1. Introduction

A wavelength division multiplexing passive optical network (WDM-PON) is considered the ultimate solution for a next generation access network. In order to realize the WDM-PON, a cost-effective WDM source that supports color-free operation (i.e. wavelength independent operation of the optical network termination) has been demonstrated successfully based on wavelength locked F-P LDs [1, 2]. In addition, a fault-localization is also required for reliable operation of the WDM-PON. The fiber fault can be detected by using optical time domain reflectometry (OTDR). However, the WDM-PON has many drop fibers after the remote node. Thus, it may be necessary to send a craftsman to the ONT side to inject an OTDR pulse into the drop fiber. This approach would require much time and effort. To solve these problems, a tunable OTDR was realized by using a tunable laser [3]. However, this is an expensive solution with respect to its application in access networks.

In this paper, we propose a new tunable OTDR for in-service monitoring of the fiber fault in the WDM-PON. A wavelength-locked Fabry-Perot laser diode (F-P LD) by an injected spectrum-sliced amplified spontaneous emission (ASE) [4] is used as an optical source for a tunable OTDR. We successfully demonstrate fault-localization in the drop fibers with the proposed OTDR. Finally we analyze the effects of crosstalk induced by the OTDR signal into the data signal.

 

Fig. 1. Experimental setup.

Download Full Size | PDF

2. Experimental setup and operation principle

The proposed fault-localization unit is shown in the dotted box in Fig. 1 in conjunction with a typical WDM-PON architecture that is based on either DFB LDs or wavelength locked F-P LDs [1, 2, 4]. The proposed scheme consists of a broadband light source (BLS), a tunable band-pass filter (BPF), a F-P LD, an optical circulator, an OTDR receiver, and other electrical parts. We used an AR-coated F-P LD to enhance the ASE injection efficiency. The laser has a cavity length of 600 μm. The AWG has 100 GHz channel spacing with a Gaussian type pass band. The tunable BPF has 3 dB bandwidth of 0.56 nm. The output of the fault-localization unit was coupled into the feeder fiber of the WDM-PON through a C/L WDM filter that passes C and S band and reflects L band (for the inputs at the common port).

In the experiment, OTDR pulse wavelength was assigned at L-band, while the upstream signal was at C-band and the downstream signal at S-band. However, we can use a single AWG for the remote node, since it has periodic transmission characteristics. A different OTDR band from the signal bands enables in service monitoring with negligible crosstalk.

Operation principle can be explained as follows. The control unit adjusted the tunable BPF to the target wavelength. The light from the BLS was then spectrum-sliced by the tunable BPF and injected into the F-P LD. We thereupon obtained a quasi single mode output from the F-P LD that was directly modulated by a series of electrical pulses. The F-P LD output was filtered by the tunable BPF to suppress residual side modes and coupled into the feeder fiber. The backscattered light from the transmission fiber was detected by the PIN photodetector (PD) and processed to find fault positions, as in a conventional OTDR. We used an oscilloscope and a computer as an A/D converter and a signal processor, respectively.

We aimed the fault detection within 25 km of the transmission fiber including the drop fibers and the AWG located at the remote node (RN). It may be noted that the maximum transmission length for an access network is usually considered as 20 km. The detection range of 25 km determines the maximum repetition rate of the OTDR pulse to about 4 kHz (the repetition rate < (2nl/c)-l, where n is the group index of the transmission fiber, l the length of the fiber, and c the speed of light). We set the repetition rate to 2 kHz and pulse width to 1μs [5] to meet the target spatial resolution of 100 m.

The required dynamic range of the OTDR can be estimated from the loss in the transmission path. The link loss was estimated to be about 12 dB (C/L WDM filter: 0.75dB, fiber: 6.25 dB, AWG: 5 dB). Thus, we designed the OTDR receiver to have a 12 dB dynamic range. The OTDR power in peak at the worst case was -4 dBm at a BLS injection power of -24 dBm/0.2 nm and the back scattering power measured at the OTDR receiver was -60 dBm. It means that the backscattering return loss is 56 dB which agrees with estimation based on Koh-ichi Aoyama’s theory [5]. Therefore it is required that the receiver sensitivity of -84 dBm to have 12 dB dynamic range for one direction. However, the measured receiver sensitivity was about -53 dBm. So it requires the noise improvement of 31 dB using averaging (-53-(-84) = 31). To achieve this, we need averaging number N of 2^20.66≈2²¹ (0.5×3×log2(N)=31 dB).

3. Experimental results

 

Fig. 2. Wavelength spectra of 4-channels after the AWG at RN.

Download Full Size | PDF

(Each spectrum results from the OTDR band, not the data signal band, and there is different detuning for each channel)

To demonstrate fault localization in the WDM-PON, we attached 4 different drop fibers with different lengths at consecutive ports of the RN, as shown in Fig. 1. Then, OTDR wavelength was tuned to the center wavelength of each channel by using tunable BPF. The measured optical spectra of OTDR pulses are shown in Fig. 2.

In order to simulate the worst case condition, the injection wavelength of channel 1 was tuned to the middle of the two modes of the F-P LD (see the first spectrum in Fig. 2). This can be compared with the best case of channel 2, when the injection wavelength matched to the lasing wavelength (see the second spectrum in Fig. 2). We also show two different spectra for comparison. As can be seen at the spectra, the combination of BLS and tunable BPF, and the F-P LD operates like a tunable laser. By coupling the light output from the fault localization unit to the feeder fiber, we obtained the ODTR traces shown in Fig. 3. The injected OTDR pulse power into the transmission fiber was -4 dBm at the worst case [Fig. 3 (a)]. Clear indication of faults at different locations was observed. The calculated fault positions match to the length of the drop fibers, i.e., 3 km, 3.4 km, 4 km, and 5.2 km from the remote node, respectively.

The worst case dynamic range was measured as 12 dB in one way at ch. 1. The peak at 20 km was resulted from reflection at the AWG. It should be noted that the measured insertion loss difference of the remote node can be explained by the difference of the filtering loss caused by different spectral width, as shown in Fig. 2. We measured the spatial resolution by resolving two pulses that are reflected at the two adjacent optical connectors. The spatial resolution was about 100 m. Thus, the realized OTDR satisfies the design targets.

Although the OTDR band is different from the data signal bands, crosstalk between the OTDR signal and the data signal can occur depending on the signal power difference. There exist four different crosstalks. The OTDR signal can induce crosstalk to the upstream and the downstream receivers, while the signals can induce crosstalk into OTDR receiver. Among them, the induced crosstalk into the upstream receiver can be negligible, since the back reflected OTDR signal is very weak. Similarly, the crosstalk into the OTDR receiver induced by the back reflected downstream signal can be negligible. In addition, the crosstalk into OTDR receiver induced by the upstream signal can be suppressed by the C/L WDM. If the isolation of the C/L WDM in not enough, we can add another C/L WDM simply. Also it may be noted that the data signal can be averaged out, since the bandwidth of the OTDR receiver is very narrow compared the data signal.

 

Fig. 3. Fault detection results for (a) 3 km, (b) 3.4 km, (c) 4 km, and (d) 5.2 km from the remote node

Download Full Size | PDF

 

Fig. 4. (a). The measured BER as a function of crosstalk (b). Theoretical and measured power penalty (@ BER=10^-10)

Download Full Size | PDF

Then, we investigate the effects of crosstalk into the downstream data by the OTDR pulse. The measured BER curves as a function of the crosstalk level are shown in Fig. 4. Here the crosstalk level is defined as the difference between the OTDR pulse peak power and the average signal power at the receiver of an ONT. We used the wavelength locked F-P LD for downstream data transmission [2]. And it was modulated by 155 Mb/s (pattern length: 2³¹-1) NRZ data. The extinction ratio of the modulated data was about 13 dB. It may be noted that many WDM-PONs were demonstrated based on wavelength locked F-P LD [1, 2]. As shown in Fig.4, the power penalty induced by the OTDR signal can be maintained below 1 dB at BER of 10^-10 when the crosstalk level is below - 7 dB. In the proposed scheme, we coupled – 4 dBm peak power into transmission power for OTDR. Then, the crosstalk level of – 7 dB can be easily achieved with a conventional C/S WDM filter, since the isolation level is more than 30 dB. In addition, we can avoid the problem of Raman scattering [6]. The measured results agree reasonably well with the analytical results based on a well known Gaussian noise approximation [7].

4. Discussion

Since the received power at the OTDR receiver is less than -50 dBm, the relative intensity noise of the OTDR source is not critical. To confirm this, we used two different sources to obtain the OTDR traces. The measured RIN of the two sources are shown in Fig. 5.

 

Fig.. 5. The measured RIN (a) wavelength-locked F-P LD with - 24 dBm / 0.2 nm, (b) spectrum-sliced F-P LD itself.

Download Full Size | PDF

 

Fig. 6. OTDR trace (a) wavelength-locked F-P LD (b) spectrum-sliced F-P LD

Download Full Size | PDF

One is the RIN of the wavelength locked F-P-LD used for the OTDR while the other is the spectrum sliced F-P LD itself, i.e., no injection. As shown in Fig. 5(b), the RIN increases by about 22 dB. However, the noise floor level of measured OTDR traces is almost the same as shown in Fig. 6. Thus we confirm that the effects of the RIN of the source are negligibly small on the performance of OTDR. However, there exists difference in the dynamic range caused by power difference of the OTDR pulse injected into the fiber. In other words, if the power of spectrum sliced solitary F-P LD were enough, it may possible to use it as an OTDR source regardless of the RIN. It should be noted that the wavelength locking process can be considered as an amplification of the injected ASE. The BLS injection power of -24 dBm/0.2 nm corresponds to -3 dBm over 26 nm, which covers 32 channels with 100 GHz spacing. This can be easily achieved at a low cost. In addition, a cost effective tunable filter can be realized by thermal tuning [8].

In this experiment, 12 dB dynamic ranges are obtained by averaging 2²¹ times. This corresponds to 17.5 minutes measurement time for a 2 kHz repetition rate. The minimum detection time is determined by the average number and the OTDR pulse repetition rate, which in turn is determined by the fiber length to be measured. Thus, to decrease the measurement time, it is necessary to decrease the average number by increasing the SNR (signal to noise ratio) of the received OTDR pulse. It may be noted that the SNR can be enhanced by increasing the OTDR signal power. It is also possible to use an avalanche photodiode instead of a PIN photodiode to increase the SNR. There are also effective detection techniques to increase the SNR, such as heterodyne detection instead of direct detection [9]. The enhancement of the SNR can be used to increase the dynamic range. Thus there is a trade-off between the dynamic range and the measurement time. The wider dynamic range may be needed for an optical fiber with a higher loss or a longer transmission length.

It may be noted that a commercial standalone OTDR has a spatial resolution of around 20 m. This is commonly used for out of service monitoring, although it is not tunable one. However, the proposed method aims to offer a cost-effective tunable OTDR for in-service monitoring. We think 100 m may be enough for this application. The spatial resolution can be improved with a shorter pulse injection.

5. Conclusion

In conclusion, we have demonstrated a tunable OTDR for in-service monitoring of fiber faults in a WDM-PON. The realized OTDR has a spatial resolution of 100 m and a dynamic range of 12 dB. Furthermore the crosstalk into the downstream data signal was negligible. The proposed fault-localization scheme can be realized cost effectively and would be useful to monitor the status of the outside plant of a WDM-PON in-service condition.