Enhanced Reflection Tolerance of Upstream Signal in a RSOA-based WDM PON by using Manchester Coding

Transcription

1 Enhanced Reflection Tolerance of Upstream Signal in a RSOA-based WDM PON by using Manchester Coding A.Murakami, Y.J.Lee, K.Y.Cho, Y.Takushima, A.Agata, K.Tanaka, Y.Horiuchi, Y.C.Chung : KDDI R&D Laboratories Inc., Ohara, Fujimino-shi, Saitama , Japan : Korea Advanced Institute of Science and Technology Guseong-dong, Yuseong-gu, Daejeon 35-71, Korea ABSTRACT We have experimentally investigated the reflection tolerance of upstream signals by comparing Manchester-encoded downstream signals with NRZ ones in a RSOA-based fiber loop back system for a WDM PON. The results showed that the reflection tolerance of the upstream signals strongly dependeds on the downstream signal modulation formats, and that the Manchester format was more tolerant than the NRZ one against the reflection of the upstream signal. Keywords: WDM PON, RSOA, reflection tolerance, Manchester INTRODUCTION The advent of video-related applications has resulted in a tremendous increase in capacity demand throughout the core, metro and access networks. For the access network, wavelength division multiplexed passive optical networks (WDM PON) have been considered as an attractive solution for future optical access networks, because of its large capacity and high security [1,]. In a WDM PON, the colorless light source is a key component to achieve low installation cost. Several colorless light sources have been proposed, such as spectrum-sliced incoherent light sources, amplified spontaneous emission (ASE)-injected Fabry-Perot lasers, and re-modulated light sources [3-7]. A re-modulated light source that has recently received a lot of attention is the reflective semiconductor optical amplifier (RSOA). This is due to the fact that the ASE-injected Fabry-Perot lasers need a high power external ASE light source to lock lasing wavelength of Fabry-Perot laser. In contrast, the re-modulated light source can generate upstream signal by re-using downstream signal, thus any ASE light source is not necessary [7]. However, the main drawback of using a re-modulated light source in a WDM PON is that optical beat interference (OBI) between the upstream signal and the interfering signals (back reflected and re-amplified upstream signals) induces a severe deterioration of the upstream performance. It has been shown that OBI is especially problematic in networks containing many reflection points (e.g. fiber connectors) which cause high back reflections [8, 9]. We show the effectiveness to apply the Manchester coding in the downstream direction in order to mitigate the effect of OBI due to reflection in a WDM PON. It has been studied that the downstream residual modulation noise components in the upstream signals could be suppressed by using Manchester instead of NRZ coding in the downstream direction [7]. This was due to the reduction of the power spectral density in the case of the Manchester encoded signal. For the NRZ signal, main lobe is present at low frequency, whereas for the Manchester signal the main lobe is located at a frequency near the data rate. As a result, the optical bandwidth of the Manchester-encoded signal spreads more widely than the NRZ signal. Thus, the Manchester format is expected to decrease the effect of OBI due to reflection. In this paper, we experimentally investigate the effectiveness of Manchester coding for the suppression of OBI in a RSOA-based fiber loop back system for a WDM PON. Comparing the reflection tolerance of the upstream signals for both the NRZ and Manchester-encoded downstream signals, we show that the degradation of the upstream performance can be drastically alleviated by using the Manchester coding in the downstream direction. Optical Transmission, Switching, and Subsystems, edited by Dominique Chiaroni, Wanyi Gu, Ken-ichi Kitayama, Chang-Soo Park, Proc. of SPIE ol. 6783, 6783I, (7) X/7/$18 doi: / Proc. of SPIE ol I-1

2 EXPERIMENTAL SETUP Fig. 1 shows the schematic diagram of OBI mechanism for a RSOA-based WDM PON with a discrete reflection point. In this study, we focused on the OBI of the upstream signals from a RSOA-based ONU to central office (CO). A portion of upstream signal light is reflected at the reflection point and goes back to the RSOA. Then, the reflected signal is reamplified by the RSOA and interfered with the originally generated upstream signal. As a result, the interference induces the OBI noise in the upstream direction. To investigate the reflection tolerance of the upstream signals for WDM PON systems using an RSOA, we implemented a fiber loopback system and measured the power penalty of the upstream signal induced by the back reflection. Fig. shows the experimental setup used for the power penalty measurement. In the downstream direction, a DFB laser diode (DFB-LD) directly modulated at a bit rate of 1.5 Gb/s or.5 Gb/s was used as an optical transmitter at a CO. The optical signal wavelength was nm, and the modulation format was non-return-to-zero (NRZ) or Manchester format ( 31-1 PRBS). The extinction ratio of the downstream signal was set to db in both formats. A polarization controller (PC) following the DFB-LD was used so as to maximize the RSOA gain at an ONU. The downstream signal power at the ONU corresponding to the input power to the RSOA was set to -9.8 dbm to attain 3 db gain compression of the RSOA. In the upstream direction, the downstream signal was re-modulated by the RSOA with a format of 6 Mb/s NRZ ( 31-1 PRBS). The extinction ratio of the upstream signal was set to 1 db. A reflection module was connected to the transmission line between the CO and ONU through a 3 db optical coupler to create back reflection light. Optical return loss was defined as the ratio of the power between the originally generated upstream signal and the reflected and re-amplified signals, as described (a) and (b) in Fig.. The module consisted of a PC, a variable optical attenuator (OA) to adjust the optical return loss in the range from 6 to 4 db and a mirror. We tuned the PC in such a way that the bit error rate (BER) was worst in order to maximize the effect of reflection. The fiber length between the ONU and mirror was around 15 m. At the receiver in the CO, the ASE noise generated by the RSOA was cut by an optical bandpass filter (OBPF) with a bandwidth of nm. CO LGGCG ruq LG-iwbjqq nblcru 1411 I LGUGCJOU bo Fig.1 Schematic diagram of OBI mechanism for RSOA-based WDM PON with discrete reflection. CO qoiulg1m i.ruuj!ugl I4K C I/\I1UCIJGGL DEThF[ I GP OL 3RG J'1K /JfJJC}JGG bc AO () (p) 141 icov DI 1K nblgruj LG(GJAGL ojj moqnj icx IJ l3bf Fig. Experimental setup. Proc. of SPIE ol I-

3 EXPERIMENTAL RESULTS First, we set the downstream bit rate to.5 Gb/s and evaluated the reflection tolerance of the upstream signals. Fig.3 (a) and (b) show the obtained upstream BER performance for different optical return losses for the downstream signal formats of NRZ and Manchester, respectively. Closed squares show the BER without reflection and open circles, triangles, crosses and reversed open triangles show the BER at optical return losses of 4, 35, 3 and 6 db, respectively. From these results, we found that the reflection tolerance of the upstream signals was highly dependent on the downstream modulation format, and that the upstream signals re-modulated with the Manchester-encoded downstream signals had much higher tolerance to the back reflection than the NRZ-encoded downstream signals. For example, the receiver sensitivity at a BER of 1-9 started to deteriorate at the optical return loss of 35 db in the case of the NRZencoded downstream signals, whereas the power penalty was negligible even at the optical return loss of 35 db in the case of using the Manchester-encoded downstream signal. Next, we set the downstream bit rate to 1.5 Gb/s and evaluated the reflection tolerance of the upstream signals to clarify the difference of downstream bit rate. Fig.4 (a) and (b) show the measured upstream BER characteristics for the downstream signal formats of NRZ and Manchester, respectively. In the same way as the.5 Gb/s downstream case, the upstream signals re-modulated from the Manchester-encoded downstream signals had much higher back reflection tolerance than the NRZ-encoded downstream signals. C ç- 9-, 8-6 OT- ZE 9 a 'I'll','''' E aa 5 s 8Z papo; aaaaaaaa. 4 9Z A RPO%,:SSOIWfllOJI)!}dO U fl ZZ 1;Mod (uzp) Hp9ZSSOIwJdo po :ssoiwo[!ldo O Iv I I I I I I I I I I I I e pticietion:16thl pi [th!t 35dB o Received power (dbm) (a) (b) Fig.3 BER performance of upstream signal for the.5gb/s downstream signals of (a) NRZ and (b) Manchester. -3 = r u- ZE I I v... I E paat3 A. spo:iw!i!ith spcr:im],th clpot:s"1wfl9jlfldo Z 9Z t'z- U- o- 1;Mod (wap) opt!ietu Ios 1dB pnim I 3dB 4 v a OpaInTOSS.35dB -1 I * 1 v I Received power (dbm) (a) (b) Fig.4 BER performance of upstream signal for the 1.5Gb/s downstream signals of (a) NRZ and (b) Manchester. Proc. of SPIE ol I-3

4 Fig.5 shows the induced power penalties of the upstream signals at a BER of 1-9, as a function of the optical return loss. Squares and circles show the power penalties with using 1.5 Gb/s NRZ and Manchester formats as downstream signals, respectively. Triangles and crosses show the power penalties with using.5 Gb/s NRZ and Manchester formats, respectively. As shown in Fig.5, the power penalties of the upstream signal caused by the back reflection were mitigated by using Manchester-encoded downstream signals. In the case of.5 Gb/s downstream, the upstream power penalty was 6.5 db at an optical return loss of 3 db by using NRZ downstream signal. In contrast, the penalty decreased to 1.6 db by using Manchester-encoded downstream signal. The reflection tolerance of the upstream signals significantly depended on the downstream modulation formats, and the difference of the power penalty was as large as about 5 db at an optical return loss of 3 db between Manchester and NRZ downstream formats. In addition, when the downstream signal bit rate was 1.5 Gb/s, at an optical return loss of 3 db, the upstream power penalty was 8.1 db and 1. db with using NRZ and Manchester downstream signals, respectively. The difference of the power penalty was about 7 db at an optical return loss of 3 db between Manchester and NRZ downstream formats. The main reason of these power penalty variations is attributed to the RF spectral shape of the downstream signal. Fig.6 shows the RF spectra of NRZ and Manchester coded signals with the bit rate of 1.5 Gb/s and.5 Gb/s. As the bit rate of downstream signal decreased from.5 Gb/s to 1.5 Gb/s, the main peak of RF spectral components shift toward the lower frequency region. Thus, the RF spectrum of 1.5 Gb/s Manchester signal has larger amount of components in lower frequency region compared to that of.5 Gb/s Manchester signal. However, when 1.5 Gb/s Manchester signal was applied in the downstream direction, the upstream power penalty performance was not degraded by comparison with the power penalty which was obtained in the case of.5 Gb/s. t=(vtkf TO Q DT3GP I4B D:T3GP t\ifucpggl D3RGP J4W D3RGP Jt\Ih11CJJGG 3 C 3 3 (JulJcJ] 4U1J '\IB) : qoiiulgjw Fig.5 Power penalty of upstream signals at a BER of 1-9 as a function of optical return loss. I I3 UP 5R UP J4K JJG4GL FJG4GL pjjqthq!agl LLd11UC) ((}H) Fig.6 Measured RF spectra of 1.5Gb/s and.5gb/s NRZ and Manchester. Proc. of SPIE ol I-4

5 As shown in Fig.5, the upstream power penalty induced by the back reflection could be drastically mitigated by using the Manchester-encoded downstream signal, in comparison with the NRZ-encoded case, for both 1.5 Gb/s and.5 Gb/s downstream cases. This was mainly due to that spectral density of OBI depended on the downstream modulation format. In comparison with the RF spectra of NRZ coded signal at the same bit rate, the majority of the RF-spectral components of the Manchester-encoded signal were shifted toward the higher frequency region, and the optical bandwidth was much wider than that of the NRZ signal. Thus, the spectrum of the OBI noise was also spread widely when the downstream light was modulated in Manchester code or in high bit rate. However, the 6 Mb/s upstream receiver effectively cut off the OBI noise components in the higher frequency region. In the case of Manchester coded signal, the low-frequency components which were within the bandwidth of the 6 Mb/s upstream receiver were substantially suppressed. As a result, we could drastically improve the back reflection tolerance of the upstream signals when Manchester-encoded signal was applied in the downstream direction instead of the NRZ-encoded signal. CONCLUSION We have experimentally evaluated the reflection tolerance of the upstream signals by using the Manchester-encoded downstream signals in comparison with the NRZ ones in a fiber loopback system with RSOA-based ONU for a WDM PON system. We have confirmed that the Manchester coding was much more effective than the NRZ one in terms of back reflection tolerance. This is mainly due to the fact that the majority of the RF-spectral components of the Manchester-encoded signal were shifted toward the higher frequency region unlike the NRZ-encoded signal. Therefore, the spectrum of the OBI noise was also spread widely and 6 Mb/s upstream receiver cut off the OBI components in higher frequency region. REFERENCES S. Han, T.Y. Kim and C.S. Park, Direct-Modulated Upstream Signal Transmission Using a Self-Injection Locked F-P LD for WDM-PON, in Proc. European Conference on Optical Communication (ECOC), We3.3.3, 5. T.J. Chan, C.K. Chan, L.K. Chen and F. Tong, A Self-Protected Architecture for Wavelength-Division-Multiplexed Passive Optical Networks, IEEE Photon. Technol. Lett., ol.15, pp , Nov. 3. K.H. Han, E.S. Son, H.Y. Choi, K.W. Lim and Y.C. Chung, Bidirectional WDM PON Using Light-Emitting Diodes Spectrum-Sliced With Cyclic Arrayed-Waveguide Grating, IEEE Photon. Technol. Lett., ol.16, pp.38-38, Oct. 4. 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., ol.1, pp , Aug.. N.J. Frigo, P.P. Iannone, P.D. Magill, T.E. Darcie, M.M. Downs, B.N. Desai, U. Koren, T.L. Koch, C. Dragone, H.M. Presby and G.E. Bodeep, A Wavelength-Division Multiplexed Passive Optical Network with Cost-Shared Components, IEEE Photon. Technol. Lett., ol.6, pp , F. Payoux, P. Chanclou, T. Soret and N. Ganay, Demonstration of a RSOA-based Wavelength Remodulation Scheme in 1.5Gbit/s Bidirectional Hybrid WDM-TDM PON, in Proc.Optical Fiber Communication(OFC), OTuC4, 6. S.Y. Kim, E.S. Son, S.B. Jun, J.H. Lee, D.H. Lee, E.D. Sim and Y.C. Chung, Effects of downstream modulation formats on the performance of a RSOA-based WDM PON, in Proc.SPIE, 6353 I, paper63531q, 6. Y.J. Lee, K.Y. Cho, A. Murakami, A. Agata, Y. Takushima and Y.C. Chung, Reflection Tolerance of RSOA-based WDM-PON, in Proc.Optoelectronics and Communication Conference (OECC), 11A1-4, 7. M. Fujiwara, J. Kani, H. Suzuki and K. Iwatsuki, Impact of Backreflection on Upstream Transmission in WDM Single-Fiber Loopback Access Networks, J. Lightw. Technol., ol.4, pp , 6. Proc. of SPIE ol I-5