Code-division Multiple-access Encoding And Decoding Of ...

IEEE PHOTONICS TECHNOLOGY LETTERS, VOL. 10, NO. 1, JANUARY 1998

171

Code-Division Multiple-Access Encoding and Decoding of Femtosecond Optical Pulses over a 2.5-km Fiber Link C.-C. Chang, H. P. Sardesai, Student Member, IEEE, and A. M. Weiner, Fellow, IEEE

Abstract—We report for the first time proof-of-concept transmission experiments for femtosecond pulse code division multiple access (CDMA) operating over kilometer lengths of optical fiber in the 1.5-m communication band. Our CDMA link consists of a femtosecond mode-locked Er-fiber laser and two chirped pulse fiber amplifiers, a pair of low-loss fiber-pigtailed pulse shapers for encoding and decoding, a 2.5-km dispersion-compensated transmission fiber, and an ultrafast nonlinear optical receiver. Our results demonstrate high fidelity tra nsmission of spectrally encoded femtosecond pulses over 2.5 km of fiber, followed by the use of the nonlinear CDMA receiver to discriminate between correctly and incorrectly decoded signals with 20-dB contrast. Index Terms— CDMA, coding, dispersion compensation, nonlinear thresholder, optical network, pulse shaping.

O

PTICAL code-division multiple-access (CDMA) communication systems offer an interesting approach for local area networks due to their unique attributes of optical processing, asynchronous transmission, potential for high information security, and the capability for multiple access [1]–[5]. Unlike time-division multiplexing and wavelengthdivision multiplexing (WDM) scheme, in CDMA signals are overlapped in both time and frequency; multiple-access is achieved by assigning each CDMA transmitter a different optical code which can be distinguished on the basis of a large intensity contrast between correctly and incorrectly decoded signals in the CDMA receivers. A femtosecond pulse CDMA scheme based on frequency-domain encoding and decoding of coherent ultrashort light pulses was previously proposed [2], [3], and such encoding/decoding was demonstrated at visible wavelengths using a pair of conjugate phase masks adjacent to each other [2]. An ultrashort pulse CDMA system of users can be operated in the following way [2], [3]: user stations are connected via a star coupler and a common fiber channel. Each user station is equipped with a CDMA transmitter including a femtosecond pulse source and an encoder which converts femtosecond pulses into lower intensity, picosecond pseudonoise bursts. Encoded signals are broadcast to the receivers of all other stations. Only the intended receiver with a matching code can decode the encoded signals back into short pulses. Incorrectly decoded Manuscript received July 21, 1997; revised September 10, 1997. This work was supported by the National Science Foundation under Grant ECS-9 312 256 and Grant ECS-9 626 967. The authors are with the School of Electrical and Computer Engineering, Purdue University, W. Lafayette, IN 47907 USA Publisher Item Identifier S 1041-1135(98)00458-3.

signals corresponding to the wrong code remain low intensity noise-like bursts which are eventually rejected by the optical thresholder. Here for the first time, we report successful experimental results on a 2.5-km femtosecond fiber-optic CDMA link operating in the 1.55- m communication band. Our experiments are devoted to the study of the key physical building blocks of a CDMA fiber network: a transmitter which can generate and encode femtosecond pulses, a fiber link that is dispersion compensated and thus allows for femtosecond pulse transmission, and a receiver which can distinguish between correctly and incorrectly decoded ultrafast optical signals. In the present work we demonstrate the first successful operation (with a contrast ratio as high as 20 dB between correctly and incorrectly decoded pulses) of a single femtosecond CDMA transmitter-receiver pair operating over kilometer lengths of fiber. Fig. 1 shows the configuration of our experiments, consisting basically of one transmitter-receiver pair connected with a 2.5-km fiber link. In the transmitter, a stretched-pulse modelocked Er-fiber ring laser [6] is used to generate 1.55m pulses as short as 65 fs at a rate of 30 MHz. To suit our CDMA experiments, these 65-fs pulses are spectrally tailored to 300-fs pulses by an interference bandpass filter. The filtered pulses are amplified to 1.2 mW by an erbiumdoped fiber amplifier (EDFA). The amplified pulses (with pulsewidth broadening to 400 fs due to gain narrowing effect in EDFA) are then passed through an encoder that is a fiber-pigtailed programmable pulse shaper containing a pair of gratings and lenses, and a liquid crystal phase modulator (LCM) in the Fourier (or mask) plane [10]. The total fiberto-fiber insertion loss of the pulse shaper is as low as 5.3 dB. The various frequency components of the pulses are spatially dispersed across the Fourier plane by the first grating and lens and recombined by the second ones. In the Fourier plane, a programmable LCM is placed to spectrally manipulate the pulses. The LCM (supplied by Cambridge Research and Instrumentation) consists of 128 phase-modulating pixels with 100- m center-to-center spacing and 3 m interpixel gaps. Each pixel can be independently programmed to provide any gray-scale phase modulation between zero and 2 . The pulse shaper is adjusted to be in zero-dispersion configuration such that when no phase function is applied to the LCM, the pulses retain their original shape and duration after passing through the pulse shaper. For our CDMA phase coding scheme, we apply -sequence pseudorandom binary phase (“ ” or “ ”)

1041–1135/98$10.00  1998 IEEE

172

Fig. 1. Block diagram for the femtosecond pulse CDMA experiments.

codes of length 31 or 63 (4 or 2 pixels per code element) to the LCM. As a result, the short pulses entering the pulse shaper are encoded into relatively wide ( 10 ps) pseudonoise bursts with reduced peak intensity. The coded pulses are then transmitted over a 2.5-km transmission fiber link that requires dispersion compensation to allow for broad-band pulse transmission. Fiber dispersion compensation is crucial for femtosecond pulse CDMA. If dispersion is too large, even successfully decoded pulses will be spread compared to the initial pulses; this will degrade the ability to discriminate between properly and improperly decoded signals. Here, we employ a linear compensation technique which uses a special compensating dispersion fiber (DCF) [10] to compensate the large anomalous dispersion in a standard 1.3- m single-mode fiber (SMF). From our previous studies the 2.5-km dispersion-compensated link with total loss of 3 dB consisting of 2060 m of SMF and 450 m of DCF has essentially zero second-order dispersion and a residual third-order dispersion 3–5 times lower than that of dispersionshifted fiber [7], [9]. The amplified pulses broaden from 400 fs to 480 fs (due to some polarization-dependent effect of the gratings in the pulse shaper) after passing through the encoder and before entering the fiber link. Fig. 2 shows the intensity cross-correlation traces of an uncoded pulse (i.e., a constant phase applied to all 128 pixels in the encoder) before (A) and after transmitting over (B) the 2.5-km SMF-DCF link. Although a small residual positive third-order dispersion still causes some pulse broadening (580 fs compared to 480 fs) and oscillation in the pulse tail, overall the fiber link is very well compensated (an equivalent length of SMF alone would broaden the input pulses from 480 fs to 350 ps). After propagating through the 2.5-km dispersioncompensated link, the encoded signals enter the decoder which is made identical to the encoder so that the same frequency components pass through corresponding pixels of the encoder and decoder LCM’s. The decoder is immediately followed by a second EDFA to boost the decoded signals for nonlinear thresholding. To properly decode the pulse, the LCM in the decoder is programmed with a phase code which is the conjugate of that applied to the encoder. Fig. 3(A) shows the intensity autocorrelation of the restored (properly decoded) femtosecond pulses measured after the second EDFA when a

IEEE PHOTONICS TECHNOLOGY LETTERS, VOL. 10, NO. 1, JANUARY 1998

Fig. 2. Input pulse to the 2.5-km fiber link (A) and output pulse from the fiber link. The small oscillation in the tail of output pulse in (B) indicates the dispersion-compensated fiber link has residual third-order dispersion.

Fig. 3. (A) Intensity autocorrelation at the decoder output for a correctly decoded pulse, (B) intensity autocorrelation at the decoder output for an incorrectly decoded pulse, (C) and (D) power spectrum after the thresholder fiber corresponding to (A) and (B) respectively. The -sequence phase code length is 63 elements in each case.

M

conjugate pair of 63-element -sequence phase codes were applied to the encoder and decoder respectively. If the pulse is improperly decoded (by a nonconjugate phase code in the decoder, as would be the case with an interfering user), the pulse remains as a pseudonoise burst as shown in Fig. 3(B). The peak intensity of the incorrectly decoded pulses is substantially less than that of the properly decoded pulses. Because correctly and incorrectly decoded pulses have comparable average power level and very short durations, ordinary photodetectors cannot be used to determine if the intended signal is received. Instead, a nonlinear thresholder is employed to distinguish correctly decoded pulses from incorrectly decoded pseudonoise bursts based on the pulse intensity and duration. Our nonlinear thresholder consists of a passive optical fiber, a spectral filter, and a photodetector [11]. The passive fiber used here is a 500-m dispersion-shifted fiber with zero dispersion wavelength of 1559 nm (which coincides

CHANG et al.: CODE-DIVISION MULTIPLE-ACCESS ENCODING AND DECODING

M

Fig. 4. Contrast ratios at the output of the nonlinear thresholder for length -sequencephase coder. The filter cut-off wavelength is 31 and length 63 1573 nm.

with the mean wavelength of the spectrum used in our CDMA system). Nonlinear effects (self-phase modulation) in the fiber cause the spectrum of the properly decoded pulse to split and spread [12] to either side of the zero dispersion point. The improperly decoded pulse propagates the same length of the thresholder fiber but experiences negligible spectral shifts due to its lower intensity and larger pulsewidth. Fig. 3(C) and (D) shows the power spectra at the output of the thresholder fiber corresponding respectively to the correctly decoded pulse and the incorrectly decoded signal. The clear differences revealed in these spectra are converted into a contrast in energy by a long wavelength pass filter followed by a slower photodetector which can operate at speeds comparable to the repetition rate of the system. Fig. 4 shows the contrast ratios between the correctly and incorrectly decoded signals for various signal power in the nonlinear thresholder fiber. Contrast ratios as high as 15 dB and 20 dB are demonstrated for 63-element and -sequences respectively when the filter cutoff 31-element wavelength is optimally chosen at 1573 nm. The nonlinearity in the thresholder can lead to enhanced contrast ratio compared to that obtained directly from the decoding. Contrary to what should be expected from matched filter operation, our 31-element codes here gives higher contrast ratios than 63element codes. The main reason for this is that LCM’s used here have higher loss when longer codes are applied and therefore the properly decoded signals for longer codes have lower intensity and thus less frequency shift in the thresholder. In the current experiments the contrast ratio is limited by the amplified spontaneous emission (ASE) from the second EDFA that spreads somewhat into the wavelength range over 1573 nm. This limitation should be alleviated by adding a fiber Bragg grating after the second EDFA to filter out the long wavelength ASE spectral components. Even with this limitation, however, we are already able to distinguish between correctly and incorrectly decoded pulses with a contrast ratio

173

of 100. For a CDMA system of users, the interference noise level should increase by times. Therefore, the contrast ratio we obtained from our single CDMA transmitter/receiver experiments provide a baseline estimate for the signal-to-noise ratio (SNR) performance of multiple-access systems. In conclusion, we have reported for the first time highfidelity transmission of spectrally encoded femtosecond pulses over kilometer lengths of fiber. We have also demonstrated the use of a nonlinear CDMA receiver to discriminate between desired and undesired waveforms with high contrast. We expect to use the current setup to implement a femtosecond CDMA testbed to characterize system performance with multiple users. We also note that the nonlinear processing inherent in our short pulse setup may allow us to extend our studies to other novel CDMA configurations, such as hybrid CDMAWDM networks in which separate CDMA and WDM traffic is overlaid within the same spectral range. ACKNOWLEDGMENT The authors thank A. M. Vengsarkar, M. Newhouse, and V. Da Silva for supplying SMF and DCF, Er-fiber, and dispersion-shifted fibers, respectively. They also thank A. W. R. Emanuel and S. Shen for calibrating the LCM’s. REFERENCES [1] P. R. Prucnal, M. A. Santoro, and T. R. Fan, “Spread spectrum fiber-optic local area network using optical processing,” J. Lightwave Technol., vol. LT-4, p. 547, 1986. [2] A. M. Weiner, J. P. Heritage, and J. A. Salehi, “Encoding and decoding of femtosecond pulses,” Opt. Lett., vol. 13, pp. 300–302, 1988. [3] J. A. Salehi, A. M. Weiner, and J. P. Heritage, “Coherent ultrashort light pulse code-division multiple access communication systems,” J. Lightwave Technol., vol. 8, pp. 478–491, 1990. [4] D. Zaccarin and M. Kavehrad, “An optical CDMA system based on spectral encoding of LED,” IEEE Photon. Technol. Lett., vol. 5, pp. 479–482, 1993. [5] R. A. Griffin, D. D. Sampson, and D. A. Jackson, “Coherence coding for code-division multiple-access networks,” J. Lightwave Technol., vol. 13, pp. 1826–1837, 1995. [6] K. Tamura, E. P. Ippen, H. A. Haus, and L. E. Nelson, “77-fs pulse generation from a stretched-pulse mode-locked all-fiber ring laser,” Opt. Lett., vol. 18, pp. 1080–1082, 1993. [7] C.-C. Chang, A. M. Weiner, A. M. Vengsarkar, and D. W. Peckham, “Broadband fiber dispersion compensation for sub-100-fs pulses with a compression ratio of 300,” Opt. Lett., vol. 21, pp. 1141–1143, 1996. [8] A. M. Weiner, D. E. Leaird, J. S. Patel, and J. R. Wullert, II, “Programmable shaping of femtosecond optical pulses by use of 128-element liquid crystal phase modulator,” IEEE J. Quantum Electron., vol. 28, pp. 908–920, 1992. [9] C.-C. Chang and A. M. Weiner, “Fiber transmission for sub-500femtosecond pulses using a dispersion-compensating fiber,” IEEE J. Quantum Electron., vol. 33, pp. 1455–1464 1997. [10] A. M. Vengsarkar, A. E. Miller, M. Haner, A. H. Gnauck, W. A. Reed, and K. L. Walker, “Fundamental-mode dispersion-compensating fibers: Design consideration and experiments,” in OFC Tech. Dig., 1994, pp. 225–227, paper Thk2. [11] H. P. Sardesai and A. M. Weiner, “A nonlinear fiber-optic receiver for ultrashort pulse code division multiple access communications,” Electron. Lett., vol. 7, pp. 610–611, 1997. [12] G. P. Agrawal and M. J. Potasek, “Nonlinear pulse distortion in singlemode optical fibers at zero-dispersion wavelength,” Phys. Rev. A., vol. 33, pp. 1765–1766, 1986.