JOURNAL OF LIGHTWAVE TECHNOLOGY, VOL. 20, NO.

6, JUNE 2002

921

Accurate Modeling of Optical Multiplexer/Demultiplexer Concatenation in Transparent Multiwavelength Optical Networks
I. Roudas, Member, IEEE, N. Antoniades, Member, IEEE, T. Otani, T. E. Stern, Life Fellow, IEEE, R. E. Wagner, Member, IEEE, and D. Q. Chowdhury, Member, IEEE

Abstract—This paper presents an accurate theoretical model for the study of concatenation of optical multiplexers/demultiplexers (MUXs/DMUXs) in transparent multiwavelength optical networks. The model is based on a semianalytical technique for the evaluation of the error probability of the network topology. The error-probability evaluation takes into account arbitrary pulse shapes, arbitrary optical MUX/DMUX, and electronic low-pass filter transfer functions, and non-Gaussian photocurrent statistics at the output of the direct-detection receiver. To illustrate the model, the cascadability of arrayed waveguide grating (AWG) routers in a transparent network element chain is studied. The performance of the actual network is compared to the performance of a reference network with ideal optical MUXs/DMUXs. The optical power penalty at an error probability of 10 9 is calculated as a function of the number of cascaded AWG routers, the bandwidth of AWG routers, and the laser carrier frequency offset from the channel’s nominal frequency. Index Terms—Error analysis, optical filters, optical receivers.

I. INTRODUCTION N TRANSPARENT multiwavelength optical networks, each lightwave signal may be optically multiplexed/demultiplexed several times during propagation from its source to its destination [1]. Optical multiplexers/demultiplexers (MUXs/DMUXs) exhibit nonideal amplitude and phase transfer functions within the optical signal band. That is, their amplitude transfer functions might present passband curvature, tilt, and ripple. In addition, their phase transfer functions might not vary linearly with frequency. These impairments are enhanced when a large number of these devices are cascaded together. Consequently, optical MUX/DMUX concatenation

I

Manuscript received May 4, 2001; revised February 21, 2002. This work was supported by the Defense Advanced Research Projects Agency under MDA 972-95-3-0027 and performed as a part of the MONET consortium. I. Roudas was with the Telcordia Technologies, Red Bank, NJ 07701 USA. He is now with Corning Inc., Somerset, NJ 08873 USA. N. Antoniades was with the with the Department of Electrical Engineering, Columbia University, New York, NY 10027 USA. He is now with Corning Inc., Somerset, NJ 08873 USA. T. Otani was with the Department of Electrical Engineering, Columbia University, New York, NY 10027 USA. He is now with the KDD Research and Development Laboratories, Saitama 356-8502, Japan. T. E. Stern is with the Department of Electrical Engineering, Columbia University, New York, NY 10027 USA. R. E. Wagner was with the Telcordia Technologies, Red Bank, NJ 07701 USA. He is now with Corning Inc., Corning, NY 14831 USA. D. Q. Chowdhury is with Corning Inc., Corning, NY 14831 USA. Publisher Item Identifier S 0733-8724(02)05399-9.

causes signal attenuation and distortion, i.e., intersymbol interference (ISI), and eventually limits the maximum number of optical network elements that can be cascaded. In the past, the cascadability of different types of optical MUXs/DMUXs was studied both theoretically [2]–[18] and experimentally [11], [18]–[24] for several bit rates, channel spacings, modulation formats, and other system considerations. Most attention was concentrated on arrayed waveguide grating (AWG) routers [6], [11], [13], [17], [19]–[22], [24], MUXs/DMUXs composed of multilayer interference (MI) filters [4], [5], [8], [10], [14], [17], [18], and MUXs/DMUXs composed of fiber Bragg grating filters [14], [15], [23]. General studies, not bound to a specific optical MUX/DMUX type, but based on arbitrary transfer functions, also exist, e.g., [7] and [16]. Previous theoretical studies contain approximations concerning the network topology and the network performance evaluation. More specifically, [4]–[6], [9], [10], and [12]–[18] study isolated chains of optical MUXs/DMUXs instead of realistic network topologies. In the latter case, the signal spectral clipping due to the concatenation of optical MUXs/DMUXs introduces an excess loss which changes the operating point of optical servo-controlled attenuators and erbium-doped fiber amplifiers (EDFAs) in the network elements. Therefore, it is necessary to study the interaction between optical MUXs/DMUXs, optical servo-controlled attenuators and EDFAs, and its impact on the power levels of the optical signal and amplified spontaneous emission (ASE) noise in the network. In addition, little effort was devoted to evaluate accurately the network performance degradation due to the concatenation of optical MUXs/DMUXs. References [2], [4], [5], [7], [9], [10], [12]–[14], [16], and [17] compute the eye pattern at the output of the direct-detection receiver by noiseless simulation and use the eye opening as a qualitative criterion of the network performance. Other authors [3], [6], [8], [11], [18] include ASE noise in their calculation with various degrees of accuracy, but assume Gaussian receiver noise statistics at the output of the direct-detection receiver. This assumption is strictly valid only for thermal-noise-limited receivers, but not for the most common case of optically preamplified direct-detection receivers. This paper presents an accurate theoretical model, free of both aforementioned limitations, for the study of the concatenation of optical MUXs/DMUXs in transparent multiwavelength optical networks. The key features of the proposed model are

0733-8724/02$17.00 © 2002 IEEE

[38]. 1(b) shows an example architecture of a reconfigurable WADM proposed in the multiwavelength optical networking (MONET) project [47]. [46]. it is possible in principle to maximize the network performance for a given number of concatenated AWG routers by jointly optimizing the optical bandwidth of the individual AWG routers and the electronic bandwidth of the receiver. As the number of cascaded AWG routers continues to increase. arbitrary optical MUXs/DMUXs and electronic low-pass filter (LPF) transfer functions. For the error-probability evaluation. It consists of a equidistant network elements separated by chain of fiber spans of loss . 6. [45].922 JOURNAL OF LIGHTWAVE TECHNOLOGY. and nonGaussian photocurrent statistics at the output of the direct-detection receiver. and the laser carrier frequency offset from the channel’s nominal frequency. THEORETICAL MODEL This section is divided into five parts. respectively. at nominal carrier frequency . To illustrate the model. 1(b). We begin our discussion with a description of the network topology. by increasing the equivalent noise bandwidth of individual AWG routers for a given bit rate at the expense of channel spacing. the concatenation of AWG routers in a transparent network element chain is studied. Network Representation The network topology is shown in Fig. The fourth topic of this section is the description of the semianalytical method for the evaluation of the average error probability. and asymptotic evaluation of the probability density function (pdf) at the output of the direct-detection receiver from the characteristic function using the method of steepest descent [26]. 20. including expansion of the ASE noise impingent upon the photodiode in Karhunen–Loève series [42]–[44]. an optical switch fabric for signal adding/dropping. In addition. These results are compared to the performance of a reference network composed of optical MUXs/DMUXs with rectangular amplitude response and zero phase response. The third topic of this section is the evaluation of the statistics of the photocurrent at the output of the direct-detection receiver. It can be shown that in the WSXC the optical signal passes from the same number of optical components as shown in Fig. 1(a). 1(b). The second topic of this section is the mathematical description for the signal and ASE noise propagation through their equivalent channels and the derivation of an analytical expression for the photocurrent at the output of the direct-detection receiver. This treatment highlights the underlying mechanisms for performance degradation and provides a rigorous definition for the excess loss resulting from the optical MUXs/DMUXs. VOL. a reference network is defined and is used for comparison with the actual network. are not parts of the optical transport layer [1]. 4) Finally. [41]. Section III presents a study case for the concatenation of AWG routers in a chain of transparent network elements. 1(c) shows the structure of an optically preamplified direct-detection receiver. 2) In the case that the equivalent noise bandwidth of the individual AWG routers is much larger than the bit rate. An example architecture of an eight wavelength 4 4 WSXC can be found in [48]. the optical bandwidth of AWG routers. In addition. when the number of cascaded AWG routers increases. For illustrative purposes and without loss of generality. an optical MUX/DMUX pair. is added at the first network element and propagates through the entire chain (dotted line). despite the inherent passband curvature of these devices. since it is the elementary building block of more complex network topologies. a chain of optical network elements is studied. An exhaustive study of network element architectures is beyond the scope of this paper. 1) It is possible to concatenate an arbitrary number of conventional AWG routers. A. the network performance can be maximized by misaligning the laser carrier frequency from the optical MUXs/DMUXs center frequency. One optical signal. and servo-controlled attenuators for power equalization. Fig. diagonalization of a quadratic form [37]. the effective equivalent noise bandwidth of the AWG router cascade becomes comparable to the bit rate and the network performance slowly degrades due to the increase of ISI. NO. for narrow optical MUX/DMUX bandwidths. The EDFA at the input of the th network element is playing the role of the receiver optical preamplifier and is omitted in Fig. It consists of two EDFAs. Section II presents the theoretical model for the study of concatenation of optical MUXs/DMUXs in transparent multiwavelength optical networks. All fibers carry wavelength division multiplexed (WDM) optical signals. II. Fig. The optical transmitter and receiver. so they are not included in Fig. The following conclusions are drawn. The rest of the paper is organized as follows. the model combines a semianalytic method [25] with an accurate statistical analysis of square-law detection [26]–[41]. JUNE 2002 the following: 1) it is based on a transparent chain of wavelength add-drop multiplexers (WADMs) or wavelength selective cross-connects (WSXCs) and 2) it uses the error probability as a criterion for the cascadability of optical MUXs/DMUXs in the network. a tunable optical . originating from a decrease in the insertion loss of servo-controlled optical attenuators. 3) Using the current model. The computation involves several steps. 1(c). numerical solution of the associated homogeneous Fredholm equation of the second kind [44]–[46]. The details of the calculations are given in Appendixes A–C. The error probability is calculated as a function of the number of AWG routers. The first topic that is treated is the representation of an optical network topology by two separate equivalent channels for the signal and the ASE noise. The error-probability evaluation takes into account arbitrary pulse shapes. initially the network performance is improved because the optical bandwidth narrowing due to AWG router concatenation reduces the power of the ASE noise at the direct-detection receiver without essentially distorting the signal. The remainder of the receiver consists of a polarizer (optional). The fifth topic of this section is the derivation of an analytical expression for the error probability of the reference network. we introduce the notion of the excess noise factor. notionally.

All intranode optical components and fibers from the initial Fig. which compensate for the exMUXs/DMUXs.g.  : signal’s nominal carrier wavelength). a photodiode. The receiver tunable and denoted by is also included. if the optical MUXs/DMUXs are replaced by optical attenuators with the same insertion loss as the optical MUXs/DMUXs exhibit at their center frequency. 2(b) shows the final block diagram of the network topology and direct-detection receiver. but not cess loss caused by the first th DMUX at the drop for the excess loss resulting from the site. This effect is described by an excess noise factor. The generalization to the case when the signal attenuation induced by the optical fibers is not compensated fully by the EDFAs at the input of the network elements (e. which cause spectral broadening (e. 2(a) MUXs/DMUXs and reveals that the signal passes through servo-controlled attenuators. is is sampled at times the bit period. LPF: electronic low-pass filter with transfer function denoted by H (f ). (c) Block diagram of an optically preamplified direct-detection receiver (Pol: Polarizer. where . DMUX: demultiplexer. VOA: servo-controlled attenuator). For simplicity. 1This paper focuses exclusively on the impact of optical MUX/DMUX concatenation on the network performance.  : sampling instant). equal to the excess loss caused with gain . The optical BPF with transfer function servo-controlled attenuators are represented by gain elements . EDFAs is amplified by The ASE noise generated by the servo-controlled the effective gain provided by the last MUXs/DMUXs. 2(a). 1. The electric fields at the output of the transmitter and the input of the receiver are denoted by and . The interplay between bandwidth narrowing due to optical MUX/DMUX concatenation and fiber effects.. Dotted line indicates the optical path of the signal under study (NE: network element. . and the bit that minimizes the error probability. self-phase modulation). is not considered here and will be addressed in future work. 1(a) and (b) is eliminated. The excess loss is compensated by adjusting the insertion loss of the servo-controlled attenuator following the MUX/DMUX pair in order to keep the average signal power at the input of the booster EDFA constant. and a decision device. respectively. Consequently. it is assumed that the excess loss per node is within the operating range of the servo-controlled attenuators..ROUDAS et al. an electronic LPF with transfer function denoted . which is introduced for the first time here and will be defined below. (b) Architecture of a WADM (EDFA: Erbium-doped fiber amplifier. in the case of WDM transmission and EDFA gain ripple) is straightforward. we consider the case of single channel transmission. Any additional signal attenuation caused by signal filtering by the optical MUXs/DMUXs is defined as excess loss [4]. bandpass filter (BPF) with transfer function denoted by . MUX: multiplexer. by the optical MUX/DMUX pair of the th node. Without loss of generality. Inspection of the simplified block diagram shown in Fig.1 Optical MUXs/DMUXs are represented by filters with transfer functions normalized to unity . Signal attenuation within the network elements arises from the insertion loss of the optical components and the band-limiting operation of the optical MUXs/DMUXs. To isolate the impact of the filtering by the optical MUXs/DMUXs. the excess loss cannot be fully compensated as assumed by previous authors [4]. (a) Chain network topology. . a sampler. BPF: tunable optical BPF with low-pass equivalent transfer function denoted by H (f ). The EDFAs are represented by independent white Gaussian noise sources . Fig. T : bit period.  propagation delay.: ACCURATE MODELING OF OPTICAL MUX/DMUX CONCATENATION 923 (a) (b) (c) Fig. whose threshold is by automatically set to minimize the error probability. . we assume that the network elements are designed so that the EDFAs at the output of each network element compensate fully the total intranode insertion loss. The above assumptions enable the reduction of the network topology to the simplified block diagram shown in Fig. The signal . The input signal and equivalent input ASE noise are propagating through different channels which are represented by the transfer and the equivalent transmittance . attenuators and is filtered by the last The effective gain provided by the servo-controlled attenuators in the ASE noise path causes an additional ASE noise amplification. is the propagation group delay through the netis the specific sampling position within work.g. function respectively. It is assumed that signal attenuation induced by the optical fibers is compensated fully by the EDFAs at the input of the network elements. [5].

2. 1(a) (H (f ) = transfer function of the signal channel. but only the relative power penalties in Figs.924 JOURNAL OF LIGHTWAVE TECHNOLOGY. will produce the same noise at the receiver input as the one provided by all the optical amplifiers in the network. The equivalent input ASE noise is defined here as a fictitious white Gaussian noise source that.. which is defined for the first time here as follows. In addition. Signal and Noise Representation For mathematical convenience. and is the transmitter’s phase quency noise. is the offset of the transmitter’s actual carrier frequency from the channel’s nomis the deviation of the transmitter’s ininal frequency . Finally. The reason for the adoption of a suboptimum reference network as a fair standard for comparison is that commercially available optical MUXs/DMUXs are designed in order to obtain maximally flat passband and minimum chromatic dispersion rather than matched filter characteristics. The amplitude transfer functions have the same insertion loss at the center frequency and the same equivalent noise bandwidth as the individual actual optical MUXs/DMUXs. if placed at the input of the network. it is assumed that the ideal optical MUXs/DMUXs do not introduce any ISI. The electric field at the output of the transmitter can be . where is the average signal written as power. The distinction should be clear from the context. and linear or zero phase transfer function. 20. ideal optical MUXs/DMUXs exhibiting rectangular amplitude transfer function. 1(a) (E (t): input signal. 9–11. n (t) = equivalent input ASE noise). JUNE 2002 (a) (b) Fig.e. chirp). g : ith attenuator gain. stantaneous frequency from the transmitter’s actual carrier fre(i. (b) Final block diagram of the network topology shown in Fig. It is well known [49] that an optimum direct-detection receiver is composed of a matched optical BPF and a wide-band electronic LPF. T (f ) = equivalent transmittance of the ASE noise channel. and plex envelope of the received electric field that can be written as a function of the normalized complex envelope of the transmitted electric field (2) is the signal effective gain provided by the reducwhere servo-controlled attenuation of the insertion loss of the . we identify vector quantities with boldface type. and the gains of the last We will compare the performance of the actual network element chain with the performance of a reference network element chain. 6. All optical MUXs/DMUXs are replaced with perfectly aligned. it is assumed that there is a polarizer in front of the tunable optical BPF in the optically preamplified receiver aligned along the received signal polarization and that the receiver electronic LPF is an integrate and dump filter with impulse response duration equal to the bit period . The received electric field vector can be written as . Matrices are also denoted by boldface type. NO. VOL. the choice of reference network does not affect the essence of the results in Section III. and that can be written as (1) where is the amplitude of the normalized complex envelope (taking values in the range ). where is the normalized Jones vector for the is the normalized comreceived state of polarization. a low-pass equivalent representation [49] of the optical signals and components is used in the following. is the normalized Jones vector for the transmitted state is the normalized complex envelope of polarization. (a) Simplified block diagram of the network topology shown in Fig. Nevertheless. n (t): ith EDFA ASE noise). To distinguish vectors from scalars. can be expressed as a funcThe equivalent transmittance optical MUXs/DMUXs transmittances tion of the last servo-controlled attenuators. B.

The received ASE noise is unpolarized and can be analyzed into two orthogonal states of polarization and . that the spontaneous emission and the EDFA gain are independent of frequency factor . We can rewrite .i. The subscript denotes the with initial condition number of fiber spans. and are the real parts of the signal where and ASE noise in the and polarizations. respectively. is calculated through the relationship In (6).) where complex bandpass Gaussian processes with two-sided power spectral density (psd) (6) is the two-sided equivalent input ASE noise psd for where is the transmittance of the equivaeach polarization and lent channel seen by the equivalent input ASE noise. The photodiode is modeled as a square law detector. complex bandpass Gaussian prowhere cesses with zero mean and unit variance and the multiplication coefficient is written as (14) where denotes the average signal energy and denotes the normalized equivalent noise represents the received opbandwidth. associated with the direct-detection of the signal. the last two terms and on the right-hand side of (16) are omitted.: ACCURATE MODELING OF OPTICAL MUX/DMUX CONCATENATION 925 tors. of the ASE noise at the output of the chain of network elements can be calculated by integration of (6) (11) is the equivalent noise bandwidth of the chain of where network elements defined as (12) The total received electric field vector the photodetector can be written as impingent upon and denote the total average input/output where powers to/from the optical MUX/DMUX pair of the th node. denotes convolution. is the term de- . where the operator The th attenuator gain is equal to (4) The variance of the orthogonally polarized components . respectively (5) and are independent identically distributed (i. that all EDFAs are identical. specifically . and is the impulse response of the cascade of optical MUXs/DMUXs and the receiver’s tunable optical BPF (3) denotes the inverse Fourier transform. respectively. Relationship (13) can be rewritten in terms of the quadrature components of the signal and ASE noise (15) . respectively. parallel and perpendicular to the received signal’s electric field.ROUDAS et al. Assuming. the two EDFAs within each network element have different gains and noise figures. can be written within the signal band. (8) (13) and are i. has gain due to the action Inspection of (8) reveals that in the of servo-controlled attenuators. and are the imaginary parts of the signal and ASE noise in the and polarizations. The final expression for the output photocurrent is – (17) where is the responsivity of the photodiode. In the general case. The quantity tical signal-to-noise ratio (OSNR) measured in a resolution bandwidth equal to the bit rate. and that as (7) where is the Planck’s constant.d. without loss of generality. where is normalized to form unity and is defined as the excess noise factor (10) In the presence of a polarizer in front of the tunable optical BPF in the optically preamplified receiver aligned along the polarization.i.d. The optical power measured by the photodetector is defined as (16) where is an auxiliary variable defined by the recursion (9) . and .

is [35]. is and is an auxiliary the root of the first derivative phase function defined as (25) by a It is also commonplace to approximate and standard deviation . and [39]. In the general case. Our modifications of [37] allow to take into account arbitrary optical MUXs/DMUXs transfer functions. C. 20.i. is an auxiliary variable that in the presence or absence of polarizer. are dewhere the coefficients . Here. As its name indicates. The Gaussian pdf with mean and standard deviation can be evaluated from mean using the relationship [49] the characteristic function (26) where . [36]. it is possible to evaluate the by taking the Fourier transform pdf of the photocurrent . Photocurrent Statistics . and – the ASE–ASE noise beating in dimensionless form denotes (23) (18) – (19) and are i. where through the characteristic function denotes the expected value. This can be done . it is assumed that the transmission channel is causal. i. respectively. The semianalytic method assumes that the ISI is caused primarily by a limited number of bits. fined in Appendix A. . zero mean and unit variance that represent the shot and thermal noise. From the characteristic function.926 JOURNAL OF LIGHTWAVE TECHNOLOGY.g. and is ature. The closed-form analytic solution for can be found using the steepest asymptotic value of descent method [46] (24) where the prime denotes differentiation with respect to . For example. Then. VOL. the mean error probability can be calculated by averaging over all the possible combinations of bits (27) is the probability of occurrence of the combiwhere and is the conditional probability nation of bits to receive the complementary symbol . Error-Probability Evaluation Finally. the method consists of a combination of simulation and analysis. Here. and the shot noise (assuming Gaussian pdf). Our derivation is based on a generalization of the analysis by [37]. The analytic evaluation of the characteristic function can be done using the same formulation as in [26]–[41]. takes the values is the equivalent noise bandwidth of the respectively. In reality. the noise cdf and the error probability are calculated analytically. 6. Given the signal waveform at the receiver. For photocurrent at the output of the sampler is the evaluation of the error probability.. a of does not exist. and . so each bit is affected by the bits. If we neglect the dark current.. D. ular combination of bits are the noncentral moments and .d. the polarizer at the receiver. [38]. NO. the load resistor. and electronic filter defined as [51] (21) The above expressions can be easily modified for avalanche photodiodes [50]. the characteristic function of is given by . given that the particwas sent. e. is the absolute temperis the electronic preamplifier noise figure. shot noise follows Poisson statistics. Noiseless simulation is used to evaluate the accumulation of ISI during signal propagation. it is necessary to calculate the statistics of the random variable (RV) . the shot noise equal to variance can be written as [50] (20) where is the electron charge. As shown in the Appendix A. consider that is the bit that previous is a sequence we want to detect and of the last received bits. . [33]. real Gaussian processes with In (17). the shot-noise variance calculated approximately assuming that the level of the received is constant within the bit period with amplitude signal . which are not included in [37]. JUNE 2002 notes the signal-ASE noise beating. the value of the At the sampling instant . we use a semianalytic method [25] for the accurate evaluation of the error probability of the network. The thermal noise is a zero mean Gaussian noise with variance [50] (22) where is the Boltzman’s constant. [31]. .e.

deperiod-optical bandwidth product. In addition. we examine whether a Gaussian function can describe well the top of the transmittance of commercially available conventional AWG routers. This section is divided into two parts. it is observed that there is a systematic insertion loss difference between channels (loss imbalance [55]). and frequency. is the central frequency offset where is the insertion loss. substitution in (28) gives erfc (30) erfc where erfc as erfc is the complementary error function [52] defined . [46]. However.. is the normalized . Disadvantages of conventional AWG routers are the relatively high insertion loss and the passband curvature [54]. and optimum threshold defined in the interval is the generalized Marcum function [49] defined by (32) denotes the modified Bessel function of the first where kind of order . III. we investigate the importance of random central frequency offsets from the channels’ nominal frequencies and the variations of the 3-dB filter bandwidth from channel to channel and from device to device. the steepest descent method [26]. again. The main purpose of this study case is to illustrate the theoretical model of Section II. the magnitude of the transfer function of each channel can be approximated at the center of the passband by a Gaussian function (see. (31) becomes identical to the error probability For of a matched filter direct-detection receiver.. STUDY CASE—CONCATENATION OF AWG ROUTERS AWG routers are a promising technology for transparent dense wavelength-division multiplexing networks due to their ability to multiplex/demultiplex a large number of channels. the mean error probability is approximately given by [31] (31) is an integer number. e. This section presents an accurate study of the concatenation of conventional AWG routers (i. i. [41].: ACCURATE MODELING OF OPTICAL MUX/DMUX CONCATENATION 927 The conditional error probability follows: is defined as (28) is the decision threshold. there is a tradeoff between passband flatness and insertion loss [54]. their negligible chromatic dispersion. Transmittance of AWG Routers Fig. and the laser carrier frequency offset from the channel’s nominal frequency. In addition. . the equivalent noise bandwidth of AWG routers. However. For example. In equivalent low-pass formulation. notes the least integer not smaller than . Several techniques to increase the passband flatness are proposed in the literature. [56]) (33) E. In the first part. approximately equal to the bit where . with round passband top) in a transparent multiwavelength optical network chain like the one presented in Fig. their compactness (i. Dotted lines indicate the periodic transmittance of one of the channels. i. which is related to the 3-dB bandwidth (full-width at half-maximum) of the transmittance through dB (34) More relationships about the properties of this type of AWG router are given in Appendix B.. the Gaussian approximation (33) does not describe the side lobes in the stopband (see Fig. Error Probability for the Reference Network For the reference network. integrated form) and their low cost [53].. is the root of the first derivative of (27) with respect to . 3).e.e. It is observed that the transmittance for each channel presents round top in the passband and side lobes in the stopband. by a Gaussian pdf with mean If we approximate and standard deviation . The optimum value of the where decision threshold . An alternative expression for the conditional error can be found by substitution of probability in (28) (29) The asymptotic value of the integrals in (29) can be found using.e. is the cutoff from the channel’s nominal frequency .g. this discrepancy becomes less important as the number of concatenated optical MUXs/DMUXs increases.ROUDAS et al. accurate description of the transmittance of an individual .. [38]. A.e. 1(a). The integrals (28) must be evaluated numerically. edge channels within the free spectral range of the device exhibit more loss than the middle channel. The second part is devoted to the evaluation of the error probability of a chain network with AWG routers as a function of the number of fiber spans. 3 shows the measured transmittance from an input port to all output ports of a commercially available 8 8 conventional AWG router. This subject was partly studied previously by [6] and [11] for 10-Gb/s nonreturn-to-zero (NRZ) and by [17] for 40 Gb/s return-to-zero (RZ) transmission. Obviously. which minimizes the mean error probability . with channels spaced 100-GHz apart.

optical MUX or DMUX within only 10 dB from the top of the passband is sufficient for the description of the overall transfer function of a chain of ten optical MUXs/DMUXs within 100 dB from the top of the passband. in (33). it is assumed that the optical AWG routers are not dispersive devices within the signal band and the phase is set equal to zero.928 JOURNAL OF LIGHTWAVE TECHNOLOGY. Fitting of the normalized transmittance measurements is used next for the extraction of the center frequency and the 3-dB bandwidth and the evaluation of their statistics. 2 Fig. Transmittance of a commercially available 8 8 AWG router with channels spaced 100-GHz apart from an input port to all eight output ports. VOL. For example. 4. In practice. 3 (points) and their fit- ting by the square of (33) (solid line). Dotted lines indicate the periodic transmittance of one of the channels. JUNE 2002 Fig. the fit is excellent. commercially available devices are often characterized by measuring the 3-dB points from the top of the passband and defining their average as the center frequency. [57]. these errors are assumed small and the behavior of the device is dominated by the curvature of the passband [21]. 3.2 For this purpose. 20. 6. and thickness from waveguide to waveguide induce amplitude and phase errors in the transfer function. Measurements within 10 dB from the top of the passband are used for the fitting. NO. measurements of the transmittance of two commer2It is worth noting that there is no universally accepted method for the measurement of the center frequency and the 3-dB bandwidth of optical MUXs/DMUXs. 3 (points) and their fitting by the square of (33) (solid line). Fig. In addition. . small variations of refractive indixes. Nevertheless. Measurements of the normalized transmittance of a sample AWG router channel from Fig. 4 shows measurements of the normalized transmittance of one AWG router channel from Fig. width. [58]. We conclude that the Gaussian approximation accurately describes the behavior of the transmittance of conventional AWG routers in the center of the passband and its vicinity.

Consequently. 6 . with standard deviation of 2. Fig. Histogram of AWG router central frequency offsets from the channels’ nominal frequencies (based on measurements of 96 transmittances).: ACCURATE MODELING OF OPTICAL MUX/DMUX CONCATENATION 929 Fig. with channels spaced 100-GHz apart. 5. are presented in Figs. This assumption is not strictly correct so the extracted center frequencies and 3-dB bandwidths are partly correlated. Fig. are fitted by the square of (33). we made the (arbitrary) assumption that the transfer functions of different channels and different input/output ports within the same device are independent. respectively. 8 GHz). Histograms for the central frequency offsets from the channels’ nominal frequencies and the 3-dB bandwidths for two commercially available 8 8 conventional AWG routers. we make no distinction between random and systematic error. with channels spaced 100-GHz apart. which is about 7% of the mean 3-dB filter bandwidth.ROUDAS et al. 5 and 6. 5 shows that central frequency offsets from the channels’ nominal frequencies are concentrated around zero in a range ( 8 GHz.7 GHz. 6. Fig. cially available 8 8 conventional AWG routers. In addition. Histogram of 3-dB bandwidths of AWG router channels (based on measurements of 96 transmittances). the following statistics provide only a rough estimation of the importance of random variations in the transmittance of AWG routers. Due to the lack of a large sample of devices. as a result of our assumption.

We conclude that. 7 shows the signal transfer function seen by the equivand the normalized transmittance alent input ASE noise (dotted lines) for one. five.930 JOURNAL OF LIGHTWAVE TECHNOLOGY.5 GHz. (For example. The impact of random variations on the system’s performance will be addressed in a future paper. three (dashed-dot line). AWG Router Cascadability The AWG router cascadability depends on the selection of network design parameters.g. ten. in a first approximation. An exhaustive study of the influence of all possible combinations of the aforementioned parameters on the network performance is beyond the scope of this example. we assumed that all attenuator gains . . absence of chromatic dispersion and nonlinearities. At the limit implies that the filtering of the ASE The behavior of noise by the optical MUX/DMUX chain is less severe than the filtering of the signal due to the distributed generation of the ASE noise in the network. it is beneficial to use a narrow-band optical BPF at the optically preamplified direct-detection receiver for rejection of the long ASE noise tails. and 12 (dashed line) fiber spans. and 15 fiber spans. To isolate the impact of the AWG router filtering from other transmission effects. It is observed that coincide in the proximity of the origin for any number (b)  as a function of the received OSNR E  =N Fig. extinction ratio. JUNE 2002 (a) Fig. fiber chromatic dispersion and nonlinearities. Since all curves present symmetry around the origin. and so forth. and 15 fiber spans. we conclude that the random central frequency offsets from the channels’ nominal frequencies and the variations of the 3-dB filter bandwidth of commercially available AWG routers are small and. 20. which exhibit a mean equivalent noise bandwidth equal to 40. for one fiber span. can be neglected. 8(a). The equivalent noise bandwidth of individual AWG routers in . of fiber spans. 6. with a standard deviation of 1. the transfer functions of concatenated AWG routers will be considered identical and perfectly aligned. the following network parameter set is used in this study: ideal NRZ pulses with infinite extinction ratio and zero chirp. Based on the above. e. the tion implies a bit rate error probability for the reference network (31) is also shown (solid line). which is 4% of the mean 3-dB filter bandwidth. 8. VOL. ASE-noise-limited direct-detection receiver.4 dB of additional power compared to the reference network are required to achieve an error probability of 10 . Appendix C presents the details of the implementation of the semianalytic method for the error-probability evaluation in this specific case study. they are very different for and their discrepancy increases as a function of the . as a Fig.6 GHz. 7. number of spans. transmitted optical pulse shape. In Fig.. receiver type. When the number of spans increases. . NO. for a large number of fiber spans. in the presence and in the absence of polarizer at the receiver respectively. shows that the mean 3-dB bandwidth is 38 GHz. chirp. 1. B. Here. (a) In the presence of polarizer at the receiver. 8(a) and (b) shows the mean error probability measured in a resolufunction of the received OSNR tion bandwidth equal to the bit rate for one (dotted line). Mean error probability P measured in a resolution bandwidth equal to the bit rate for one (dotted line). Away from the origin. equivalent noise bandwidth ). (b) In the absence of polarizer at the receiver (condition: B = 4R ). ten. In the following. negligible shot and thermal noises. and 12 (dashed line) fiber spans.) For a comparison. 3. only the positive frequency semiaxis is displayed. three (dashed-dot line).e. five.. are unity (case of wide AWG routers compared to and the signal spectral occupancy). Signal transfer function H (f ) (solid lines) and normalized transmittance T (f ) (dotted lines) of the chain of network elements seen by the equivalent input ASE noise for one. the above assumpGb/s. in the case of these graphs is assumed the AWG routers shown in Fig. initially the sensitivity is improved because . and a fourth-order Bessel electronic LPF with cutoff frequency (i. It is observed depends on the effective attenuator gains from (6) that . (solid lines) Fig.

61 dB power penalty compared to the reference network. where there is only 0. where there is only 0. in Fig. However. For a matched filter direct-detection receiver. . the OSNR measured in a resolution bandwidth equal to the bit rate required to achieve an error probability of 10 is approximately 38.47-dB power penalty compared to the corresponding reference network. the actual network for three fiber spans performs slightly better than the reference network. depending on traffic demands. in all cases examined in the present paper. a broad maxthe imum sensitivity is achieved after 12 fiber spans. and 15 (dashed-dot line) fiber spans in the presence of polarizer at the receiver. Fig. a broad maximum sensitivity is achieved after four fiber spans. a broad maximum sensitivity is achieved after fifteen fiber spans. such optimization is impractical because in transparent optical networks. as expected. This is due to the fact that AWG filtering changes the properties of the ASE noise at the input of the direct-detection receiver. 9. Notice that. and . in principle. The results case fall in between the and for the cases. i.6 dB at an error probability of 10 . and 15 (dashed-dot line) fiber spans in the presence of polarizer at the receiver. equipment failures. Optical power penalty compared to the reference network.. i.85 dB). in reconfigurable optical networks. to maximize the network performance for a given number of concatenated AWG routers by jointly optimizing the optical bandwidth of the individual AWG routers and the electronic LPF bandwidth of the receiver. B = 4R . The role terms of the polarizer is more important for larger individual AWG router equivalent noise bandwidths and smaller number of fiber spans. Similar conclusions can be drawn from Fig. in the presence of polarizer. and so forth. where there is only 0. in the presence of polarthe receiver.e. This apparent paradox is due to the fact that the reference network is suboptimum. These results indicate that it is possible to concatenate an arbitrary number of conventional AWG routers. a minimum power penalty 0. the performance improvement due to the polarizer is marginal (less than 0. due to the elimination of the ASE–ASE noise beating and on the right-hand side of (16). A matched filter direct-detection receiver [not depicted in Fig. presence (solid lines) or absence (dotted lines) of polarizer at case. In the absence of polarizer. Fig. It is observed that the slope of the error-probability curves changes as a function of the number of fiber spans. despite the inherent passband curvature of these devices. it is possible. which requires an OSNR of 16. ten (dotted line).61 dB is achieved after three fiber spans. 8(a) and (b) to avoid clutter] presents superior performance than both the actual and the reference network at all error probabilities. ten (dotted line). the power penalty slowly increases due to the increase of ISI. In addition. at high error probabilities (i.87-dB power penalty compared to the reference network.16-dB power penalty compared to the reference network. five fiber spans. 9 shows the optical power penalty compared to the reference network at an error probability of 10 as a function of the number of fiber spans for three individual AWG router equiv. 6R . A maximum sensitivity is achieved after three fiber spans. 8(b). and 8R . at an error probability of 10 . As the number of spans continues to increase.e. for three individual AWG router equivalent noise bandwidths. Fig. 8(a).. as a function of the equivalent noise bandwidth of the individual AWG routers for five (solid line). it is instructive to examine the impact of the equivalent noise bandwidth of individual AWG routers on the performance of the network. In addition.5 (15.: ACCURATE MODELING OF OPTICAL MUX/DMUX CONCATENATION 931 Fig. by increasing the equivalent noise bandwidth of individual AWG routers for fixed bit rate at the expense of channel spacing. Using the current model. in the presence (solid lines) or absence (dotted lines) of polarizer. In the absence of polarizer. 10. a minimum sensitivity is achieved at where there is only 0. 10 shows the optical power penalty at an error probability of 10 (compared to the corresponding reference network at the optimum operating point) as a function of the equivalent noise bandwidth of individual AWG routers for constant . where the signal passes through five (solid line). the optical bandwidth narrowing reduces the power of the ASE noise without essentially distorting the signal. low OSNRs).e.41-dB power penalty compared to the reference network. Optical power penalty compared to the reference network. in the alent noise bandwidths. where there is only 0.ROUDAS et al. as a function of the number of fiber spans. We consider electronic LPF equivalent noise bandwidth three different paths in the network. In case.. In the case of ten fiber spans. Nevertheless. it is observed that an ideal polarizer always improves the performance of the optically preamplified receiver. For . each optical signal passes through a different number of concatenated AWG routers. In the izer. respectively. . However. each optical signal may be rerouted through several different paths during the network’s lifetime.5 dB) and does not justify the implementation cost of this device. at an error probability of 10 .

28 b/s/Hz in a 10. a min.g. but different signs. [46]. ten (dotted line). 11. the model predicts a minimum optical power penalty of 1. a minimum sensitivity is achieved at . when the signal passes through five (solid line). the bandwidth of AWG routers.932 JOURNAL OF LIGHTWAVE TECHNOLOGY. arbitrary optical MUX/DMUX and electronic LPF transfer functions. the model predicts a monotonic increase in the optical power penalty as the laser carrier frequency offset increases.2-Tb/s transmission experiment [60].5 dB in all cases. However. The model is based on a semianalytical technique [25] for the evaluation of the error probability. and the laser carrier frequency offset from the channel’s nominal frequency. Finally. where there is only 0.g.9 dB at an error probability of 10 . Preliminary simulations performed by the authors indicate that this effect is universal and can be observed for other modulation formats and optical MUXs/DMUXs types as well. ten (dotted line). see [26] and the references therein) and later in the context of Fig. only the optical power penalty for positive misalignments is displayed. numerical solution of the associated homogeneous Fredholm equation of the second kind [44]–[46]. including expansion of the ASE noise impingent upon the photodiode in Karhunen–Loève series [42]–[44]. The model can be used to derive specifications for arbitrary optical MUXs/DMUXs in order to achieve a prescribed power penalty in conjunction with different modulation formats. and 15 (dashed-dot line) fiber spans in the presence of polarizer at the receiver (condition: B = 4R ). respectively. the power penalty is the same for laser misalignments of the same magnitude. 20. The computation of the error probability involves several steps. at an error probability of 10 . which requires an OSNR of 17 dB at an error probability of 10 . VOL. taking into account arbitrary pulse shapes. this effect was proposed as a means to achieve . and asymptotic evaluation of the pdf at the output of the direct-detection receiver from the characteristic function using the method of steepest descent [26]. the concatenation of conventional AWG routers in a transparent multiwavelength optical network chain is studied. in the case of ten and 15 fiber spans.. was first observed theoretically and experimentally by [6] and was attributed to the filtering of the vestigial side band. In the case of five fiber spans. 6. Fig. for the first time. compared to the reference network for a and . which occurs when the effective equivalent bandwidth of optical MUXs/DMUXs cascade becomes comparable to the bit rate. In addition. NO. as a criterion for the cascadability of optical MUXs/DMUXs. respectively. In the case of 15 fiber spans.34 and 1. as a function of the laser carrier frequency misalignment from the channel’s nominal frequency. the present technique offers superior accuracy. laser carrier frequency offset from the channel’s nominal frequency. Due to the symmetry in the optical signal spectrum and the AWG transfer function. measurements of the transmittance of commercially available AWG routers are fitted in order to extract the values of the channels’ central frequencies and 3-dB bandwidths and evaluate their statistics. at an error probability of 10 . which guarantees an optical power width equal to penalty less than 0. were used in the past. IV. The sensitivity degrades sharply for narrower equivalent noise bandwidths. Then.97 dB. Approximate calculations of the error probability with semianalytical techniques. JUNE 2002 ultrahigh-capacity WDM transmission [59] and was used to obtain record spectral efficiency 1. the notions of effective gain and excess noise factor due to servocontrolled attenuators are introduced. The model helps to identify the underlying mechanisms for performance degradation and to provide a rigorous definition for the excess loss due to spectral clipping resulting from optical MUX/DMUX concatenation. e. Therefore.. This surprising result. e. [45]. APPENDIX A DERIVATION OF THE CHARACTERISTIC FUNCTION OF THE OUTPUT PHOTOCURRENT The analytical evaluation of the characteristic function of the filtered noise at the output of a square-law detector was performed initially in the context of radio communications (e.06-dB power penalty compared to the corresponding reference network.g.25-dB power penalty compared to the corresponding reference network. Statistical results indicate that first-order random variations in the transfer functions of AWG routers from device-to-device can be neglected. diagonalization of a quadratic form [37]. Recently. [41]. where there is only imum sensitivity is achieved at 0. Gaussian RZ pulses and MUXs/DMUXs composed of MI filters. To illustrate the model. and 15 (dashed-dot line) fiber spans in the presence of polarizer at the receiver. 11 shows the optical power penalty compared to the reference network. First. However. a reference network with ideal optical MUXs/DMUXs is defined for comparison.. the error probability is evaluated as a function of the number of AWG routers. [3] and [6]. as a function of the laser carrier frequency offset from the channel’s nominal frequency when the signal passes through five (solid line). We conclude that a possible compromise is to choose the AWG router equivalent noise band. SUMMARY This paper presents a general theoretical model for the study of concatenation of optical MUXs/DMUXs in transparent multiwavelength optical networks. Optical power penalty compared to the reference network. and non-Gaussian photocurrent statistics at the output of the direct-detection receiver assuming ideal square-law detection [26]–[41]. which requires an OSNR of 16.

always degenerate. i. new i.d. thus. we rewrite the photocurrent given in (17) at the output of the sampler – (35) where .d. Here. The method consists in using a quadrature rule for the integral in the left-hand side of (39) and in discretizing the time at the same nodes in order to create a matrix eigenvalue equation. the same notation as in [37] is used. In the following... the derivation of the expression (23) for the characteristic function is described briefly. with diagonal elements equal to the standard deviations of is a column vector with elements . where and are new RVs following a normal distribution. RVs (39) (43) where (44) in (43) are multiplication coefficients that deObviously. For the sake of completeness. Gaussian RVs using a Karhunen–Loève expansion of the equivalent input ASE noise.e. The orthonormal funczero mean and variance and the variances of are the eigenfunctions tions and the eigenvalues. we adapt the formalism of [37] to the study of optical MUX/DMUX concatenation. involving the signal provided from digital simulation. substituting in (41) yields – (42) are the eigenvalues of . For the quadrature.ROUDAS et al. well suited for the evaluation of integrals (44). of the homogeneous Fredholm equation of the second kind [45] is the autocorrelation function of and can where be evaluated by the real part of the inverse Fourier transform of .g. respectively. It can be shown that are uncorrelated and. e. where can be written in series in terms of the Similarly.: ACCURATE MODELING OF OPTICAL MUX/DMUX CONCATENATION 933 optically preamplified direct-detection receivers. the equivalent input . into Substituting the Karhunen–Loève expansion of (37) yields – (40) . . . where we defined Relation (40) is a quadratic form that can be rewritten in matrix formulation [37] (36) – (37) – (41) is a diagonal matrix where denotes transpose. due to optical MUX/DMUX center frequency offsets or systematic passband tilts. the polarizer at the receiver. we use the trapezoidal rule because it allows for equally spaced nodes and is. It is worth noting that the eigenfunctions and the eigenvalues for the real and imaginary parts of the and polarizations are . where are new column vector with elements new RVs following a normal distribution.i.e. the shot and thermal contribution are added to the characteristic function. the convolutions in (18) and (19) describing the low-pass filtering of the signal–ASE and ASE–ASE noise beatings are transformed into sums of i..g. Wherever possible. and are a function of the pend on the unitary transformation projection of the received signal on the orthonormal functions For the evaluation of the statistics of (37). tions and the corresponding coefficients are cross correlated [49. In the first part. [27]–[41] for different types of optical and electronic LPFs. borrowing elements from [27]–[41]. independent [61]. where is the duration of the impulse response of the electronic LPF. As a starting point. the new RVs Finally. Our modifications of [37] consist of the inclusion of arbitrary optical MUXs/DMUXs transfer functions. However. the characteristic function of the signal–ASE and ASE–ASE noise beatings is calculated. ASE noise quadrature components are represented in the time interval . Equation (39) can be solved analytically only in certain spe[44]. where denotes imaginary part. i. In the second part. [43] (38) are independent Gaussian RVs with where the coefficients [42]–[44]. The interested reader can find more details in [37]. where is a transformation matransformation and is a trix whose columns are the eigenvectors of . (39) must be resolved numerically [46].e. and the shot noise variance.i. i. The derivation is divided into the following parts. e. The quadratic form (41) can be diagonalized using a unitary . . it is presents Hermitian symmetry. as is assumed that the case for perfectly aligned AWG routers with symmetric transfer functions. therefore. then the real and imaginary parts of the ASE noise along the and polariza. . Finally. 153–157]. if does not present Hermitian symmetry around the origin. by a Karhunen–Loève expansion [42]. pp. In the case of arbitrary optical cial cases of MUX/DMUX transfer functions. so this cross correlation is negligible...

[41]. 6) Evaluation of the mean error probability by averaging over the conditional error probabilities for all bits. the equivalent noise bandis slightly larger than the 3-dB bandwidth. 7) Finally. 1) Evaluation of the signal waveform at the output of the receiver by noiseless simulation of the the block diagram of Fig. VOL. 6. assuming ideal square-law detection and using an expansion of the ASE noise impingent upon the photodiode in Karhunen–Loève series [42]–[44]. Notice that depend also on the sampling instant . we find (51) where . equal to dB concatenated AWG routers is The impulse response of given by (50) is the transfer function of the cascade of optical where MUXs/DMUXs. 2) Analytical evaluation of the ASE noise psd at the photodiode. In the simulation. zero chirp. we can evaluate analytically the equivalent noise bandwidth of one AWG router by substitution of (33) in (48) and use of the in [62] relationship (49) Notice that. (47) APPENDIX B ANALYTICAL EXPRESSIONS FOR AWG ROUTER PARAMETERS The equivalent noise bandwidth of a BPF is defined as (48) denotes the equivalent low-pass transmittance of where the BPF. The algorithm for the semianalytic evaluation of the error probability involves several computational steps. the characteristic function of the and can be evaluated as – For the simulation described in Appendix C. NO. a maximal-length pseudorandom sequence with an additional zero added after the longest string of zeros. respectively. [38]. and infinite extinction ratio. By substitution of (33) in (50) and use of the relationship in [62]. numerical minimization of the mean error probability by optimization of the decision threshold. where the ASE noise is more important.e. From (42) and (43). a pseudorandom sequence with period up to 2 is used to modulate the optical signal. relationship (4) is used. since it is assumed it is much higher than the ASE noise power for normal network operating conditions. the duration of the impulse response (51) is arbitrarily defined as eight times the root-mean-square (rms) width [63] (52) where is defined as (45) in the presence or absence of where takes the values polarizer. 2(b). but this dependence is implied in order to simplify the notation. i. fiber Fabry–Perot . The optical waveform is assumed NRZ with perfect rise and fall times. If we neglect the periodicity of the AWG router transmittance. for the shot and thermal The characteristic functions noises can be readily evaluated [49] (46) Finally. This will introduce a small error for high error probabilities. 20.. 4) Asymptotic evaluation of the pdf of the photocurrent at the output of the receiver for each bit from the corresponding characteristic function using the method of steepest descent [26]. For simplicity. The number of coefficients of each FIR filter is calculated so that the duration of the impulse response is equal to eight times the rms width (see Appendix B).934 JOURNAL OF LIGHTWAVE TECHNOLOGY. 5) Evaluation of the conditional error probability for each bit by numerical integration of the pdf tail.g. which is width . For the evaluation of the effective gain provided by each servo-controlled attenuator. JUNE 2002 . AWG routers are represented by finite impulse response (FIR) filters. 3) Analytical evaluation of the characteristic function of the noise at the output of the receiver for each bit. This is a de Bruijn sequence [64]. only the signal is considered in the calculation of the average power.. Commercially available optically preamplified receivers usually have a narrow tunable optical BPF (e. [46]. for one AWG router. Successive iterations with different sequence lengths show that longer sequences increase computing time and provide a negligible increase on accuracy. the conditional characteristic function is given by of It is straightforward to show that APPENDIX C IMPLEMENTATION OF THE SEMI-ANALYTIC METHOD and (53) .

K. 11. M. Tong. M. pp.. Mohs.” IEEE Photon. Jul. Oct. Goodman. 21–22. Munich. v. 2000. A. MD. S.” IEEE J. P.. . H. Miyata. and F. Leisching. this requires the numerical solution of the homogeneous Fredholm equation of the second kind (39) [46]. Freund. J. Wagner. Aug.” IEEE Photon. CA. For the evaluation of the characteristic function from (23). Li. and T. Garrett for providing the measurements of Fig. M. Burtsev. Feb. 10. Technol. Lett. Wang. 1997. N. “Fabry-perot filter bandwidth effects on end to end transmission performances in a multiwavelength transport network. Testa. R. P. S. Otsuka. 9. vol. First.ROUDAS et al. Litchinitser. Optical Fiber Communication Conf. “Performance of cascaded misaligned optical (de)multiplexers in [5] multiwavelength optical networks. for simplicity. The autocorrelation of the ASE noise is evaluated using the inverse fast Fourier transform (FFT) with 1024 points. Agrawal. Dallas. Aug. Technol. G.” IEEE Photon. C. Optical Fiber Communication Conf. Dallas. and Y. Tomkos. Boscovic for helpful discussions. Antoniades. C. R. 1996. vol. pp. Roudas. vol. H. D. vol. [11] C. K. and G. Sept. C. 1997. M. K. vol. J. and A. Froberg. For the quadrature. A. [16] K. vol. Feb. Conf. 1027–1030. Meeting. 327–329. Topics Quantum Electron.. Garrett. 3. E. pp. Ho. June 1996. and G. can be assumed. Technol. 20–21. “Dispersion of cascaded fiber gratings in WDM lightwave systems. “Influence of filtered ase noise and optical filter shape on the performance of a WADM cascade. R. 2001. “Influence of realistic optical filter characteristics on the performance of multiwavelength optical networks. E. Tong. E. Here.. P. 1523–1529. Geiger. the propagation delay is calculated by correlation of input and output sequences. Wagner. Lee. Cartledge. Lett. Baltimore.” in Optical Fiber Communication Conf. and B. Aug. M. E. “Comparison of cascadability of different types of commercially available wavelength (de)multiplexers. 1997. Tomkos. Nissov. I. F. pp. pp.” IEEE Photon. R. “Four-channel 10 Gb/s transmission over 15-wavelength-selective cross-connect paths and 1175 -km dispersion compensated standard single-mode fiber links. “MONET: Multiwavelength optical networking. Optical Fiber Communication Conf. Krüger. Szemiot for corrections in the syntax of the manuscript. The thermal and shot noise variances are set to zero. and D. “Modeling of waveform distortion due to optical filtering. Optical Communications. Eggleton. the noiseless eye opening at the output of the receiver is used to determine the optimum sampling instant . pp. 1999. and L. Boskovic. A jitter window around . D. Lightwave Technol. 143–146. The evaluation of pdf’s from the characteristic function is done using the method of steepest descent. Strebel. 2. M. Chikama. N. Foisel. Conf. and K.. is calculated in order to maximize the eye opening. 542–543. vol. Lett. pp.. Litchinitser. The optimum threshold is evaluated numerically. Shehadeh. D. Nov. Eggleton. “Cascadability of array-waveguide grating (de)multiplexers in transparent optical networks. For simplicity. 1998. 1649–1651. and P. vol. 14. Antoniades. J. Sept. Stoll. “Misalignment distortion penalty in moderate-speed directly modulated WDM systems and its reduction using fiber Bragg grating lasers. pp. Slusher. Lett. [18] I. B. Wagner.” in Proc. Germany. Khrais. Feb. J. we use the trapezoidal rule with 64–512 nodes. due to the clock recovery circuit. B. Fischer. vol. 1403–1405. [22] K. Elrefaie. I. Merli. pp. it is assumed that the equivalent noise bandwidth of the receiver optical BPF is much wider than the equivalent noise bandwidth of each optical MUX/DMUX. vol. in the following simulations. Pagnan. “System impact of cascaded optical add/drop multiplexers based on tunable fiber Bragg gratings. and T. Antoniades. 1997. pp. [8] I. P. Henion. B. “Implications of fiber grating dispersion for WDM communication systems. vol..” in Proc.” IEEE Photon. Foisel. and T. Jorgensen. [2] C. and R. A fourth-order Bessel LPF with cutoff is used at the receiver. 1997. 1997. L... and . pp. N. Eur.” in Proc. Freundand. and Ms.K.. 33.: ACCURATE MODELING OF OPTICAL MUX/DMUX CONCATENATION 935 with optical bandwidth of the order of [65]) with an autotracking circuit that allows the filter to align its transmission peak with the signal carrier frequency. K. Mar. [6] C. M.. Dallas. M. E. [9] B. Lett.. 2000. 7. 223–226. vol. vol..” IEEE Photon. F. 8. Technol. REFERENCES [1] R. Vodhanel. Rauschenbach. [3] F. Aug. no.” Electron. [14] M. L. N. Antoniades. Chen. Eur. 3. F. Onaka. no. TuE4. Roudas. Lett. Saleh. pp. B. Khrais. 16. Optical Fiber Communication Conf. Messerschmitt. The evaluation is performed in two steps. Liu. Habiby. pp. Edinburgh. S. Patterson. pp. ACKNOWLEDGMENT The authors would like to thank Dr. Lenz. Alferness. [17] H. TuE3. Foisel. 296–298. Habiby. 1997. R. 19–20. [4] N. F. U. T. . 1998. Caspar. D.” J. [13] T. F. The evaluation of conditional error probabilities directly from the characteristic function using (29) is oscillatory. 1992. 2000. [20] M. vol. IEEE/LEOS Annu. Strebel. “Impact of external modulator in an eight-wavelength eight-node WDM ring network. p. vol. Stern. Lett. 9. H. “Impact of optical demultiplexers on DWDM transmission of 40 Gb/s channels. [19] F. 1348–1350. P. vol. Therefore. Then. [7] H. [15] H. J. Pedersen. S. 2. Lett. Roudas. Nov. TuE2. Liaw.. 1994.” in Proc. 1349–1355. TX. Technol. and B. 1038–1040. unless a successful guess of the vicinity of the root is found. A. E. 1624–1625. vol. 6. Furst. R. Otani.. pp. Helmolt. A. and R. A. Ahmed. vol. pp. F. G. E.. shown in Appendix A. 1995. 11. no. 19. vol. Stern. S. A. S. The evaluation of the equivalent transmittance is done analytically using the recursive relationship (8). Its sole purpose is to eliminate the periodicity of the ASE spectrum inherent in AWG routers and does not cause additional insertion loss or signal filtering. Wagner.. 8. The pdf’s computed by the steepest descent method are in excellent agreement with the pdf’s given by the FFT of the characteristic function. The optimum threshold occurs at the intersection of the sum of pdf’s of the ZEROs and the sum of pdf’s of ONEs of the de Bruijn sequence.. Feb. numerical integration of the pdf tail is preferable. Technol. Lett. Nov. 1997. and M. N. N. R. U. S. Kuznetsov. Sugaya. R. Hesse. “Transmission performance through cascaded 1-nm arrayed waveguide multiplexers at 10 Gb/s. 9. TX.” IEEE Photon. Drs. Mar.” in Proc. C. . 6.” IEEE Photon. and S. vol. Elrefaie. Lightwave Technol. K. 1997. pp. pp. and A. D.. K. Chiao. “Bandwidth and ripple requirements for cascaded optical (de)multiplexers in multiwavelength optical networks. G. “Impact of filter concatenation on the performance of metropolitan area networks utilizing directly modulated lasers. 1414–1416. Ho.” J. Optical Fiber Communication Conf. ThP3. Strebel. As the coefficients . Anaheim. vol.” J. [21] C.” in Proc. S. pp./Apr. no. WBB4.” in Proc. 8. Optical Communications. 6. Dec. WM29. S. [12] N. “Performance degradation of multiwavelength optical networks due to laser and (de)multiplexers misalignments. H. Aug.” in Proc. Caspar. 1148–1161. F. [10] I. Richter. Technol. J. Bock. Suzaki. Technol. Bock. N. are needed. S. and B. 1073–1075. Fischer. Lightwave Technol. H. San Jose. R. and G. pp. “Channel capacity optimization of chirp-limited dense WDM/WDMA systems using OOK/FSK modulation and optical filters. K. Caspar. “Cascadability of passband-flattened arrayed waveguide-grating filters in WDM optical networks. . CA. 16... 1999. vol. frequency must be found through a The optimum sampling instant numerical minimization of the error probability. Liu and W. TX. N. Select. N. Nov. “Power penalty for optical signals due to dispersion slope in WDM filter cascades. Soole. C. L. 1411–1413.

P. Frignac. Shitz. [31] P. M. Bigo. Veith. and J. [39] S. Digital Transmission Theory. New York: Academic. Shift Register Sequences.936 JOURNAL OF LIGHTWAVE TECHNOLOGY. C. 11.7 Gb/s PDM/WDM) transmission over 100 km teralight fiber with 1. 12. CA. Anaheim. D. R. Humblet and M. 170–170. arbitrary pulse shape and pre. T. Fiber Technol. and K. Gross. vol. 8. Nov. Bayard. photograph and biography not available at the time of publication. [24] H. pp. 450–463. 363–389. vol. Chowdhury (S’89–M’89). Optical Communications. Stern. Optical Fiber Communication Conf. Nov. 1804–1812. Koch. vol.. Bertilsson.” Opt. “10. A. Feb. 1. pp. 1987. pp. Roudas. F. Mar. 11. Van Trees. New York: Wiley. D. vol. Y. 1996. Sept. and Products. no. C. Arfken.. Handbook of Mathematical Functions.. G. no. pp. Abramovich and P. Piscataway. 6. San Francisco. “Optical preamplifier receivers: Application to long-haul digital transmission. 7. 1. pp. B. pp.” J. 6. P. vol. Baussmann. Chinn. “On the bit error rate of lightwave systems with optical amplifiers.. Nice. Nicolas. Optical Fiber Telecommunications IIIB. [48] I. Optical Fiber Communication Conf.” IEEE Trans. and V. Fondeur. vol. Ryzhik. vol. Sarlet. Jacobsen. “Sensitivity of a direct WDM-system with a frequency selective optical receiver and optical preamplifier. Anaheim. 32–36. Communication Systems and Techniques. Idler. Henry. 4. 1995. Schwartz. 1998. Apr..” IEEE Trans. 1. New York: McGraw-Hill. A. Lightwave Technol. “Performance bounds and cutoff rates of quantum limited ook with optical amplification. Theory. Biglieri. 4.” IEEE J. 1981. Saito. 1994. “Distribution of the filtered output of a quadratic rectifier computed by numerical contour integration. Methods of Mathematical Physics. . F. vol. 20. An Introduction to the Theory of Random Signals and Noise.. Richards. 1998. Wagner (S’84–M’85). Houston. E. Hayashi. Gehler and K. D. pp. M. 153–170. pp. Fiber Technol.-S. France.and postdetection filtering. Technol. pp.. [55] O. Einarsson. Nov.” in Proc. vol. 1989. [47] I. Jan. 1895–1906. [44] C.” J. New York: Academic. [51] S. pp.. H. pp. 2000. Davenport Jr and W. 3rd ed. vol. 13. [61] M. Hilbert. [25] M. “Calculation of bit-error probability for a lightwave system with [29] optical amplifiers and post-detection Gaussian noise. pp. and C. 1509–1511. M. 1987. Digital Communications. Commmun. Jul. Helstrom.” IEEE Photon. [59] W. vol. Areas Commun. vol. vol. 1493–1503. 10. Oct. [49] J. [50] G. 5. Bennett. “An efficient simulation model of the erbium-doped fiber for the study of multiwavelength optical networks. vol. Rocha. Stein. pp. [42] H. Lightwave Technol. Oda. “Techniques for estimating the bit error rate in the simulation of digital communication systems. Helstrom. Jackel. photograph and biography not available at the time of publication. E. vol. “Performance degradation of long-distance optical WDM networks due to bandwidth limitations of wavelength multiplexers and demultiplexers. 510–530. no. no. Eur. K. [30] F. Kaminow and T. “Evaluating the error probability in lightwave systems with chromatic dispersion. 1995. and M. D. 2000. 1999. W. T. N. Eds. pp. 1319–1321. 18. and R. Lightwave Technol. 1995. “Some statistical remarks on the derivation of ber in amplified optical communication systems. B. vol. Antoniades (S’91–M’92). Jul. 31.” in Proc. Wagner. J. NY: Pergamon Press. Jackel.” J. Oct. pp. Lightwave Technol. Lightwave Technol. R. B. and Y. 1997. Fundamentals of Photonics. H. vol. no. 1984.” J. Erbasch. Otani.. S. I368–369.-S. 977–981. R. G. Q. “Error-rate performance of optical amplifiers with fabryperot filters. pp. T. L. Habiby. Benedetto. “WDM transmission system performance: Influence of non-Gaussian detected ASE noise and periodic DEMUX characteristic. W. Mar. S. New York: Wiley. and G. “Performance evaluation of edfa preamplified receivers taking into account intersymbol interference. Optical Fiber Communication Conf. E. Durhuus. pp.. “Detailed noise statistics for an optically preamplified direct detection receiver. Ben-Eli. Frignac. pp. Dallal. F. 2nd ed. N 2N 2 2 I. Xiaopin. 348–362. Roudas (M’00).. Select. [45] R. Herbst. W. Poehlmann. 15. 13. 1032–1034. [36] S. [62] I. Eur. Lösch.2 Tb/s 256 42. Lightwave Technol. Statistical Theory of Signal Detection. Detection. France. and Z. 1968. and T. Aug. Y. Lee and C. “Vestigial side band filtering for ultra high capacity (0. 1990. vol.. [54] C. 2nd ed.” J. Estimation and Modulation Theory. Courant and D. Root. Lightwave Technol. CA.28 b/s/Hz spectral efficiency. 10. Simonneau. P. vol. T. [32] J. Sept. L. pp. Tada. CA: Holden-Day. Sept. F. 1. W. R. SAC-2. Y. Inform. E. 13. New York: Wiley. W. Lenz. Sumida. G. Areas Commun. 505–513. Conf.” IEEE J. S. 14.” Electron. K. B. L. PD25.” IEEE Photon. S. Dragone. VOL. [52] M. no. Ishida and H. 1816–1823. 1224–1229. Castellani.” in Proc. Nice. JUNE 2002 [23] G. 1155–1163. Y. photograph and biography not available at the time of publication. Jourdan..64 bit/s/Hz) transmission of 128 40 Gb/s channels. D.” J. Wagner. 59–71. 2467–2474. Englewood Cliffs. Toba. Takahashi. R. [27] P. Lawetz and J. S.” J. 1992. A. Select. pp. no. Antoniades. B. 1967. 447–455. Fiber-Optic Communication Systems.. [56] H.” in Proc. pp. Stegun. L. Maeda. Lightwave Technol. W. Lett.” in Proc. Meissner. “Dispersion measurement of arrayed waveguide gratings by fourier-transform spectroscopy. “Bit-error-rate analysis of optically preamplified receivers using an eigenfunction expansion method in optical frequency domain.. Series. Lightwave Technol. Hamaide. 16. Mathematical Methods for Physicists. Jeruchim. 16. photograph and biography not available at the time of publication. Technol. [38] I. vol. vol. Mikkelsen. N. Tarrytown. T. Idler. Optical Communications. [41] E. Gradshteyn and I.. J. Eggleton. P. Borne. 10. [40] G. Sept. [46] G. [57] B. Takahashi. pp.” IEEE J. pp. and V. Dec. vol..” J. E. 756–757. no. Danielsen. vol. NJ: Prentice-Hall. A. Conf. R. no. “Derivation of analytical expressions for the bit-error probability in lightwave systems with optical amplifiers. pp. . Cella. 1998.. vol. Marcuse. [64] S. 11. Nov. R. Tran. Lightwave Technol. vol. no. Azizoglu. 1968. Elrefaie.. Stern (S’54–M’57–SM’67–F’72–LF’95). J. 1999. Teich..” Opt. [26] C. vol. Bayvel. E. IT-32. Proakis. [34] S. MM3.” J.” J. 9. 1997. New York: Academic..” J. NO. K.. S. “Performance of optically preamplified receivers with fabry-perot optical filters. “Transmission characteriswavelength multiplexer. Oct. pp. “Bit error evaluation of optically preamplified direct detection receivers with fabry-perot optical filters.. TX. N. 9. 1995. [33] L. Forestieri. [37] C. New York: Dover. J.. no.” J. Granlund. pp. 1986. [35] D. 12.. Pinto. 1994. Dischler. 3. “Effect of process variations on arrayed waveguide grating dispersion. Monroy and G. “Dispersion penalty measurements of narrow fiber Bragg gratings at 10 Gb/s. Apr. Abramowitz and I. Joergensen. C. NJ: IEEE Press. no. pp. New York: Wiley. 13. “Loss-imbalance equalization in arrayed-waveguide-grating (AWG) multiplexer cascades. June 1995. E. “Efficient techniques for widening the passband of a wavelength router.” J. New York: McGrawHill. C. Agrawal. P. photograph and biography not available at the time of publication. [58] J. no. Saleh and M. E. 1989. L. Ribeiro. S. 1991. 225–232. “Wavelength-domain simulation of multiwavelength optical networks (invited paper).. 1546–1553. 1999. Lightwave tics of arrayed-waveguide Technol. Veith.. 1991. Topics Quantum Electron. Lightwave Technol. Jan. G. 45. 8. vol. 1991. no. [65] Y. Shim. Lightwave Technol. 2001. H./Apr. Apr. Cartledge. 1972. 1985. A. Stubkjaer. Tables of Integrals. 2. 3rd ed. L. Park and S. Antoniades. 16. 6. [63] B. 1966. J. 1999. 1998. Lightwave Technol. photograph and biography not available at the time of publication. Inoue. [53] I. no. Franz. Nykolak. [43] W. I328–329. May 1995. vol. and Y. 1. Strasser. 9. Lett. and S. THK3. Bigo. 1576–1582. 11.” J. Lett. and S. [60] S. no. “Error-rate performance of optical amplifiers. and A. 5. E. Select. no. Oct. C. H. [28] D. Roudas. 1997. 1995. 2001. Golomb. vol. Richards. 13.