We report the attempt of optical quantization and coding in 5-bit parallel format for photonic A/D conversion. The proposed system is designed to realize generation of 32 different optical codes in proportion to the corresponding signal levels when fed a certain range of amplitude-varied input pulses to the setup. Optical coding in a bit-parallel format made it possible, that provides 5bit optical codes from 32 optical quantized pulses. The 5-bit parallel operation of an optical quantization and coding module with 5 multi-ports was tested in our experimental setup.
© 2011 OSA
Various optical approaches have been attracting interest as breakthrough toward the future technology including photonic networks . As recent tremendous growths of the ultrawide-bandwidth applications, Analog-to-Digital (A/D) conversion has been investigated as a key interface technology in any electronic or photonic systems . A/D conversion is basically composed of three steps; sampling, quantization, and coding. Its performance is evaluated by rate of sampling and resolution of quantization and coding. While remarkable progress of electrical circuit technology realizes the high-resolution A/D conversion whose resolution is eight or less bits at gigasamples per second (GS/s), there is an unavoidable trade off between the sampling rate and the resolution bit due to electrical jitter of the sampling aperture and ambiguity of the comparator .
Optical approaches to A/D conversion is expected to provide many performance advantages over their electrical counterparts [3–5]. A ultrafast optical sampling technique has come to achieve a sampling rate over 100(GS/s) [6–8] and such a technique would solve the problem of timing jitter in sampling. Therefore, it makes optical replacement in quantization and coding all the more valuable to complete a high resolution A/D conversion with keeping the high throughput advantage in optical sampling [9–18]. We have also proposed optical quantization using soliton self-frequency shift (SSFS) in a fiber  and the several coding schemes [15–17] that could follow the proposed optical quantization. While there is a report for 5bit all-optical conversion, it is limited to demonstration for a single output port and the transfer functions are not experimental results but calculated ones . In this work, 5-bit optical quantization and coding in bit parallel format are experimentally demonstrated and the 5 transfer functions are experimentally obtained. Concerning potential of possible sampling rate, bit-error-rate assessment of one of 32 quantized levels in simulation shows an error free operation for a 10 Gbps 211-1 PRBS data signal. In addition, comparison of degradation of signal quality after quantization and coding shows almost degradation-free operation for less than 20ps pulses.
2. Principle of Optical Quantization and Coding
Since a sampling rate over 100(GS/s) has been achieved by optical sampling technique [6–8], the proposed subsystem is focused on optical quantization and optical coding after optical sampling process. The proposed schematic diagram for optical quantization and optical coding is shown in Fig. 1.
Our optical quantization method is equivalent to intensity-to-wavelength conversion based on the soliton self-frequency shift (SSFS) . Since the SSFS in a fiber can generate a different color signal proportional to an input peak power, an input peak power can be identified by an output signal color from a fiber. Once each input peak power is identified by each color, a dispersion devise can separately split each different colored signal into a different port as a level identification signal. The SPM-based spectral compression (SSC) in a fiber can improve the resolution of the color separation because it enables to compress the bandwidth of an input pulse . To upgrade the number of bit of optical quantization from 4bit  to 5bit, we need to improve a spectral compression ratio more than twice as good as that for 4bit optical quantization and we introduce a multi-stage approach of SSC here. Optical coding can be realized by vertically-stacked pulse shapers . The level identification signal is fed to the vertically-stacked pulse shapers and each pulse shaper generates a binary pulse signal for each bit. The vertically-stacked pulse shapers can be constructed by plural wavelength-division multiplexing (WDM) devices such as arrayed waveguide gratings (AWGs) as well as bulk pulse shapers.
We experimentally demonstrated 5-bit parallel operation of an optical quantization and coding module with 5 multi-ports. The experimental setup is shown in Fig. 2. We used an optical pulse irradiated from a fiber laser (IMRA America Inc.) as a light source. The pulse width, the center wavelength, and the repetition rate were 500 fs, 1559 nm, and 50 MHz, respectively. Here, we used a relatively low repetitive frequency laser to provide a sufficient peak power for generation of SSFS because this scheme has the inherent feature of sampling rate transparency. We prepared input pulses with pseudo-continuously varied peak power as a substitute of an input analog signal. To prepare sampled analog pulses, we adjusted a power of an optical pulse using a variable optical attenuator. We varied the peak power of optical pulses at least 4 steps per one level of quantization from 27.2 W to 44.0 W at the regular interval 0.139 W.
As described the design principle in Section 2, we use SSFS for optical quantization and a multi-stage SSC for twice improvement of the resolution of optical quantization. Input sampled analog pulses are propagated in a high-nonlinear fiber (HNLF) for the generation of SSFS. In this setup, 1km-HNLF (dispersion:D=7(ps/nm/km), dispersion slope:S=0.03(ps/nm 2/km), nonlinearity:γ=16(/W/km)) is used for SSFS generation. SSFS changes the center wavelength of each sampled analog pulse to a longer wavelength side proportional to the input peak power of each sampled analog pulse. The colored sampled analog pulses are propagated in a set of single mode fiber (SMF) and highnonlinear fiber (HNLF) for SPM-based spectral compression (SSC). In this setup, the SSC part is composed of 4 SMFs and 4 HNLFs and they are connected in the following order; 2m-SMF (dispersion:D=17(ps/nm/km), dispersion slope:S=0.06(ps/nm 2/km), nonlinearity:γ=1.5(/W/km) ), 92m-HNLF (dispersion:D=−0.00185(ps/nm/km), dispersion slope:S=0.0029(ps/nm 2/km), nonlinearity:γ=15(/W/km)), 10m-SMF (dispersion:D=17(ps/nm/km), dispersion slope:S=0.06(ps/nm 2/km), nonlinearity:γ=1.5(/W/km)), 110.5m-HNLF (dispersion:D=−0.266(ps/nm/km), dispersion slope:S=0.0268(ps/nm 2/km), nonlinearity:γ=9.3(/W/km)), 30m-SMF (dispersion:D=17(ps/nm/km), dispersion slope:S=0.06(ps/nm 2/km), nonlinearity:γ=1.5/W/km)), 400m-HNLF (dispersion:D=0.044(ps/nm/km), dispersion slope:S=0.029(ps/nm 2/km), nonlinearity:γ=20(/W/km)), 100m-SMF (dispersion:D=17(ps/nm/km), dispersion slope:S=0.06(ps/nm 2/km), nonlinearity:γ=1.5(/W/km)), and 116m-HNLF (dispersion:D=-0.325(ps/nm/km), dispersion slope:S=0.031(ps/nm 2/km), nonlinearity:γ=6.2(/W/km)). SSC compresses the bandwidth of each colored sampled analog pulse for resolution improvement.
After SSC, each colored sampled analog pulse is vertically expanded by a set of cylindrical lenses CL1(f=50mm) and CL2(f=125mm) and it is evenly fed to vertically-stacked pulse shapers. The vertically-stacked pulse shapers are composed of 5 pulse shapers which have own different frequency filters. These 5 different frequency filters make up the spatial mask M for coding. A set of cylindrical lenses CL3(f=50mm) and CL4(f=150mm), a set of spherical lenses L1(f=200mm) and L2(f=200mm), and a set of cylindrical lenses CL5(f=100mm) and CL6(f=100mm) are used for additional vertical expansion, spatial frequency filtering, and relay lenses, respectively.
The first grating G1(groove frequency of 600 lp/mm) separately splits each colored sampled analog pulse into the corresponding direction. After passing through CL3, L1, and CL4, each colored sampled analog pulse is led to the corresponding spatial position on the spatial mask M as a level identification signal. Each level identification signal has vertically expanded line shape and it is aligned along with the horizontal axis proportional to an input peak power.
The spatial mask M filters level identification signals so as to provide appropriate binary pulse signals according to the binary conversion table (Table 1). The spatial filter in each pulse shaper is designed according to the binary conversion table for 5-bit A/D conversion. It allows us to broadcast a level identification signal to each spatial port so as to provide multiple-bit binary code in a bit-parallel format. Figure 3 shows a set of spatial filters (;the spatial mask M) used in this scheme. Consequently, it can be regarded as that in the previous approach by Tsunoda and Goodman . After passing through CL5, L2, CL6, and G2(groove frequency of 600 lp/mm), each filterd level identification signal is emitted as multiple-bit binary codes in a bit-parallel format. Here, each pulse shaper takes a role of generation of each bit signal.
4. Experimental Results
The output spectrums of colored level identification signals were measured by an optical spectrum analyzer (AQ8403; Yokogawa Electric Corporation) to confirm operation of optical quantization. Experimental results of optical quantization are shown in Fig. 4. Figure 4(a) indicates that the amount of center frequency shift is lineally increasing with each input peak power. From Fig. 4(b), we can confirm that 32 independent colored level identification signals are successfully provided which enable to achieve 5bit optical quantization. The multiple-bit binary codes in a bit-parallel format from vertically-stacked pulse shapers were observed by an InGaAs-Camera (C10633-13; Hamamatsu photonics K. K.). Experimental results of the 32 output multiple-bit binary codes in a bit-parallel format are shown in Fig. 5.
Output ports b1-b5 are numbered beginning at the top of each captured image. From Fig. 5(a), we can confirm that each level identification signal is successfully encoded as the corresponding multiple-bit binary code in a bit-parallel format. Since, however, a Gaussian beam is fed to vertically-stacked pulse shapers, each obtained output multiple-bit binary codes has a nonuniform output intensity. Especially, the 1st and the 5th bit outputs are seriously affected by this influence because they use the edge of a Gaussian beam. Figure 5(b) shows the transfer function of each output port b1-b5. From Fig. 5(b), we can successfully confirm the 8 periods transfer function at the first bit port b1. Since, however, gratings G1 and G2 have a certain amount of polarization dependency, each output intensity of an output multiple-bit binary codes fluctuates depending on the polarization of an input pulse. From these results, we have successfully demonstrated 5-bit parallel operation of optical quantization and coding module with 5 multi-ports for photonic A/D conversion.
In this experiment, an InGaAs-Camera is used for observation of outputs from 5 ports because it was difficult to prepare 5 high-speed PDs and simultaneously measure outputs from 5 ports. Instead, we focused on a certain level signals (the output center wavelength;1593nm) and measured temporal profiles of signals before and after coding by using a photodetector and an oscilloscope ( 86100A; Agilent Technologies Ltd.) to confirm signal quality after quantization and coding. To measure the typical encoded signal, we coupled a spatial beam signal at the third port b3 to a fiber connected to a photodetector via an EDFA. Figure 6(a) and 6(b) show typical temporal profiles of signals after quantization (equivalent to before coding) and coding. Comparison of degradation of signal quality in Fig. 6(b) with that in Fig. 6(a) confirm sufficient signal quality can be kept in 5bit cascade operation of optical quantization and coding for photonic A/D conversion. From Fig. 6(a) and 6(b), pulse widths of both signals can keep less than 20 ps and it indicates that this scheme would have a potential for at least 40 GS/s cascade oepration of optical quantization and coding. While the needed average power for generation of SSFS at 40 GS/s is estimated to be around several hundreds mW, development of SSFS geneartion devices would be expected to solve this power consumption issue.
In order to assess the system ability, we achieve simluation of bit-error-rate (BER) measurements for a certain level signals of the back-to-back and the third port b3 (the output center wavelength;1593nm) after optical quantization and coding. The assessments were performed in the RSoft’s OptSim environment and the simulation setup consists of 10GHz mode-locked pulse laser, a modulator driven by a pulse pattern generator (PPG), a PPG generating a 211-1 pseudo-random bit sequence (PRBS), EDFA, our system for optical quantization and coding, an AWG, a receiver, and a BER tester (BERT). In this assessment, we used the same condition of the experimental system except the sampling rate. Transmission simulation of our system for optical quantization and coding was carried out under a transmission rate of 10 Gbps and a PRBS of 211-1 word length. Figure 7 shows BER characteristics under the same conditions. An error-free operation (BER< 10−9) with power penalty of 2.5 dB after transmission was achieved in simulation.
Concerning the resolution performance of this experimental system, we estimated the integral nonlinearity error(INL), the differential nonlinearity error(DNL), and the effective number of bits resolution (ENOB) which is associated with signal-to-noise ratio (SNR). The INL, the DNL, and ENOB are derived from the experimentally obtained transfer function as shown in Fig. 8. In this estimation, we adopted to the metrics as in the previous work.
In the estimation of the ENOB, we assume the oversampling condition which the input bandwidth is less than half of the sampling rate. A uniformly distributed probality density function is used as the input analog signal [3, 18]. The ENOB under the condition is estimated as,
We experimentally demonstrated 5-bit optical quantization and coding in bit parallel format. Experimental results showed that a certain range of amplitude-varied input pulses can be converted into 5-bit parallel binary codes corresponding to each input peak power. The results of experiments indicate that the proposed photonic ADC has the potential of both high sampling rate and high resolution applications. Since the proposed coding system has a highly-flexibile feature, we can promptly demonstrate other binary coding scheme by preparing an appropriate spatial mask in vertically-stacked pulse shapers.
References and links
1. Y. Liu, M. T. Hill, N. Calabretta, E. Tangdiongga, R. Geldenhuys, S. Zhang, Z. Li, H. De Waardt, G. D. Khoe, and H. J. S. Dorren, “All-optical signal processing for optical packet switching networks,” Proc. SPIE 5907, 59070J (2005). [CrossRef]
2. R. H. Walden, “Analog-to-digital converter survey and analysis,” IEEE J. Sel. Areas Commun. 17, 539–550 (1999). [CrossRef]
3. B. L. Shoop, Photonic Analog-to-Digital Conversion (Springer-Verlag, 2001).
5. H. F. Taylor, “An electro-optic analog-to-digital converter,” Proc. IEEE 63, 1524–1525 (1975). [CrossRef]
6. C. S. Langhorst and H.-G. Weber, “Optical sampling techniques,” J. Opt. Fiber. Commun. Rep. 2, 86–114 (2005). [CrossRef]
7. T. Kawanishi, T. Sakamoto, and M. Izutsu, “470GHz optical clock signal generation using reciprocating optical modulation,” ECOC2006, We4.6 (Canne, 2009).
8. H. Ji, M. Pu, H. Hu, M. Galili, L. K. Oxenlowe, K. Yvind, J. M. Hvam, and P. Jeppesen, “Optical waveform sampling and error-free demultiplexing of 1.28 Tb/s serial data in a nanoengineered silicon waveguide,” J. Lightwave Technol. 29, 426–431 (2011). [CrossRef]
9. J. -M. Jeong and M. E. Marhic, “All-optical analog-to-digital and digital-to-analog conversion implemented by a nonlinear fiber interferometer,” Opt. Commun. 91, 115–122 (1992). [CrossRef]
10. P. P. Ho, Q. Z. Wang, J. Chen, Q. D. Liu, and R. R. Alfano, “Ultrafast optical pulse digitization with unary spectrally encoded cross-phase modulation,” Appl. Opt. 36, 3425–3429 (1997). [CrossRef] [PubMed]
11. T. Konishi, K. Tanimura, K. Asano, Y. Oshita, and Y. Ichioka, “All-optical analog-to-digital converter by use of self-frequency shifting in fiber and a pulse-shaping technique,” J. Opt. Soc. Am. B 19, 2817–2823 (2002). [CrossRef]
12. S. Oda, A. Maruta, and K. Kitayama, “All-optical quantization scheme based on fiber nonlinearity,” IEEE Photon. Technol. Lett. 16, 587–589 (2004). [CrossRef]
13. K. Ikeda, J. M. Abdul, H. Tobioka, T. Inoue, S. Namiki, and K. Kitayama, “Design considerations of all-optical A/D conversion: Nonlinear fiber-optic sagnac-loop interferometer-based optical quantizing and coding,” J. Light-wave Technol. 24, 2618–2628 (2006). [CrossRef]
14. J. Stigwall and S. Galt, “Demonstration and analysis of a 40 gigasample/s interferometric analog-to-digital converter,” J. Lightwave Technol. 24, 1247–1256 (2004). [CrossRef]
15. T. Nishitani, T. Konishi, and K. Itoh, “Optical coding scheme using optical interconnection for high sampling rate and high resolution photonic analog-to-digital conversion,” Opt. Express 15, 15812–15817 (2007). [CrossRef] [PubMed]
17. T. Nishitani, T. Konishi, and K. Itoh, “Resolution improvement of all-optical analog-to-digital conversion employing self-frequency shift and self-phase-modulation-induced spectral compression,” IEEE J. Sel. Top. Quantum Electron. 14, 724–732 (2008). [CrossRef]
18. Y. Miyoshi, S. Takagi, S. Namiki, and K. Kitayama, “Multiperiod PM-NOLM With dynamic counter-propagating effects compensation for 5-Bit all-optical analog-to-digital conversion and its performance evaluations,” J. Light-wave Technol. 28, 415–422 (2010). [CrossRef]
19. B. R. Washburn, J. A. Buck, and S. E. Ralph, “Transform-limited spectral compression due to self-phasemodulation in fibers,” Opt. Lett. 25, 445–447 (2000). [CrossRef]
20. A. M. Weiner, “Femtosecond pulse shaping using spatial light modulators,” Rev. Sci. Instrum. 71, 1929–1960 (2000). [CrossRef]