepeated transmission distance for a bit rate of 10.709-Gb/s to 304 km, overcoming 58 dB of link loss . The generation of DPSK signals typically uses a LiNbO3 Mach-Zehnder modulator (MZM) biased at extinction and driven by a nonreturn-to-zero (NRZ) electrical drive signal with a peak-to-peak voltage of 2Vπ . Simple, compact and low cost DPSK transmitters that avoid the use of external modulation are advantageous for many applications in passive optical networks. Some time ago the use of a directly modulated laser (DML) to generate an NRZ-DPSK signal was experimentally demonstrated . A bipolar drive signal with a width of 50 ps for a bit rate of 10-Gb/s and appropriate driving voltage amplitude induced a ± π phase change at bit transitions producing an NRZ-DPSK optical signal .
To improve the performance or to increase the bit rate of a DML based DPSK transmitter, a laser with high modulation bandwidth and precise chirp tuning is needed. Passive feedback lasers (PFLs) which utilize a photon-photon resonance have been shown to exhibit a 3-dB modulation bandwidth in excess of 35 GHz, enabling intensity modulation up to 40-Gb/s [5, 6]. The PFL is a two-section device consisting of a distributed feedback (DFB) laser and an integrated feedback section . If the full modulation bandwidth potential of the PFL is not used, it provides an additional degree of freedom by allowing the chirp, bandwidth and extinction ratio to be tuned, albeit not independently, via the feedback section current .
In this paper we demonstrate the generation of DPSK signals using a directly modulated passive feedback laser at 10.709-Gb/s, 14-Gb/s and 16-Gb/s. The quality of the DPSK signals is assessed using both noncoherent detection for a bit rate of 10.709-Gb/s and coherent detection with DSP involving a look-up table (LUT) pattern-dependent distortion compensation scheme. We extend our previous work  to show transmission of the generated DPSK signal at 10.709-Gb/s over a passive link spanning 100 km of standard single mode fiber (SMF) with pre-amplified coherent detection and less than a 0.2 dB penalty relative to the back-to-back performance.
2. Experimental set-up
The experimental set-up is shown in Fig. 1 . A programmable arbitrary waveform generator (AWG) which consists of a memory block interfaced to a Micram VEGA digital-to-analog converter (DAC) was used for generating the drive signal for the PFL. The DAC has a 6-bit resolution and a variable sampling rate. The sampling rate was set to 21.418-GSa/s, 28-GSa/s and 32-GSa/s for bit rates of 10.709-Gb/s, 14-Gb/s and 16-Gb/s, respectively. A 218 de Bruijn bit sequence was differentially encoded and used in synthesizing the bipolar drive signal for the DFB section of the PFL (denoted IDFB). The strength of the feedback was controlled through the DC current IFB. A variable optical attenuator (VOA) was used to vary the received optical power (ROP) prior to a pre-amplified receiver consisting of an erbium doped fiber amplifier (EDFA) and noncoherent or coherent detection. In the case of noncoherent detection, a DI and balanced photo-detectors followed by an automatic gain control (AGC) amplifier were used to detect the NRZ-DPSK signal. The bit-error-ratio (BER) was measured using an error detector. In the case of coherent detection, the received signal was applied to a polarization controller (PC), to accommodate the use of a dual-polarization receiver which consists of a local oscillator (LO) laser, two 2 × 4 90° hybrids, and balanced photo-detectors. The LO laser had a nominal linewidth of 100 kHz. The in-phase and quadrature components of the electrical field were digitized by two 80-GSa/s analog-to-digital converters (ADCs) using a real-time sampling oscilloscope with 32 GHz electrical bandwidth.
DSP was performed offline in a sequence of steps including (i) normalization and filtering (ii) down-sampling to 2 samples per bit, (iii) dispersion compensation using a 13-tap finite impulse response (FIR) fixed equalizer , (iv) digital square and filter clock recovery , (v) amplitude distortion compensation using the constant modulus algorithm , (vi) carrier frequency recovery using a spectral domain algorithm , (vii) phase recovery using a 4th power sliding window algorithm  for initial phase estimation and (viii) a pre-decision directed LUT pattern-dependent phase distortion compensator . Finally, the BER was determined by direct bit error counting and the error vector magnitude (EVM) was calculated.
2.1 PFL operating condition
The concept of generating a DPSK signal using a DML is illustrated in Fig. 2 . At a bias current well above threshold, a bipolar drive signal shown in Fig. 2(a) with a width of Ts/2 (where Ts denotes the bit period) and appropriate driving voltage amplitude induces a ± π phase change at bit transitions producing DPSK signals at a bit rate of 1/Ts bits per second. The time resolved depiction of the optical intensity, chirp and phase is illustrated in Fig. 2(b). For a DML based DPSK transmitter, the optical phase assumes an NRZ format, with some amount of inherent intensity modulation accompanying the DPSK generation.
To utilize the PFL as a phase modulator, it is essential to operate with a high frequency modulation (FM) efficiency and low extinction ratio (ER). At a bias current well above threshold, the nonlinear modulation dynamics of the PFL are suppressed and the laser exhibits a low ER using the requisite modulation current, while ensuring the magnitude of the transient contribution to the output optical chirp is minimized. On the other hand, a high FM efficiency increases the magnitude of the adiabatic contribution to the output optical chirp and in turn minimizes the residual amplitude distortion inherent to the phase modulation.
The complex-coupled PFL used in this experiment was operated at 1532 nm with a threshold current of 5 mA. The DFB section of the PFL was biased at 60 mA and operated at 20°C. Under these conditions the PFL exhibits a large linewidth enhancement factor, which results in significant adiabatic chirp and negligible transient chirp. The magnitude of this adiabatic chirp can be controlled by varying the feedback current IFB. The measured adiabatic chirp and ER are illustrated in Fig. 3 for 23 mA peak-to-peak modulation current at 1-Gb/s with the bit sequence [1010 . . .]. Initially, increasing the feedback current IFB results in a decrease in the adiabatic chirp and increase in the ER prior to an abrupt change around IFB = 9.5 mA, indicating the onset of the second 2π cycle of the feedback phase . The two points of operations suitable for DPSK generation, which exhibit high FM efficiency and low ER are at IFB = 1 mA and IFB = 10 mA.
The measured small signal intensity modulation (IM) frequency responses of the PFL at these selected points of operation are shown in Fig. 4 . The 3-dB modulation bandwidth was 5.3 GHz at IFB = 1 mA and 7.8 GHz at IFB = 10 mA. Therefore, the feedback current IFB was set to 1 mA for modulation at 10.709-Gb/s and 10 mA for modulation at 14-Gb/s and 16-Gb/s. The modulation potential of the PFL (3-dB bandwidth > 35 GHz) was not fully exploited, as the laser was operated at a point favourable for DPSK generation exhibiting a high FM efficiency and low ER.
2.2 Pattern-dependent distortion compensation
The use of a LiNbO3 MZM to generate DPSK signals results in nearly perfect phase transitions, rendering the system more tolerant to drive voltage imperfections . However, for DML based DPSK transmitters any slight asymmetry in the driving signal or long streams of 1’s or 0’s result in a pattern-dependent distortion. This is eventually manifested as nonlinear phase distortion and can be observed by plotting the unwrapped phase after frequency offset compensation of the coherently detected 10.709-Gb/s DPSK signal as illustrated in Fig. 5 . To compensate for this pattern-dependent phase distortion a decision directed LUT scheme is used as part of the DSP for the coherent receiver . A LUT is an appropriate nonlinear processing unit for mitigating the large signal modulation distortion in DMLs . In the case of noncoherent detection, the pattern-dependent phase distortion is eliminated through the differential detection.
In our particular implementation, the LUT was used in a pre-decision mode as shown in Fig. 6 . The LUT contains the required phase correction ΔФ for the central complex data sample within a specific incoming complex sequence of length 2n + 1. Initially the LUT phase corrections are all zero. A sliding window Фin[k-n:k:k + n] is used to group n symbols to the left and to the right of the incoming phase estimate of interest Фin[k]. A pre-decision on all 2n + 1 complex samples in the sliding window is used to form an address N of the LUT. The phase correction ΔФ[k] is the difference between the incoming central phase sample and its corresponding pre-decision output. The phase correction ΔФ[k] is subsequently added to the corresponding table entry. With a specific bit sequence appearing numerous times the phase corrections will accumulate at a specified LUT entry. A LUT counter (C) is used to track the number of updates at that specified LUT entry. The average correction is subsequently used to correct Фin[k] and produce the output phase Фout[k] prior to BER counting. The LUT is operated on a symbol by symbol basis.
The bipolar drive signal generated using the AWG was designed to synthesize the first derivative of an NRZ raised-cosine pulse train with a roll-off factor of 1.0. The eye-diagram of the electrical drive signal at 10.709-Gb/s is shown in Fig. 7(a) . The eye-diagram of the noncoherently detected DPSK signal is shown in Fig. 7(b) for ROP = −30.4 dBm, while the normalized output optical spectrum is shown in Fig. 7(c).
The constellation diagram of the 10.709-Gb/s DPSK signal after coherent detection and initial carrier phase estimation is shown in Fig. 8(a) , while the final recovered constellation diagram after a 5-bit (n = 2) decision directed LUT distortion compensator is illustrated in Fig. 8(b). It evident from Fig. 8(a) that the pattern dependent effect is manifested as nonlinear phase distortion resulting in an error vector magnitude (EVM) of 18%. This pattern dependent effect is mitigated using the LUT approach, reducing the nonlinear phase distortion as shown in Fig. 8(b) and resulting in an EVM of 5.3%. In Fig. 8(c), the steady-state value of the phase correction ΔΦ is plotted against the LUT binary address for the 5-bit LUT (n = 2), revealing the specific pattern dependent sequences of 1’s and 0’s requiring strong averaged phase correction.
In the context of coherent access networks, the dependence of the BER on the ROP for both noncoherent and coherent detection is presented in Fig. 9 for a 10.709-Gb/s NRZ-DPSK signal at back-to-back. The performance of the system with coherent detection after 100 km of SMF transmission is also plotted. In all the coherent detection measurements, 267,500 bits were used to estimate each BER value and a 5-bit (n = 2) LUT was employed.
It is evident from Fig. 9, that the PFL DPSK transmitter can generate high quality signals intended for both noncoherent and coherent detection with no apparent BER floor. Assuming a forward error correction (FEC) limit at BER = 3.8 × 10−3, coherent detection offers a 5.7 dB improvement in the receiver sensitivity over noncoherent detection at back-to-back. A small penalty (< 0.2 dB) at the FEC limit was induced after 100 km transmission, where dispersion compensation was performed at the receiver using a 13-tap FIR equalizer. The ROP at the FEC limit was −45 dBm. The output power of the PFL was 4 dBm; this allows the PFL-DPSK transmitter to operate within a 49 dB loss margin. The coherent receiver sensitivity could be improved using an optimized receiver, as opposed to a real-time oscilloscope.
To illustrate the potential of exploiting the increased PFL bandwidth obtained for IFB = 10 mA, 14-Gb/s and 16-Gb/s DPSK signals were generated and coherently detected without exhibiting a BER floor and with EVMs of 5.7% and 6.8% respectively. The recovered constellation diagrams after using the LUT pattern dependent distortion compensation are shown in Fig. 10 .
In this paper, we have utilized the large modulation bandwidth and fine chirp tuning capabilities of a PFL in generating DPSK signals at 10.709-Gb/s, 14-Gb/s and 16-Gb/s. The DPSK signal was subsequently recovered using either noncoherent detection for a bit rate of 10.709-Gb/s or coherent detection with DSP involving a LUT pattern dependent distortion compensator. Transmission over 100 km of SMF was achieved with a loss margin of 49 dB.
We wish to thank MICRAM for their support.
References and links
1. H. Rohde, S. Smolorz, J. S. Wey, and E. Gottwald, “Coherent optical access networks,” Proc. Conference on Optical Fiber Communication, paper OTuB1 (2011).
2. C. Wree, S. Bhandare, D. Becker, D. Mohr, and A. Joshi, “Repeaterless 10.7-Gb/s DPSK transmission over 304 km of SSMF using a coherent receiver and electronic dispersion compensation,” IEEE Photon. Technol. Lett. 20(6), 407–409 (2008). [CrossRef]
3. A. H. Gnauck and P. J. Winzer, “Optical phase-shift-keyed transmission,” J. Lightwave Technol. 23(1), 115–130 (2005). [CrossRef]
4. R. S. Vodhanel, A. F. Elrefaie, M. Z. Iqbal, R. E. Wagner, J. L. Gimlett, and S. Tsuji, “Performance of directly modulated DFB lasers in 10-Gb/s ASK, FSK, and DPSK lightwave systems,” J. Lightwave Technol. 8(9), 1379–1386 (1990). [CrossRef]
5. J. Kreissl, V. Vercesi, U. Troppenz, T. Gaertner, W. Wenisch, and M. Schell, “Up to 40 Gb/s directly modulated laser operating at low driving current: buried-heterostructure passive feedback laser (BH-PFL),” IEEE Photon. Technol. Lett. 24(5), 362–364 (2012). [CrossRef]
6. J. Kreissl, U. Troppenz, W. Rehbein, T. Gaertner, P. Harde, and M. Radziunas, “40 Gbit/s directly modulated passive feedback laser with complex-coupled DFB section,” Proc. European Conference on Optical Communication, We8.1.4. (2007).
7. A. S. Karar, Y. Gao, K. P. Zhong, J. H. Ke, and J. C. Cartledge, “Generation of DPSK signals using a directly modulated passive feedback laser,” Proc. European Conference on Optical Communication, Tu.4.A.1 (2012).
9. H. Meyer, M. Moeneclaey, and S. A. Fechtel, Digital Communications Receivers ,(Wiley-Interscience, 1997).
10. M. Morelli and U. Mengali, “Feedforward frequency estimation for PSK: a tutorial review,” Eur. Trans. Telecommun. Relat. Technol. 9(2), 103–116 (1998). [CrossRef]
11. Y. Gao, A. P. T. Lau, S. Yan, and C. Lu, “Low-complexity and phase noise tolerant carrier phase estimation for dual-polarization 16-QAM systems,” Opt. Express 19(22), 21717–21729 (2011). [CrossRef] [PubMed]
12. C. Li, Z. Zhang, F. Zhu, and Y. S. Bai, “Method and apparatus for carrier phase estimation and correction in a coherent optical system,” US. Patent Application No. 0008952 A1 (2012).
13. A. S. Karar, M. Yañez, Y. Jiang, J. C. Cartledge, J. Harley, and K. Roberts, “Electronic dispersion pre-compensation for 10.709-Gb/s using a look-up table and a directly modulated laser,” Opt. Express 19(26), B81–B89 (2011). [CrossRef] [PubMed]