Survey
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
International Journal of Performability Engineering, Vol. 3, No. 4, October 2007, pp. 501-503. ©RAMS Consultants Printed in India On the Use of Gaussian Approximation for Reliable Performance Evaluation in Optical DPSK Systems Qun Zhang* and Han-Way Huang Electrical and Computer Engineering Department Minnesota State University, Mankato, MN 56001, USA (Received on August 7, 2007) Abstract – In this paper, we propose to extend the Guassian approximation (GA) method for reliable system performance evaluation from the traditional optical on-off keying (OOK) systems to the emerging optical differential phase shift keying (DPSK) systems. The proposed method can be used to guide efficient numerical estimate as well as experimental measurement of the noise-loading back-to-back DPSK system performance where the inter-symbolinterference (ISI) is not significant. Keywords: Bit-error-rate (BER) evaluation, optical fiber communication systems, differential phase-shift keying (DPSK) 1. Introduction Reliable back-to-back performance evaluation/measurement is crucial in optical transceiver design. Also, it is the starting point in the link budget design for optical communication systems [1]. In the traditional optical on-off keying (OOK) systems, a simple and yet reliable method, i.e. the Gaussian approximation (GA) method was extensively used for simulation based performance evaluation and experimental bit error rate (BER) measurement [2-3]. However, it has been shown that this method cannot be used directly for the next generation optical differential phase shift keying (DPSK) systems [4]. The reason is simple. For OOK systems, the pdf’s of spaces and marks have respectively, central and non-central χ 2 shapes. With GA and a decision threshold at which the two types of probability of error are equal, i.e., Pr(1|0)=Pr(0|1), the approximated BER could be within one order of magnitude compared to the exact BER, although the detection threshold may deviate significantly from the its true values. For DPSK systems, the pdf’s of spaces and marks have similar non-central χ 2 shapes. As a result, the decision threshold derived from the GA method could be highly accurate, while the approximated BER values could deviate significantly from the true values. In this paper, we propose to modify the GA method so that it can be used for the optical DPSK systems. The proposed method can guide efficient numerical performance estimate as well as experimental performance measurement for the noise-loading back-to-back DPSK system where the inter-symbol-interference (ISI) is not significant. 2. Proposed Method For a back-to-back DPSK system, we employ a recently developed Karhunen-Loéve series ______________________________________________________________________ * Corresponding author’s email: [email protected] 501 Q. Zhang and H-W Huang 502 expansion (KLSE) method [5] to accurately calculate the exact probability density function (pdf) of the detected photo voltage for the single shot case. The exact bit error rate (BER) and thus the exact Q-factor obtained by simply inverting the BER, can be obtained. We then approximate the exact pdf’s by the Gaussian pdf’s to calculate the approximate Q-factor. The difference between the exact and the approximate Q-factor values, i.e., the ∆Q can be obtained next and recorded. Now we investigate the pulse train case. For typical back-to-back systems that usually have weak ISI, we can apply the GA approximation to obtain the approximated Qfactor values. We then use the recorded ∆Q to correct the approximated Q-factor. We expect that at high optical signal-to-noise (OSNR) levels, the Q-factor values obtained by using the proposed method may exaggerate the noise effect due to the existence of ISI. However, for practical systems that usually can tolerate a higher noise power level, the proposed method should be able to give good estimate of system performance. We validate our proposed method using the following example. 3. Example The analyzed system model is illustrated in Fig. 1. The electrical 40 Gb/s information data passes through a differential encoder and a driving circuit, and drives the Mach-Zehnder modulator (MZM) to generate 50% RZ optical signals. The rise time of the trapezoidal NRZ shaped electrical driving signal is assumed to be ¼ of the bit interval. The low pass response of the driving circuits is assumed to be 5th order Bessel filter with a 32 GHz 3-dB bandwidth. The modulator model is based on a Fujitsu EO modulator, and the 3-dBO bandwidth is 33 GHz. The DPSK receiver consists of a 2nd order super-Gaussian shaped optical BPF with a 3dB passband bandwidth of 75 GHz, a delay line interferometer (DI), a balanced photodetector, a LPF the same as the one for the driving circuits, a sampler, and the decision circuit. To take into account the bit pattern effects, we use a 8 bit De Bruijn sequence that contains all 3-bit patterns each with exactly one occurrence. No receiver imperfection is considered. Fig. 1: System Setup - RZ-DPSK Transmitter and Optically Pre-amplified Receiver Using the proposed method, we first obtained the exact pdf’s (denoted as pdfexact) at OSNR = 10 dB for the transmitted 00 and 01 single shot signals, as shown in Fig. 2. Also shown is the approximated Gaussian pdf’s, denoted as pdfGA . We can clearly see that by using GA, the BER will be overestimated. Fig. 3 shows the single shot Q values calculated from respectively pdfexact and pdfGA, vs. OSNR. As expected, we have Qexact > Qexact. Fig. 3 also shows that ∆Q = Qexact − QGA is not a constant, but rather OSNR-dependent. Next we study the system performance using the 8-bit pulse train generated. Fig. 4 shows the system performance from the exact pdf and BER calculation, as well as that from the GA. Apparently BERGA is dramatically different from the BERexact, and thus is not reliable in indicating the system performance. However, BERGA-cor, which is obtained by correcting BERGA with the help of single shot ∆Q, matches BERexact very well at all studied OSNR levels. pdf Detected voltage (a.u.) Fig. 2: Single Shot pdf’s OSNR (dB) Fig. 3: Single Shot Q’s and ∆Q BER ∆Q (dB) On the Use of Gaussian Approximation for Reliable Performance Evaluation in Optical DPSK Systems Q (dB) 503 OSNR (dB) Fig. 4: System BER 4. Conclusions and Future Work We demonstrate a novel GA method for reliable performance estimation/measurement in back-to-back Optical DPSK Systems where the ISI may not be significant. For systems with strong ISI, we could refer the method in [6] and calculate ∆Q for all possible patterns that is long enough to account for system memory. This is left as a future work. References [1] Golovchenko, E. A., A. N., Pilipetskii, N. S. Bergano, C. R. Davidson, F. I. Khatri, R. M. Kimball, and V. J. Mazurczyk, Modeling of Transoceanic Fiber-Optic WDM Communication, IEEE Journal of Selected Topics in Quantum Electronics, Vol. 6, No. 2, pp. 337–347, 2000. [2] Humblet, P. A. and M. Azizoglu, On the bit error rate of lightwave systems with optical amplifiers, IEEE Journal of Lightwave Technology, vol. 9, no. 11, pp. 1576–1582, 1991. [3] Bergano, N. S., F. W. Kerfoot, and C. R. Davidson, Margin measurement in optical amplifier systems, IEEE Photonics Technology Letters, vol. 5, no. 3, pp. 304–306, 1993. [4] Kim, H. and P. J. Winzer, Nonlinear Phase Noise in Phase-Coded Transmission, Technical Digest OFC 2005, Vol. 4, paper OThO3, Anaheim, CA, USA. [5] Zhang, Q. and C. R. Menyuk, A Rigorous Optical DPSK Transceiver Model and BER Calculation Using the KLSE Method, Proceedings of 41st Conference on Information Sciences and Systems, paper TP4, Baltimore, MD, 2007. [6] Anderson, C. J. and J. A. Lyle, Technique for Evaluating System Performance using Q in Numerical Simulations Exhibiting Intersymbol Interference, IEE Electronics Letters, Vol. 30, pp. 71-72, 1994. Qun Zhang is currently an assistant professor at the ECE department of the Minnesota State University (MSU), Mankato, MN, USA. Prior to joining the MSU, he worked in research laboratories in PhotonEx Corporation, Corvis Corporation (now part of L-3 Communications), and Tyco Telecommunications. Dr. Zhang received his Ph. D. degree in electrical engineering from the University of Virginia, Charlottesville, VA, USA in 2001. Han-Way Huang is a professor at the ECE department of the Minnesota State University, Mankato, MN, USA. He received his Ph. D. degree in computer engineering from the Iowa State University, Ames, IA, USA in 1988.