1. Introduction

Mobile devices, such as smartphones and tablets, have changed personal lives to a great extent. At the same time, these mobile devices had a deep impact on mobile backhaul transport, which brought a bottleneck in the mobile network; this was because the current backhaul infrastructure was not optimized to handle this drastic increase traffic [1-2]. It has been widely considered that the ultimate choice for facilitating large capacity is to use a fiber, owing to its nearly unlimited transmission bandwidth [3-4]. Among the different fiber based solutions available, wavelength division multiplexing-passive optical network (WDM-PON) has been considered to be the most suitable technology for meeting the current demand of high bandwidth. This is because it can provide dedicated user bandwidth up to 10 Gbit/s per subscriber using a designated wavelength regardless of transmission protocols. Furthermore, the reason for the WDM-PON to be considered as one of the most promising candidates as a fiber based backhaul solution is that it can solve the power budget problems and can mitigate the complicated bandwidth allocation or ranging problems in conventional time-shared optical access networks [5].

In order to make WDM-PON cost-effective and practical for widespread deployment, many researchers have been developing colorless light sources for optical network units (ONUs). Colorless light sources can be categorized into tunable lasers and reflective type broadband sources, such as reflective semiconductor optical amplifier (RSOA), and Fabry-Perot laser (FP-LD) [6–12]. Among these light sources, tunable lasers are considered as an attractive solution for use in WDM-PON, owing to their flexibility and scalability [8–12]. However, a wavelength initialization process is necessary before starting communication using tunable lasers; this is because the output wavelength of the tunable lasers is not fixed. A straightforward and simple way to achieve wavelength initialization is to use a lookup table, which is usually predetermined and loaded in the memory of the tunable transceiver. A lookup table has to be generated for each of the lasers because of the manufacturing variations. Moreover, the value of the control parameters in the lookup table need to be adjusted because of either laser aging or temperature changes. The time for generating a lookup table depends on the tuning mechanism of the laser diodes, and although, some methods have been proposed for generating a lookup table in a short time, the overall generating process is exhaustive and requires a time-consuming scanning process [13]. Because of this, the device packaging cost increases. Furthermore, remote control between the central office and subscriber is required to use the lookup table; this can impair the unique advantage of protocol transparency that the WDM-PON provides. Hence, from a practical point of view, it is strongly preferred to find a wavelength initialization method that keeps operator intervention to a minimum. Several research groups have proposed wavelength initialization methods in the optical domain [14–17]. One of the group achieved wavelength initialization by alignment of the wavelength by maximizing the back-scattered optical power from the feeder fiber, whereas another group achieved wavelength initialization by maximizing the back-reflected optical power. There was another proposal regarding the use of the beating phenomena between the lasing light output and intentionally added light.

In this paper, we propose a self-wavelength initialization method for Bragg-grating based tunable light sources. The principle and the feasibility of the proposed initialization method are explained in section 2. The effectiveness of the proposed method along with the transmission results is experimentally investigated in section 3. Finally, this paper will be summarized in section 4.

2. Principle and feasibility of self-wavelength initialization method

In the proposed self-wavelength initialization method, we have used the unique characteristics of the Bragg-grating reflector and the WDM-PON. The Bragg-grating, a wavelength selective reflector in the external cavity laser, reflects light not only in the direction toward the inside of the cavity, but also in the direction away from the cavity. Because the optical line terminal (OLT) and the optical network unit (ONU) are logically point-to-point connected in the WDM-PON, the wavelength of the light that arrives at the ONU comes from the OLT and passes through the WDM multiplexer (MUX) and demultiplexer (DeMUX). Therefore, the wavelength that passes through the WDM MUX becomes the target wavelength for the tunable laser in the ONU. As shown in the inset of Fig. 1 , the lasing wavelength of the Bragg-grating based tunable external cavity laser is mainly determined by the peak wavelength of the Bragg reflector. Therefore, by monitoring the amount of the optical power reflected from the Bragg reflector, one can simply check whether the lasing wavelength of the tunable laser is well aligned.

 

Fig. 1 Simplified configuration of the WDM-PON using the Bragg-grating based tunable external cavity laser. Inset is the lasing principle of the Bragg-grating based tunable external cavity laser

Download Full Size | PDF

The basic concept of the proposed initialization method appears to be similar to the previously demonstrated method [14]. The difference between the previous method and the proposed method is the type of target light used as a reference [18]. In the previous method, the target light was a downstream signal. If the wavelengths of the upstream and downstream signals are the same, then the transmission performance would degrade owing to the beating noise. Therefore, the wavelength band is normally separated by the multiples of the free spectral range (FSR) of the arrayed waveguide grating (AWG) in the WDM-PON, especially when single-mode lasers are used as transmitters. Under such conditions, the downstream signal can barely be used as a target for the upstream signal. Furthermore, the previous method cannot be applied to the tunable laser in the central office. In the current proposed method, we have used the incoherent light as the reference target light. The incoherent lightcan be a downstream signal or an additional low-power broadband light source depending on the link configuration, as shown in Fig. 2 . The main difference between Fig. 2(a) and Fig. 2(b) is the downstream transmitter. In Fig. 2(a), both the upstream and the downstream transmitters are tunable lasers. On the other hand, in Fig. 2(b), only the upstream transmitters are tunable lasers. The link configuration shown in Fig. 2(b) can be implemented any time the operators want to upgrade the seeded WDM-PON. Although tunable lasers have superior transmission characteristics and offer flexibility to the network operators, they are still considered as a costly solution for use in optical access network. Therefore, some people have assumed that the reflective type colorless transmitters based WDM-PON will be commercialized before the tunable lasers based one. Some telecom companies in Korea used RSOAs to start commercial service already [7]. Furthermore, the standardization regarding seeded WDM-PON has been ongoing under the management of ITU-T. However, the seeded WDM- PON has some constraints such as scalability and long-reach transmission [19]. This is because the transmission performance of these reflective type transmitters is been strongly dependent on the power level of the seed light. Although many research groups have tried mitigating the above-mentioned problems, it would be reasonable to consider the link configuration shown in Fig. 2(b), wherein reflective type transmitters for the downstream transmission and tunable lasers for the upstream transmission are used.

 

Fig. 2 WDM-PON link configurations (a) tunable lasers for both upstream- and downstream transmission (b) tunable lasers for upstream transmission and reflective type transmitters for downstream transmission. The simplified structures of the tunable transceiver are sketched in the inset of the Fig. 2(a) and Fig. 2(b).

Download Full Size | PDF

In order to demonstrate the proposed wavelength initialization method, we have used the waveguide-Bragg-grating (WBG) based external cavity laser. The reflection peak of the WBG was tunable in the C-band, and therefore, the laser output also is tunable in the C-band. Thesimplified structure and the operating principle of the tunable external cavity laser are shown in Fig. 1. A laser diode chip was used as a gain medium and a polymer WBG was used as a wavelength selective reflector. The polymer material, which has a large thermo-optic coefficient, was used in order to obtain a wide tuning range. For thermal tuning a Cr/Au thin-film heater was deposited on the top of the polymer WBG. By applying an electric current to the heater, the peak wavelength in the reflection spectrum was tuned, almost linearly proportional to the electrical heat power. Owing to this thermal tuning mechanism, the tuning time would increase by a few seconds. Therefore this type of tunable laser would be not appropriate for a wavelength agile architecture where fast switching is mandatory. However, it can be applied in the WDM-PON because tunable lasers have been used as a substitute for the fixed wavelength transmitter. The details of the structure and the performance of this polymer WBG based tunable external cavity laser have been well described in other papers [10–12].

In Fig. 3(a) and Fig. 3(b), a simplified diagram of the tunable transceiver and a flow chart for the initialization process are shown, respectively. The band splitter in Fig. 3(a) was used to split the different wavelength bands. The band splitter is not required in the link configuration shown in the Fig. 2(b). As shown in Fig. 3(b), the first and foremost thing to do is make sure the laser diode is turned off before starting the initialization process. This is because the monitor PD (mPD) will not be able to distinguish the reflected light from the laser output. The heater power at which the maximum photocurrent is achieved can be determined by a scan of the heater power. Although a detail scanning procedure would be dependent on the peaksearch algorithm, Fig. 3(b) shows an example by which the maximum reflected point can be determined.

 

Fig. 3 (a) Simplified diagram of the transceiver in the configuration shown in Fig. 2(a), and (b) flow chart for the wavelength initialization process

Download Full Size | PDF

In order to demonstrate the feasibility of the proposed method, we have used a WDM MUX, at 200 GHz channel spacing. As shown in the optical spectrum (right-hand side) in Fig. 4 , because of the residual reflectance in the Bragg reflector, some optical power is measured although the peak wavelength of the reflector is misaligned with the center of the WDM multiplexer. The reflected optical power ratio in dB for the aligned and misaligned cases is determined by the characteristics of the Bragg reflector. Hence, the proposed method can be applied even with low power. The proposed wavelength initialization procedure is shown in Fig. 4. We have controlled the heater power by using the adaptive step-size to find the peak. As mentioned above, the initialization time would be dependent on the peak search algorithm.

 

Fig. 4 the reflected optical power versus heater power, the optical spectrum of the reflected output when the wavelength of the Bragg reflector and the external light is aligned and misaligned.

Download Full Size | PDF

3. Experiments and discussion

The experimental setup for demonstrating the effectiveness of the proposed method is shown in Fig. 5(a) and Fig. 5(b), which is similar to the setup shown in Fig. 2(a). We have assumed that the wavelength band for the upstream and the downstream signal have been separated by the FSR of the AWG. Therefore, we separately examine their transmission performance, as shown in Fig. 5(a) and Fig. 5(b). When the tunable lasers are located in the central office with the broadband light source, the optical power injected into the laser cavity might be too strong. This is because there is no place to put isolators to adopt the proposed initialization method. Because the refractive index of the semiconductor lasers is strongly dependent on the carrier density, the phase or amplitude of the laser light can be varied by the strong injection power. Therefore, the influence of external injection into the laser cavity has been examined for the downstream transmission with the experimental setup shown in Fig. 5(a). When the tunable lasers are located in the subscriber region, as shown in Fig. 5(b), the upstream transmission performance degrades by back-scattered broadband light source.

 

Fig. 5 Experimental setup to investigate two different cases: (a) tunable lasers located in the central office and (b) tunable lasers located in the subscriber region.

Download Full Size | PDF

We have used an erbium doped fiber amplifier (EDFA) as a reference broadband light. For the WDM MUX and DeMUX, we have used thin-film filters, with a channel spacing of 100 GHz, and 200 GHz. The transmission performance was evaluated using 20 km and 40 km single-mode fibers. A 3 dB power splitter was used in front of the external cavity laser to align the wavelength of the laser. The wavelength of the laser was aligned to 1542.94 nm. The laser was directly modulated at 1.25 Gbit/s NRZ data with a pseudorandom bit sequence (PRBS) pattern length of 2⁷-1. The extinction ratio and the output were about 9 dB,and about 3 dBm, respectively. APD based optical receiver was used to measure the BER plots.

In order to examine the influence of the external optical injection into the cavity, we have controlled the optical variable attenuator to vary the injection power. The extent of performance degradation would be a function of the power ratio between the external injected light and the laser output. However it is difficult to obtain the exact value of this ratio, because of coupling losses and the wavelength dependency of the WBG reflectivity. Moreover the external injection, originating from the reference incoherent light and the backscattered portion of the laser output itself is hard to separate. Therefore, we have overcome these problems by simply defining an injection ratio based on externally measurable parameters. Here, we have defined the external injection power ratio as P_{external_injected}/P_output. The superposition of the laser output is shown in Fig. 6(a) and Fig. 6(b). Because the spectral width of the 100 GHz spaced MUX is about half of that of the 200 GHz spaced MUX, the spectral density of the externally injected power is different in both cases, even the external injection power ratio was calculated to be the same. Therefore, the peak wavelength of the laser output has changed with the MUX spacing of 100 GHz.

 

Fig. 6 Superposition of the laser output spectrum under different external optical injection ratios. The channel spacing of the MUX/DeMUX were (a) 200 GHz and (b) 100 GHz

Download Full Size | PDF

The BER plots were obtained for different lengths of the fiber to examine the degradation in the transmission performance under such strong injection. Because of the narrow linewidth of the laser output, there was negligible penalty due to the spectral width of the WDM MUX, as shown in Fig. 7(a) . A comparison of Fig. 7(a) and Fig. 7(b), the dispersion penalty after transmission using the single-mode fiber of 40 km is about 0.5 dB when there is no external injection, and when there is a strong external injection, the dispersion penalty after 40 km is about 0.9 dB with the MUX spacing of 200 GHz. When the MUX spacing of 100 GHz is used, the dispersion penalty after 40 km is more than 1.5 dB. The additional power penalty of 0.4 dB and 1 dB were cause by the spectral width of the signal output, as shown in the Fig. 6(b). If the laser is modulated at higher bit rate than 1.25 Gbit/s, the allowable transmission length will be limited owing to the dispersion induced power penalty. Although we examine the degradation effect due to strong optical injection, it may be noted that the high power broadband light source is not necessary in the proposed initialization process. This is because the proposed method only monitors the optical power difference in a dB scale, and not the amount of optical power reflected.

 

Fig. 7 Downstream bit error rate plots after (a) 0 km transmission and (b) 40 km transmission

Download Full Size | PDF

In order to investigate the effectiveness of the proposed method, when the tunable lasers are placed in the subscriber regime, we measured the BER plots using the experimental setup, shown in Fig. 5(b). In such a case, the transmission performance is expected to degrade by the back-scattered broadband light source. The beat noise originates from the mixing or beating of the coherent laser signal with the incoherent broadband light in the same polarization [20]. Because the amount of incoherent light, which arrives at the receiver, depends on the optical bandwidth of the WDM, it is assumed the denser spaced MUX would exhibit the larger power penalty. By varying the launching power into the fiber, the BER plots were measured, as shown in Fig. 8(a) and Fig. 8(b). There was power penalty lower than 0.5 dB for the MUX with 200 GHz spacing. This penalty of 0.5 dB includes the dispersion penalty after 20 km and the back-scattered induced penalty. For the MUX with 100 GHz spacing, it was difficult to obtain error-free transmission with a launching power of 20 dBm, as shown in the Fig. 8(b). For the upstream application, the maximum allowable launching power of the broadband light would be limited by the beating noise. Similarly, in the downstream case, the minimum allowable launching power would be limited by the sensitivity of the mPD.

 

Fig. 8 Upstream bit error rate plots using MUX spacing of (a) 200 GHz and (b) 100 GHz

Download Full Size | PDF

4. Summary

We have proposed a self-wavelength initialization method for tunable WBG based external cavity lasers. An additional broadband light source was required in the central office to use the proposed method. We have demonstrated the feasibility of the suggested wavelength initialization method. The transmission performance when the tunable lasers were placed in the central office and in the subscriber region was investigated. The power penalty due to the high power injection was negligible at the back-to-back transmission. However, the additional dispersion penalty after 40 km was measured to be about 0.5 dB and 1 dB, with the MUX channel spacing of 200 GHz, and 100 GHz, respectively. When the tunable lasers are used as a upstream transmitter, the back-scattered deterioration was another reason that the power penalty might increase. The degradations in performance occurred only with the high power broadband light. The minimum allowable output power of the broadband light would be higher than the sum of the detection limit of the mPD and insertion loss between the broadband light source and the mPD. We strongly expect that our novel initialization method will contribute to the cost-effective realization of wavelength tunable WDM-PON in the near future.