1. Introduction

Reflective electro-absorption modulator (R-EAM) has been considered as an attractive solution for a colorless upstream source in high-capacity wavelength-division multiplexed passive optical networks (WDM-PONs) [1–4]. This is mainly because it could provide a wide modulation bandwidth and a low chirp, compared to Fabry-Perot laser diode (FP-LD) and reflective semiconductor optical amplifier (RSOA) which have been widely used for the implementation of colorless WDM-PON systems [5, 6]. Therefore, the colorless WDM-PON systems operating at a 10 Gb/s and beyond have been demonstrated with R-EAM based upstream optical sources, recently [1–4]. However, in the R-EAM based WDM-PON systems, a high insertion loss of R-EAM could induce a significant degradation of optical signal-to-noise ratio (OSNR). To resolve this problem, a SOA has been integrated with R-EAM to provide an optical gain for the compensation of insertion loss [1, 2, 4]. In addition, only wavelength-specific optical sources have been used as seed sources for the recent demonstration of R-EAM based WDM-PON systems. Previously, it has been reported that broadband light source (BLS) would be more cost-effective seed source than wavelength-specific sources for the colorless WDM-PON systems [5]. In a typical BLS seeded optical source configuration, an array of semiconductor devices, such as FP-LD and RSOA, could generate a number of downstream or upstream WDM signals with a single BLS [5, 6]. However, both excess intensity noise (EIN) and inherent wide source bandwidth of BLS seeded optical source could degrade severely the performance of WDM-PON systems operating especially at a 10 Gb/s and beyond. Therefore, in this paper, we have evaluated the performance of R-EAM based optical source with a BLS seed source for the high-capacity colorless WDM-PON applications. Like an integrated R-EAM and SOA [1, 2, 4], an RSOA was also used for the implementation of our BLS seeded optical source to suppress the EIN within BLS and to provide an optical gain for the compensation of R-EAM insertion loss. First, we have found that the 3-dB optical bandwidth of our BLS seeded R-EAM optical source should be wider than 0.8 nm to achieve an error-free transmission of a 1.25 Gb/s signal. Thus, we could conclude that the 3-dB optical bandwidth of BLS seeded R-EAM source should be increased to be wider than 8 nm for a 10 Gb/s signal error-free transmission, since the SNR of BLS based optical source is determined by the ratio of optical bandwidth and electrical bandwidth [7–9]. Then, to evaluate the impact of the wide source bandwidth of the BLS seeded R-EAM source, we have also measured and calculated the power penalties induced by a chromatic dispersion and an in-band crosstalk, respectively. From the results, we confirmed that the wide source bandwidth of the BLS seeded R-EAM source limited the maximum reach of system due to the chromatic dispersion of transmission fiber, but provided a better tolerance to the in-band crosstalk than distributed feedback laser diode (DFB-LD) seeded R-EAM source.

2. Results and discussion

Figure 1 shows the experimental setup for the performance measurement of WDM-PON system using an R-EAM based optical source. As a seed source, an amplified spontaneous emission (ASE) from a BLS was spectrally-sliced with an optical bandpass filter, and then launched into an R-EAM and a RSOA with a 4-ports circulator. The R-EAM was used to modulate the spectrum-sliced BLS output at a 1.25 Gb/s non-return-to-zero (NRZ) pseudo-random pattern of length 2³¹-1. The bias and modulation voltages of the R-EAM were set to be −1.2 V and ± 1.5 V, respectively. The 3-dB RF bandwidth of the R-EAM was ~8 GHz, which could be used for a 10 Gb/s signal generation. However, we decided to use a 1.25 Gb/s signal for the performance evaluation of the BLS seeded R-EAM source, since a broad bandwidth of seed source was needed to obtain an error-free transmission of 10 Gb/s signal even in a back-to-back configuration. The insertion loss of the R-EAM including a circulator insertion loss and a modulation loss was measured to be ~10 dB. The RSOA was also used to amplify the modulated signal for the compensation of the R-EAM insertion loss, and suppress the EIN within a spectrum-sliced BLS. In our measurement, the current and gain of the RSOA were set to be 24 mA and ~20 dB, respectively. The amplified signal at the wavelength of 1556.7 nm was transmitted through a conventional single-mode fiber (SMF) and then demultiplexed by using a second optical bandpass filter. Finally, the demultiplexed signal was detected by use of a PIN-based receiver having a 940 MHz electrical filter.

 

Fig. 1 Apparatus used for the performance measurement of an R-EAM based WDM-PON system. Acronyms are receiver, Rx; and bit-error-rate test sets, BERTs.

Download Full Size | PDF

At first, bit-error-rate (BER) curves have been measured with the different 3-dB optical bandwidths of the BLS seeded R-EAM optical source in a back-to-back configuration, as shown in Fig. 2. Here, the 3-dB optical bandwidth of the BLS seeded R-EAM source was estimated from the optical spectrum measured just before an optical receiver, in order to include the cascading effect of optical bandpass filters located at both transmitter and receiver sides. Previously, it has been reported that the SNR of BLS based optical source could be determined by the ratio of optical and electrical bandwidths [7–9]. Thus, in order to evaluate the effect of optical source bandwidth on the improvement of SNR and the reduction of EIN in the BLS based optical signal, we used three different types of optical bandpass filter; an arrayed waveguide grating (AWG) and two tunable optical bandpass filters. The 3-dB bandwidth of each filter was measured to be 0.4 nm for an AWG, 0.96 nm for one optical bandpass filter and 2.94 nm for the other optical bandpass filter, respectively. These three different types of optical filters were used at the transmitter and/or receiver sides to get a different 3-dB bandwidth of the BLS seeded R-EAM source, as shown in Fig. 1. Usually, AWGs at both transmitter and receiver sides were used for the generation of colorless WDM source and the multiplexing/demultiplexing of signals in conventional colorless WDM-PON systems [5, 6]. However, even in our back-to-back measurement, a BER floor at 10⁻⁶ was observed with the AWGs having a 3-dB passband of 0.4 nm (@ 0.32 nm in Fig. 2 due to the cascading effect of two AWGs). This is mainly because of SNR degradation induced by a high insertion loss of R-EAM in our optical source. In addition, due to the low input power of modulated signal into the RSOA, the EIN in the spectrum-sliced BLS could not suppress sufficiently [10]. Thus, in order to improve the SNR of signal, the wide bandwidth filters were used at both transmitter and receiver sides, as shown in Fig. 2. However, even in a case of 0.66 nm bandwidth, a BER floor at 10⁻⁹ was still observed. From the measurements, we found that the 3-dB bandwidth of our BLS seeded R-EAM optical source should be wider than ~0.8 nm to achieve an error-free transmission of a 1.25 Gb/s. Thus, we conclude that the 3-dB bandwidth of BLS seeded R-EAM optical source should be increased to be wider than ~8 nm to maintain this SNR and achieve an error-free transmission of a 10 Gb/s signal. In order to reduce the required 3-dB bandwidth of the BLS seeded R-EAM optical source and improve the dispersion-limited maximum reach, some compensation and/or equalization techniques might be helpful especially for a 10 Gb/s signal transmission.

 

Fig. 2 Measured BER curves with different 3-dB bandwidths of a BLS seeded R-EAM optical source in a back-to-back configuration.

Download Full Size | PDF

Figure 3 shows the optical spectrum of the BLS seeded R-EAM source measured after passing through a second optical bandpass filter. The 3-dB optical bandwidth of the source was measured to be ~0.88 nm. Using this source bandwidth, we could achieve an error-free transmission of a 1.25 Gb/s signal, as shown in Fig. 2. Previously, we have reported that the performance of BLS based WDM-PON systems could be estimated properly with an equivalent optical bandwidth rather than with a 3-dB optical bandwidth [9, 11, 12]. Thus, from the measured optical spectrum of Fig. 3, the equivalent optical bandwidth of the BLS seeded R-EAM source was estimated to be 1.06 nm. Using this equivalent optical bandwidth B_o, we believe that we can calculate the power penalties induced by a chromatic dispersion of transmission fiber and an in-band crosstalk with the following equations [11, 13];

 

Fig. 3 Optical spectrum of a BLS seeded R-EAM source measured at a receiver side after passing through a second optical bandpass filter.

Download Full Size | PDF

Then, in order to evaluate the impact of wide equivalent source bandwidth (@ 1.06 nm) on the system’s performance, we have measured the power penalties induced by a chromatic dispersion and an in-band crosstalk in a WDM-PON using the BLS seeded R-EAM optical source. At first, the dispersion-induced penalties were measured by increasing the transmission distance of SMF with the equivalent source bandwidth of ~1.06 nm and the experimental setup shown in Fig. 1. Then, both calculated and measured penalties were represented in Fig. 4, for comparison. In our calculation with Eq. (1), we assumed that the dispersion value of SMF was −17 ps/nm/km. A dispersion-induced penalty of ~2.5 dB was measured after a 25-km SMF transmission. Moreover, a BER floor at 10⁻⁹ was observed again after a 30-km SMF transmission. However, as shown in Fig. 4, we re-confirmed that the calculated penalty with the Eq. (1) underestimated the measured one, like the cases of BLS seeded RSOA and FP-LD sources [10]. In order to reduce this discrepancy between the measured and calculated penalties with the BLS seeded optical source, it is desirable to take into account the effect of dispersion-induced EIN increase in the Eq. (1) especially with high values of chromatic dispersion [10]. From the results, we found that the chromatic dispersion limited the maximum reach to be less than 20 km even with a 1.25 Gb/s NRZ signal and a 1-dB penalty guide line.

 

Fig. 4 Measured and calculated dispersion-induced penalties of a 1.25 Gb/s NRZ signal as a function of SMF transmission distance.

Download Full Size | PDF

We have also measured the in-band crosstalk-induced penalty using an experimental setup shown in Fig. 5(a). The paths of original signal and in-band crosstalk components were separated by using two optical splitters; one 90: 10 coupler and one 50:50 coupler. The power level of the in-band crosstalk component was simply adjusted by using a variable optical attenuator, and a 25 km of SMF was used to de-correlate the in-band crosstalk component with the original signal. In our measurement, the R-EAM based optical sources seeded by a BLS and a DFB-LD were used to evaluate the effect of seed source on the in-band crosstalk-induced penalty [13]. For comparison, two calculated power penalties were also represented in Fig. 5(b). The Eq. (2) was used for the calculation of the in-band crosstalk-induced penalty in the BLS seeded R-EAM source while the well-known equation of in-band crosstalk penalty for an externally-modulated source was used for the case of DFB-LD seed source [13]. As it can be seen in Fig. 5(b), the BLS seeded R-EAM source were measured to be ~8 dB more tolerant to the in-band crosstalk than the DFB-LD seeded one. In case of the BLS seeded R-EAM source, the measured penalty agreed well with the calculated one especially at low levels of crosstalk-to-signal ratio. However, at a high crosstalk-to-signal ratio of >-10 dB, the discrepancy between the measured and calculated penalties was observed due to the low extinction ratio of the R-EAM. In the penalty calculation of the BLS seeded R-EAM source, we used two different values of κ to take into account the effect of polarization states of signal [12, 13]. However, the difference in penalties between two polarization states was quite small even at a high crosstalk-to-signal ratio. Thus, we found that the penalty in our BLS seeded R-EAM source was mainly determined by the second term of the Eq. (2) due to the wide equivalent source bandwidth of the BLS seeded R-EAM source. In case of DFB-LD seeded R-EAM source, the discrepancy between the calculated and measured penalties was observed, as shown in Fig. 5(b). We believe that this is mainly because the equation for the DFB-LD case does not include the effect of chirp-induced spectral broadening [13].

 

Fig. 5 (a) Experimental setup for measurement of in-band crosstalk-induced penalties and (b) measured and calculated penalties as a function of crosstalk-to-signal ratio. The symbols and lines represent the measured and calculated penalties, respectively.

Download Full Size | PDF

3. Summary

The performance of BLS seeded R-EAM optical source has been evaluated by taking into account the effect of various noises, such as EIN, chromatic dispersion of transmission fiber and in-band crosstalk. Due to the SNR degradation and insufficient EIN suppression caused by a high insertion loss of R-EAM, we found that the equivalent optical bandwidth of the BLS seeded R-EAM source should be wider than ~1 nm even for a 1.25 Gb/s signal error-free transmission. Thus, for a 10 Gb/s signal transmission with our BLS seeded R-EAM source, we believe that the equivalent bandwidth should be increased to be >10 nm unless any compensation or equalization techniques would be used for an error-free transmission. We have also evaluated the effect of this wide source bandwidth on the system’s performance. From the results, the dispersion-limited reach was measured to be shorter than 20 km of SMF even at a 1.25 Gb/s signal transmission. On the other hand, due to the wide bandwidth, the BLS seeded R-EAM source was more tolerant to the in-band crosstalk than DFB-LD seeded R-EAM source.