The wavelength division multiplexing passive optical

Transcription

1 Lee et al. VOL., NO. 6/JUNE 010/J. OPT. COMMUN. NETW. 381 Decision Threshold Control Method for the Optical Receiver of a WDM-PON Hoon-Keun Lee, Jung-Hyung Moon, Sil-Gu Mun, Ki-Man Choi, and Chang-Hee Lee Abstract We demonstrate an improvement of optical receiver performance for an optical signal with beat noises. The decision threshold level is adjusted automatically according to the detected average optical power. This simple method improves optical receiver performance under a wide range of input signal conditions, e.g., beat noise level and extinction ratio. Thus this proposed method is useful to optical receivers for WDM-PONs. By employing this control circuit, we successfully demonstrate a WDM-PON based on a wavelength-locked Fabry Perot laser diode at 1.5 GbÕs per channel with 100 GHz spacing. Index Terms Wavelength division multiplexing passive optical network (WDM-PON); Wavelengthlocked Fabry Perot lasers; Decision threshold control; Beat noise. I. INTRODUCTION The wavelength division multiplexing passive optical network (WDM-PON) has been extensively investigated for use as a next-generation access network. One of the important issues for WDM-POM is cost-effective colorless or color-free optical sources [1]. Recently, a wavelength-locked Fabry Perot laser diode (F-P LD) with an injected spectrum-sliced amplified spontaneous emission (ASE) light was proposed [], and the optical systems based on this optical source have already been commercialized [3,4]. It guarantees a 15 Mb/s data rate for both the upstream and downstream directions and accommodates 3 subscribers with 100 GHz channel spacing. The proposed optical source provides an attractive solution for WDM-PON due to its cost-effectiveness and colorfree characteristics. Unfortunately, the received optical signal contains optical beat noises induced by the Manuscript received January 5, 010; revised April 1, 010; accepted May 7, 010; published May 5, 010 Doc. ID 131. Hoon-Keun Lee, Jung-Hyung Moon, Sil-Gu Mun, and Chang-Hee Lee ( changheelee@kaist.edu) are with the Department of Electrical Engineering, Korea Advanced Institute of Science and Technology (KAIST), Daejon, , South Korea. Ki-Man Choi is with the Network Infra Research Depratment, Korea Telecom Network, Daejeon, , South Korea. Digital Object Identifier /JOCN injected spectrum-sliced ASE light [5,6]. Generally, the beat noises exist in many types of WDM-PONs due to optical backreflection [7,8], optical amplifiers [9], and spectrum filtering [10 1]. The optimum decision threshold level of the optical receiver with the beat noise is a function of the incident optical power, since the noise power depends on the signal optical power. Therefore we need to adjust the decision threshold level according to the noise level or the optical power. Until now, various optical receivers that could control the decision level have been proposed [13 15]. However, these receivers are not suitable for the access network because of the relatively high cost and the complexity of the receiver configurations. In this paper, we demonstrate a simple decision threshold control circuit that can adjust the decision threshold level automatically according to the detected average optical power. The proposed method can be applicable for a wide range of signal conditions, such as the beat noise level and extinction ratio (ER). By using this method, we demonstrate the improvement of the optical receiver performance for a WDM- PON based on the wavelength-locked F-P LD. This paper is organized as follows. In Section II, the noise characteristics of the wavelength-locked F-P LD are presented with the simulation results, and the configuration and operation of the decision threshold control circuit are given. Then, we present the experimental results in Section III. An application of the proposed optical receiver to a WDM-PON is presented in Section IV. Finally, the conclusion of this paper is drawn in Section V with discussion. II. THEORY A. Noise Characteristics of the Wavelength-Locked F-P LD In optical systems based on wavelength-locked F-P LDs, the 1 level noise is larger than the 0 level noise due to an ASE-ASE beating noise generated by the spectrum-sliced ASE light. Then, the optimum decision threshold level is lower than that of the thermalnoise-limited systems and decreases as the detected /10/ /$ Optical Society of America

2 38 J. OPT. COMMUN. NETW. / VOL., NO. 6/ JUNE 010 Lee et al. average power increases [16,17]. This optimum decision threshold level can be expressed as I th = 0I I , 1 where I th is the decision threshold level, and 0 and 1 are the standard deviation of noise for the 0 and 1 levels, respectively. I 0, I 1 represent the output current level of the 0 and 1 levels. Because the dominant noise factors of the wavelength-locked F-P LD system are thermal noise and beating noise [in this case the relative intensity noise (RIN)], 0 and 1 can be expressed by 0 = thermal = einc B, 1 = thermal + beating, where einc is the equivalent input noise current, and B is the electrical bandwidth of the receiver. Here, assuming the RIN value is x 0 db/hz, the beat noise is given by [18] beating = R P 1 10 x/10 B, P 1 = 1+ P rec, P 0 = 1+ P rec, where R is the receiver responsivity, and P 1, P 0 are the 1 level and 0 level optical powers, respectively. =P 0 /P 1 and P rec are the inverse of the ER of the optical signal and the received average optical power, respectively. From the above Eqs. (1) (5), we know that the optimum decision threshold level is changed according to the RIN, the ER, and the optical received Fig. 1. (Color online) Optimum decision threshold levels for WDM- PON based on the wavelength-locked F-P LD. The character D represents the decision threshold level of the proposed decision threshold control circuit TABLE I SIMULATION PARAMETERS a Symbol Parameter Value P rec Received power 3 to dbm R Responsivity 0.8 A/ W einc Equivalent input noise current 3.4 pa/ Hz Extinction ratio 10 db RIN Relative intensity noise Best: 11.5 db/ Hz Worst: db/ Hz a The extinction ratio and relative intensity noise are based on the measured value. power. Here, the RIN value is determined by a detuning and an ASE injection power. The detuning means the wavelength difference between the spectrumsliced ASE and the lasing mode of the F-P LD (i.e., ASE F-P LD )[19]. To investigate the noise characteristics of the wavelength locked F-P LD, we simulated the optimum decision threshold level as a function of the received power as shown in Fig. 1. For this simulation, we used measured RIN values for the noise model of the wavelength-locked F-P LD output. The best case was observed when the injection wavelength matched to one of the lasing wavelengths of the F-P LD, while the injection wavelength was close to the center of the two neighborhood modes for the worst case. The parameters used in this simulation are shown in Table I. It may be noted that the optimum decision threshold level is positioned at the center of the 1 level and the 0 level for a thermal-noise-limited case as shown in the dotted dashed line with A. However, it decreases as the received optical power increases for the wavelength-locked F-P LD cases [ B, C ]. Furthermore, the optimum level of the worst case C is positioned at a lower level than that of the best case B. This is because of the increased intensity noise by the detuning. Then, an optimization method was proposed to handle this complicated noise characteristic more easily and effectively [16,17]. Since the overall system performance is determined by the worst case of the receiver sensitivity, it may be possible to enhance the total system performance by following the decision threshold level of the worst case. However, as seen in the decision level of C, the decision threshold level is not a linear function of the received input power. For simple realization, we propose a linear change of the decision threshold as indicated by the decision level of D as a function of the received input power. It may be noted that the receiver performance can be enhanced further, if we follow the decision level of C more closely with a log amplifier in the decision control circuit.

3 Lee et al. VOL., NO. 6/JUNE 010/J. OPT. COMMUN. NETW. 383 B. Configuration and Operation of the Decision Threshold Control Circuit Based on the above observations, we designed and implement the decision threshold control circuit. The configuration of the optical receiver with the decision control circuit is shown in Fig.. It consists of a conventional receiver part, a decision threshold control part, and a power monitoring part. The conventional receiver part comprises a PIN photodiode (PD), a transimpedance amplifier (TIA), a limiting amplifier (LA), and clock and data recovery (CDR). The decision threshold control circuit is composed of an electrical linear amplifier (K), an adder, a dc voltage control circuit (VC), and a low-pass filter (LPF). The resistor (R) is used to check the received average power of the optical receiver as a monitoring part. The operation for the decision threshold control circuit is as follows: First, the transmitted signals from the optical transmitter are detected by the PIN PD. At the same time, the received average optical power is monitored by the resistor (R). The monitored voltage is amplified by the difference linear amplifier (K) to determine the amount of the decision threshold level to be changed. The amplified signal V o1 and a reference dc voltage V DC are added by using an adder. This reference voltage is used to set a reference level of the decision threshold. The output of the adder V o is then used to change the decision threshold level after passing through the LPF. As a result, the decision threshold level V th is controlled automatically according to the received average optical power. The proposed scheme can be compared with a wellknown automatic gain-control (AGC) circuit. The AGC was developed to provide a constant input to the limiting amplifier and/or decision circuit [0]. Thus it does not a have decision threshold control function. C. Simulation Results To estimate the receiver performance with the proposed decision threshold control method, we plot the bit error rate (BER) curves by calculating the Q factors of the 0 and 1 levels. The Q factors for the 0 level Q 0 and the 1 level Q 1 are given by Q 0 = I th I 0 0, Q 1 = I 1 I th 1. 6 By assuming Gaussian noise distribution, the BER is given by BER = 1 4 erfc Q erfc Q exp Q 0 Q exp Q 1 Q 1. Figure 3 shows the simulation results of the receiver performance according to the decision threshold levels. The characters A D represent the decision threshold level conditions in Fig. 1. First, we investigate the enhancement of the system performance with the optimum decision threshold level of curve C. These results are shown in Fig. 3(a) with the conven- 7 Fig.. (Color online) Configuration of the optical receiver with the automatic decision threshold control circuit. Fig. 3. (Color online) Simulation results for the BER performance according to the decision threshold levels for the best case and the worst case. A D represent the decision level conditions in Fig. 1.

4 384 J. OPT. COMMUN. NETW. / VOL., NO. 6/ JUNE 010 Lee et al. tional receiver case (without the decision control) for comparison. The receiver performance is improved dramatically as seen in the simulation result. However, for the simple realization of the decision control circuit, we used a linear curve as a function of the received average input power [decision level: D ]. These BER results of the best case and the worst case are shown in Fig. 3(b). The performance of the receiver is also improved considerably, although a linear approximation causes a performance degradation compared with the optimal decision control level case [decision level: C ]. However, it is possible to have error-free transmission with the proposed decision threshold control method. A. Experimental Setup III. EXPERIMENTS Figure 4 shows an experimental setup based on the wavelength-locked F-P LD to investigate the performance of the decision threshold control circuit. A broadband light source (BLS) was an ASE light generated by a pumped erbium-doped fiber (EDF). The BLS output was spectrum sliced by an arrayed waveguide grating (AWG) and injected into the F-P LD located at the optical network unit (ONU) via an optical circulator. The ASE power injected into the F-P LD was about 13.5 dbm/0. nm (total power: 7 dbm) at the peak wavelength (channel number: 5, center wavelength: nm). Channel 5 had the lowest ASE injection power among the 16 channels. Then the output signal of the wavelength-locked F-P LD was sent to AWG located at the central office (CO) after passing through AWG1. The channel spacing and its 3 db bandwidth were 100 GHz and 0.61 nm, respectively. We used a flattop passband AWG to reduce the beating noise of the input ASE light [1]. The F-P LD was directly modulated at 1.5 Gb/ s [pseudorandom binary sequence (PRBS): 31 1] with non-return-tozero (NRZ) data. The measured extinction ratio was about 10 db. The mode spacing and front facet reflectivity of the LD were about 0.57 nm and 0.1%, respectively. The modulation current of the F-P LD was.0 I th, where I th is the lasing threshold current. We measured BER curves for the best case 48 C and the worst case 45 C within one mode spacing of the F-P LD (45 C to 51 C) to investigate color-free operation. The BER performance is estimated according to the decision threshold levels defined in Fig. 1. B. Experimental Results The experimental results in Fig. 5 show good correspondence with the simulation results. The solid triangle ( ) and the solid circle ( ) show the BER performance with the conventional receiver [decision level: A ] for the best case and the worst case, respectively. In these cases, we could not achieve an errorfree transmission. However, when the decision level was controlled with the optimal decision threshold level of the curve C, we got an error-free transmission within the one mode spacing of the F-P LD (best case:, worst case: ) as shown in Fig. 5(a). We also measured the BER curves with the proposed control circuit [decision level: D ]. These results are shown in Fig. 5(b). Using the linear approximation with the decision control circuit caused power penalties of 1.3 and 0.6 db in comparison with the optimal decision threshold level case for the best case and the worst case, respectively. However, if we consider the total system performance, the degradation due to the linear Fig. 4. (Color online) Experimental setup with the proposed optical receiver. Fig. 5. (Color online) Experiment results for the BER performance according to the decision threshold levels for the best case and the worst case. The inset shows the measured eye diagram of the worstcase optical signal for the back-to-back condition.

5 Lee et al. VOL., NO. 6/JUNE 010/J. OPT. COMMUN. NETW. 385 approximation is small. The inset shows the measured eye diagram of the worst-case optical signal at 10 dbm. As expected, the 1 level noise is broader than the 0 level noise because of the beat noise. As shown in Section II, the optimum decision threshold level is changed according to the average received power and the input signal conditions (such as RIN or ER). To investigate this relationship, we simulated the optimum decision threshold levels in accordance with the variation of ER and RIN as a function of average received power in Figs. 6(a) and 6(b), respectively. The reference means the decision threshold level of the control circuit [decision level: D ] atthe signal condition with an ER of 10 db and a RIN value of db/hz. As shown in Fig. 6(a), the optimum decision threshold level is increased (or decreased) from the reference as the ER becomes low (or high) while maintaining the RIN level of db/hz. In contrast with the ER, for a case of the RIN, the optimum decision threshold level is decreased (or increased) as the RIN level becomes high (or low) while Fig. 7. (Color online) Power penalty contours as a function of the signal condition at 1.5 Gb/ s. The symbols represent the experimental results and the solid lines show the simulation results. maintaining the ER of 10 db as shown in Fig. 6(b). Unfortunately, the variation of the input signal conditions (RIN or ER) causes a power penalty resulting from the change of the optimum decision threshold level. Thus, we calculated the power penalties at a given BER by comparing them with that of the reference. Figure 7 shows the power penalty contours as a function of the signal conditions. The decision threshold level was adjusted according to the received average input power by following the line D in Fig. 1. The symbols and lines represent the measured and calculated data at the BER, respectively. As expected, the BER performance is degraded as the signal conditions depart from the references. However, the power penalty can be negligible when the signal condition is better than the reference condition (i.e., RIN db/hz and ER 10 db). If we accept 1 db power penalty, the receiver can be operated when the RIN is less than 109 db/hz with the ER higher than 8.7 db. We also investigate the power penalty contours with 0 km of transmission. The changes of signal conditions after transmission were +0. and 0.3 db for the RIN and the ER, respectively. These changes induce very small penalty as shown in Fig. 7, when we use the receiver with the decision control circuit that the decision threshold level is set at the back-to-back reference signal conditions. It may be noted that the measured dispersion penalty was less than 0.3 db after 0 km transmission. IV. APPLICATION TO A WDM-PON SYSTEM Fig. 6. (Color online) Optimum decision threshold levels according to variation of (a) ER and (b) RIN. The reference indicates the decision threshold level of D in Fig. 1. To demonstrate performance improvement in a WDM-PON with the proposed receiver, we implement a high-speed WDM-PON based on the wavelengthlocked F-P LDs. It consists of 16 channels with the wavelength regions from to nm. We transmitted five channels with 100 GHz channel spac-

6 386 J. OPT. COMMUN. NETW. / VOL., NO. 6/ JUNE 010 Lee et al. ing. The transmission data rate was 1.5 Gb/ s per channel. The temperature of the F-P LD was maintained around 43 C by using a heater. Figure 8 shows the measured BER curves after passing through 0 km of single-mode fiber (SMF). The same optical receiver was used to measure the BER for all channels. The hollow black circle ( ) shows the BER curve without the control circuit at channel 5. It shows an error floor at the BER of However, all the measured BER curves show error-free transmission with the help of the decision control circuit. The sensitivity difference between channels is less than 0.8 db. This difference may be a result of the various factors such as the ASE injection power, the injection position with respect to the F-P LD s lasing mode, the front facet reflectivity of the F-P LDs, and so on. The inset shows the measured spectra of the five channels. The spectra show different shapes resulting from the different ASE injection power and the detuning. The ASE injection power was varied from 13.5 to 11.8 dbm/0. nm channel by channel according to the different insertion loss of the AWG and the spectral profile of the BLS output. The ER of each channel was more than 10 db. A. Discussion V. DISCUSSION AND CONCLUSION We also investigate the application of the proposed decision threshold control method at a higher data rate of 10 Gb/s. The simulation result of the decision threshold control is shown in Fig. 9. In this simulation, we assume an einc of 1.5 pa/ Hz, an ER of 10 db, and a RIN of 10 db/hz. The curve B indicates the optimum decision threshold level of the transmitted signal condition and the line C means the decision threshold level to be set by the control circuit. Based on the decision level of the line C, we also plotted the power penalty contours according to the variation of the signal conditions in Fig. 10. These Fig. 9. (Color online) Optimum decision threshold level at 10 Gb/s. The character C represents the decision threshold level of the proposed control circuit. contours imply that a simple control method can be applied to the wide range of the signal conditions and various bit rates. However, the total area of the 1 db power penalty is reduced because of the increased data rate. For example, if we consider the higher RIN level from the reference at the ER of 10 db, the acceptable RIN variation range is about. db at 1.5 Gb/s. However, at 10 Gb/s transmission, the acceptable RIN variation range is reduced to 1.4 db. In the cases of the ER, the acceptable ER range is about 1.3 db (at 1.5 Gb/s) and 1.5 db (at 10 Gb/s) from the reference at the RINs of 111 and 10 db/hz, respectively. The RIN level of 10 db/hz can be obtained by using the noise suppression method []. However, we need dispersion compensation either in the optical domain or in the electrical domain for 10 Gb/ s transmission. The transmission performance of a WDM-PON using a seed light is not only degraded by signal conditions (such as RIN and ER), but also optical backre- Fig. 8. (Color online) BER curves of five upstream channels with the decision threshold control circuit. The inset shows the measured spectra of five channels. Fig. 10. (Color online) Power penalty contours as a function of the signal condition at 10 Gb/s.

7 Lee et al. VOL., NO. 6/JUNE 010/J. OPT. COMMUN. NETW. 387 flection [7,8]. The beat noise resulting from the backreflection can be given by [7] E reflection = R P E 1 E P E P 3 1 n 1 10 n 1 G+R /10 n=1, where E 1, E, and E 3 are the electrical fields of the signal, the reflected field by reflection-i (backreflection of the injected seed light to the CO), and the reflected field by reflection-ii (backreflection of the modulated upstream signal to the ONU), respectively. P and P 3 are the normalized power spectral density of the beating noise by reflection-i and reflection-ii within the receiver bandwidth. G is the ONU gain and R is the optical return loss. From Eq. (8), the 1 level noise increases as the received optical power increases. This means that the optimum decision threshold level is also decreased with a slope according to the received power. This slope is determined by the beat noise resulting from reflection-i and reflection-ii. It may be noted that the noise distribution resulting from the backreflection can be approximated to the Gaussian distribution [3]. Thus, the decision threshold control method can be applied to a WDM-PON based on the seed light injection to mitigate the backreflection effects. B. Conclusion In conclusion, we proposed and demonstrated a simple decision threshold control method according to the detected average optical power. It improves optical receiver performance in a wide range of input signal conditions, such as the beat noise level (RIN) and ER. The proposed method can be useful for a low-cost optical receiver with optical signals containing a backreflection-induced noise, a signal-ase beat noise, and an ASE-ASE beat noise, etc. We have successfully demonstrated the enhancement of the receiver performance for the WDM-PON based on the wavelengthlocked F-P LD at 1.5 Gb/s transmission with the proposed decision threshold control method. REFERENCES [1] C.-H. Lee, W. Sorin, and B. Y. Kim, Fiber to the home using a PON infrastructure, J. Lightwave Technol., vol. 4, no. 1, pp , Dec [] H.-D. Kim, S.-G. Kang, and C.-H. Lee, A low-cost WDM source with an ASE injected Fabry-Perot semiconductor laser, IEEE Photon. Technol. Lett., vol. 1, no. 8, pp , Aug [3] H. J. Park, H. S. Yoon, T. S. Park, S. J. Park, and J. H. Kim, Recent research activities of WDM-PON in Korea, in Optical Fiber Communication Conf. and the Nat. Fiber Optics Engineers Conf., Anaheim, CA, 007, paper OWL1. 8 [4] C.-H. Lee, S.-M. Lee, K.-M. Choi, J.-H. Moon, S.-G. Mun, K.-T. Jeong, J. H. Kim, and B. Kim, WDM-PON experiences in Korea [Invited], J. Opt. Netw., vol. 6, no. 5, pp , Apr [5] X. Cheng, Y. J. Wen, Y. Dong, Z. Xu, X. Shao, Y. Wang, and C. Lu, Optimization of spectrum-sliced ASE source for injectionlocking a Fabry-Pérot laser diode, IEEE Photon. Technol. Lett., vol. 18, no. 18, pp , Sept [6] M. Fujiwara, H. Suzuki, K. Iwatsuki, and M. Sugo, Noise characteristics of signal reflected from ASE-injected FP-LD in loopback access networks, Electron. Lett., vol. 4, no., pp , Jan [7] J.-H. Moon, K.-M. Choi, S.-G. Mun, and C.-H. Lee, Effects of back-reflection in WDM-PONs based on seed light injection, IEEE Photon. Technol. Lett., vol. 19, no. 4, pp , Dec [8] M. Fujiwara, J. Kani, H. Suzuki, and K. Iwatsuki, Impact of backreflection on upstream transmission in WDM single-fiber loopback access networks, J. Lightwave Technol., vol. 4, no., pp , Feb [9] P. C. Becker, N. A. Olsson, and J. R. Simpson, Erbium-Doped Fiber Amplifiers. San Diego, CA: Academic Press, [10] A. D. McCoy, B. C. Thomsen, M. Ibsen, and D. J. Richardson, Filtering effects in a spectrum-sliced WDM system using SOA-based noise reduction, IEEE Photon. Technol. Lett., vol. 16, no., pp , Feb [11] J. S. Lee, Y. C. Chung, and D. J. DiGiovanni, Spectrum-sliced fiber amplifier light source for multichannel WDM applications, IEEE Photon. Technol. Lett., vol. 5, no. 1, pp , Dec [1] H.-K. Lee, H.-J. Lee, and C.-H. Lee, A simple and color-free WDM-passive optical network using spectrum-sliced Fabry- Pérot laser diodes, IEEE Photon. Technol. Lett., vol. 0, no. 3, pp. 0, Feb [13] Y. Matsumoto, T. Kuriyama, D. Inami, and M. Ohta, An adaptive decision threshold control of the optical receiver for multigigabit terrestrial DWDM transmission systems, in Optical Fiber Communication Conf. and the Nat. Fiber Optic Engineers Conf., Anaheim, CA, 001, paper TuR. [14] M. Kawai, H. Watanabe, T. Ohtsuka, and K. Yamaguchi, Smart optical receiver with automatic decision threshold setting and retiming phase alignment, J. Lightwave Technol., vol. 7, no. 11, pp , Nov [15] I. Hatakeyama, T. Nagahori, K. Miyoshi, Y. Nukada, T. Shine, T. Watanabe, A. Uda, and K. Shiba, A 1-channel data-formatfree 1 Gb/s/ch parallel optical receiver, in IEEE Int. Solid- State Circuits Conf., San Francisco, CA, 1999, paper WP.. [16] S.-G. Mun and C.-H. Lee, An optimization method for the decision threshold level in optical receivers for WDM-PONs, Proc. SPIE, vol. 6354, paper 63549, 006. [17] H.-K. Lee, J.-H. Moon, S.-G. Mun, K.-M. Choi, and C.-H. Lee, An automatic decision threshold control circuit for WDM-PON based on the wavelength-locked F-P LDs, in Proc. 14th Opto- Electronics and Communication Conf., Sydney, Australia, 008, paper ThL-4. [18] J. C. Palais, Fiber Optic Communications, 5th ed. Upper Saddle River, NJ: Pearson Prentice-Hall, 005. [19] S.-M. Lee, M.-H. Kim, and C.-H. Lee, Demonstration of a longreach DWDM-PON for consolidation of metro and access networks, J. Lightwave Technol., vol. 5, no. 1, pp , Jan [0] D. K. Mynbaev and L. L. Scheiner, Fiber-Optic Communications Technology. Upper Saddle River, NJ: Prentice-Hall, 001. [1] K.-Y. Park, J.-S. Baik, T.-W. Oh, and C.-H. Lee, Intensity noise suppression and 1.5 Gb/ s transmission using a wavelength locked Fabry-Perot laser diode with filtered ASE injection, in Proc. 10th OptoElectronics and Communication Conf., 004, paper 13P-44, pp

8 388 J. OPT. COMMUN. NETW. / VOL., NO. 6/ JUNE 010 Lee et al. [] J.-S. Jeong and C.-H. Lee, Optical noise suppression techniques for wavelength-locked Fabry-Perot laser diode, in Proc. of the 15th Asia-Pacific Conf. on Communications, Shanghai, China, 009, paper 14. [3] C. J. Rasmussen, F. Liu, R. J. S. Pedersen, and B. F. Jorgensen, Theoretical and experimental studies of the influence of the number of crosstalk signals on the penalty caused by incoherent optical crosstalk, in Optical Fiber Communication Conf., Anaheim, CA, Feb. 1999, paper TuR5 1. Hoon-Keun Lee received the B.S. degree in electronics from Kyungpook National University, Daegu, South Korea, in 006. He is currently in an M.S. Ph.D. joint program at Korea Advanced Institute of Science and Technology (KAIST), Daejeon, South Korea. His research interests include lightwave systems and optical access networks based on WDM-PON. Jung-Hyung Moon received the M.S. degree from the Department of Electrical Engineering, Korea Advanced Institute of Science and Technology (KAIST), Daejeon, South Korea, in 010. His research interest includes lightwave systems and optical access networks based on WDM-PON. Sil-Gu Mun received the M.S. and Ph.D. degrees from the Department of Electrical Engineering, Korea Advanced Institute of Science and Technology (KAIST), Daejeon, South Korea, in 005 and 010, respectively. She is currently working at KAIST as a Postdoc. Her research interest includes lightwave systems and high-speed optical access networks. Ki-Man Choi received the M.S. and Ph.D. degrees from the Department of Electrical Engineering, Korea Advanced Institute of Science and Technology (KAIST), Daejeon, South Korea, in 004 and 008, respectively. She spent 6 months at KAIST as a Postdoc. She has been working as a Researcher at the Network Infra Research Department, Korea Telecom Network. Her current research interests are WDM-PON and network management. Chang-Hee Lee received the M.S. and Ph.D. degrees from the Department of Electrical Engineering, Korea Advanced Institute of Science and Technology (KAIST), Daejon, South Korea, in 1983 and 1989, respectively. He spent a year at Bellcore (Bell Communications Research) as a Postdoc. He worked with the Electronics and Telecommunications Research Institute from 1989 to 1997 as a Senior Researcher. Since 1997, he has been a Professor with KAIST. His major interest is optical communications and networks. He has spent over 0 years as an engineer in the area of optical communications, including semiconductor lasers. He was Technical Leader for the.5 Gb/ s and 10 Gb/ s optical-transmission-system development, including optical amplifiers at the Electronics and Telecommunications Research Institute. He is the author of over 00 journal and conference papers. He is the holder of 8 U.S. patents and more than 60 additional patents are pending in the U.S. Prof. Lee is a fellow of IEEE and a member of OSA.