## Abstract

The pre-compensation on power fading effect of a colorless laser diode (CLD) carried 40-Gbit/s 256-QAM OFDM transmission during 25-km is demonstrated. By offsetting the DC bias to thrice the threshold (I_{th}) and increasing the injection to 0 dBm, the CLD not only enhances its coherence but also suppresses modulation throughput declination and reduces the relative intensity related noise floor to −50 dBm. Modeling the receiving power of the delivered 256-QAM OFDM subcarriers is established, indicating that raising the bias to 3I_{th} down-shifts the power fading induced notch to 8.8 GHz. This further degrades the OFDM subcarrier peak power by −2.9 dB after 25-km transmission, and the corresponded signal-to-noise ratio (SNR), error vector magnitude (EVM) and bit-error-rate (BER) are 26.1 dB, 4.9% and 6.5 × 10^{−3}, respectively. Pre-leveling the OFDM subcarrier as well as the modulation throughput effectively compromises the over-bias enlarged power fading to promote transmission. With a pre-leveled power slope of 1.5 dB/GHz for 256-QAM OFDM data, the modulation throughput declination of the high biased CLD significantly mitigates under BtB transmission, enabling the receiving sensitivity at −7.2 dBm with SNR, EVM and BER of 29.9 dB, 3.1% and 1.5 × 10^{−4}, respectively. Increasing the pre-leveling slope to 3.2 dB/GHz minimizes the fiber dispersion induced power fading, which improves the receiving SNR, EVM and BER to 27.4 dB, 4.2% and 2.6 × 10^{−3}, respectively, with receiving sensitivity of −3 dBm and power penalty of 4.2 dB after 25-km SMF transmission.

© 2015 Optical Society of America

## 1. Introduction

With rapid increase on demand of real-time multimedia applications, internet service providers are required to demonstrate high-speed and low-cost access networks for delivering multimedia services. At present, the dense wavelength-division-multiplexed passive optical network (DWDM-PON) with high capacity and security has been developed for constructing next-generation optical access networks [1–3]. For cost and complexity reduction of the DWDM-PON, the directly modulated laser transmitter with compact structure and colorless feature has shown its absolute superiority to meet the demand. Up to now, many directly modulated colorless laser transmitters have been proposed [4–9], including the mostly emphasized reflective semiconductor optical amplifier (RSOA) [5–7] and the Fabry-Perot laser diode (FPLD) [8, 9]. The RSOA exhibits wide wavelength coverage to serve multiple DWDM channels [10, 11]; however, it suffers from weak coherence and insufficient modulation bandwidth although which can be extended by high-power injection-locking [12, 13]. Injection-locking the directly modulated FPLD can provide strong coherence and high bandwidth [14–16], but its high end-face reflectance induces strong resonance in few modes to limit the injection-locking efficiency for colorless operation. The appropriate anti-reflection coating on the front-facet of a typical FPLD is a useful solution to approach colorless wavelength selectivity without using abundant injection-locking power, which releases the strong cavity effect to extend the gain spectral linewidth [17, 18]. Therefore, a long-cavity colorless laser diode (CLD, also called weak-resonant-cavity FPLD) with 0.5-2.5% front-facet and 95% rear-facet reflectances was implemented for DWDM-PONs [19, 20]. Such a CLD is assemble in a TO-can package with limited bandwidth of <6 GHz for cost reduction [21]. To effectively use the available bandwidth, a high spectral-efficiency data format which combines quadrature amplitude modulation (QAM) with orthogonal frequency division multiplexing (OFDM) can be employed to modulate the CLD [22, 23]. After injection-locking, the CLD with suppressed relative intensity noise (RIN) further improves for carrying the 16-QAM OFDM data at 20 Gbit/s [24].

Although the injection-locked CLD (hereafter referred as IL-CLD) can provide high-speed OFDM transmission, its receiving power sensitivity is strictly degraded by the dispersion induced power fading effect after long-distant single-mode fiber (SMF) transmission [25]. The power fading effect of the directly modulated laser diode is also correlated with bias current [26, 27] and linewidth enhancement factor [28, 29] which is described as the real to imaginary part ratio of the carrier induced susceptibility variation [30]. Enlarging the bias of IL-CLD concurrently mitigates the RIN and the negative power-to-frequency slope [31–33]; however, such operation strengthens the power fading effect to impact the quality of transmitted OFDM data. A trade-off is set between the power delivering and power fading when optimizing the IL-CLD carried OFDM data after SMF transmitting, and the compromise among bias point, pre-leveling slope and power fading frequency has yet not been surveyed.

In this work, by adjusting bias current and pre-level slope to pre-compensate the fiber dispersion induced power fading, the IL-CLD delivered 256-QAM OFDM transmission in a 25-km SMF at 40 Gbit/s is demonstrated. Optimizing the DC bias (I_{bias}) suppresses the injection-locking induced modulation throughput declination and reduces the RIN introduced high-frequency noises, which concurrently improves the signal-to-noise ratio (SNR) and suppress the bit-error-rate (BER) of the 256-QAM OFDM data. To compromise the power fading strengthened by enlarged bias and dispersion, analytical modeling on the receiving power for the IL-CLD carried 256-QAM OFDM after 25-km SMF transmission is performed. By comparing the BER and SNR of the 25-km SMF transmitted 256-QAM OFDM data under the influence of power fading induced notch on transmission throughput, the appropriate OFDM subcarrier pre-leveling is implemented to pre-compensate the power fading induced degradation on the receiving power. After pre-leveling the OFDM subcarrier, the improvement on BER, SNR and error vector magnitude (EVM) performances of back-to-back (BtB) and 25-km SMF transmissions are compared. The improved power sensitivity and relatively low receiving power penalty are performed.

## 2. Experimental setup

To enable the demanded functionality of covering numerous DWDM channels, the homemade CLD with a low front-facet reflectance of 2% to allow low-throughput-loss injection and a long cavity length of 600 μm to provide efficient modal gain was selected. By using a homemade cooling module, the temperature of the CLD was controlled at 22°C to stabilize the output wavelength. The free-running CLD was biased at 2 times of its threshold current (I_{th} = 22 mA), which reveals a wide gain spectrum ranged from 1530 to 1590 nm with a central wavelength of 1562 nm and longitudinal mode-spacing of 0.6 nm, as shown in Fig. 1(a).

After coherently injection-locking at 0 dBm, the IL-CLD behaves like a single-mode light source with a side-mode suppression ratio (SMSR) as high as 51 dB. If further enlarges the bias current to 3I_{th}, the intense gain competition of unwanted side-modes in the IL-CLD occurs, which conversely reduces the SMSR to 50.1 dB to verify the injection-locking performance. This causes the residual side-modes existed to induce modal dispersion during fiber transmission. Therefore, the optimized bias current is expected to range between 2I_{th} and 3I_{th} for IL-CLD. In addition, the power-to-current response of the CLD is shown in Fig. 1(b), which indicates that if the bias current is enlarged from 2I_{th} to 3I_{th}, the CLD enlarges its output power from 1.7 to 3.3 mW to favor OFDM data encoding. After injection-locking at 0 dBm, the output powers of the IL-CLD at bias points of 2I_{th} and 3I_{th} are slightly enhanced from 1.7 to 1.8 mW and from 3.3 to 3.4 mW, respectively. Such an enlarged stimulated emission slightly releases the lasing criterion of the IL-CLD to suppress its threshold current from 22 mA to 11 mA, which favors the transmitter to avoid from the waveform clipping effect when modulating with 256-QAM OFDM data.

A DWDM-PON testing bench architected for the IL-CLD carried 40-Gbit/s 256-QAM OFDM transmission is shown in Fig. 2. At optical line terminal (OLT), the OFDM data with a total bandwidth of 5 GHz and an OFDM subcarrier number of 214 was exported by an arbitrary waveform generator (AWG, Tektronix, AWG70001A) with a sampling rate of 12 GS/s, which was used to directly encode the CLD. To maximize the modulation depth, a pre-amplifier (Picosecond, Model 5865) with a bandwidth of 12 GHz, a power gain of 26 dB and a noise figure of 5.8 dB was added behind the AWG to adjust the data amplitude. A tunable laser source (TL, HP, 8168F) with various wavelengths from 1450 to 1590 nm and output power of 0 dBm was employed as a master source to injection-lock the CLD after a proper adjustment on polarization, which provides a modal linewidth of 100 kHz for suppressing intensity and phase noises of the slave CLD to improve the transmission performance of the carried QAM OFDM data.

After transmitting through a 25-km single-mode fiber (SMF, Corning SMF-28), the optical OFDM data was received by a photodetector (PD, Nortel, PP-10G) at optical network units (ONUs). To enhance the detection sensitivity, a post-amplifier with a power gain of 18 dB and a noise figure of 8 dB was added behind the photodetector to amplify the received OFDM data. Finally, the amplified OFDM data was real-time resampled by the digital serial analyzer (DSA, Tektronix, DSA71604) with a sampling rate of 100 GS/s, which is analyzed by a homemade MATLAB demodulation program to explore its BER, SNR and EVM performances.

In experiment, an OFDM subcarrier pre-leveling technology is introduced not only to equivalently improve the declined frequency response of the IL-CLD but also to compensate the fiber dispersion caused power fading, which can significantly improve the 25-km SMF transmitted OFDM data quality. Moreover, the pre-leveling technology effectively helps the IL-CLD to extend its usable bandwidth for further OFDM encoding. A schematic diagram of pre-leveling operation for the transmitted OFDM data is shown in Fig. 3. By pre-setting a positive power-to-frequency slope for all electrical 256-QAM OFDM subcarriers, the OFDM subcarrier powers at high-frequency band can be enlarged at a cost of slightly decreasing power at low-frequency band after 25-km SMF transmission, which effectively improves the average SNR performance.

## 3. Results and discussions

#### 3.1 Optimization of the IL-CLD with enhanced throughput power and suppressed RIN

By using IL-CLD as a universal DWDM-PON transmitter for delivering the 256-QAM OFDM data, its allowable modulation bandwidth becomes very important. Figure 4 illustrates the frequency response of the free-running and injection-locked CLD at different bias conditions. After injection-locking at 0 dBm, the throughput power of the IL-CLD biased at 2.0I_{th} is decreased from −46.4 to −53.1 dBm at a modulation frequency of 5 GHz. In addition, the negative power-to-frequency slope always occurs for the IL-CLD [34, 35], which inevitably reduces the throughput intensity. With increasing the bias point of the IL-CLD to 3I_{th}, the gain competition among all lasing modes becomes severe to reduce the injection-locking efficiency of the specified mode, which decreases the throughput power degradation of modulation response from 6.7 to 4.9 dB.

In principle, the RIN is inversely proportional to the relaxation oscillation frequency of a CLD [36]. For the IL-CLD, both increased bias point and enlarged injection power can extend its modulation bandwidth by suppressing its RIN level and up-shifting its RIN peak away from (beyond) the 256-QAM OFDM data bandwidth [24, 37]. To optimize the wavelength detuning of the injection-locking for surveying the RIN properties, the effect of wavelength detuning (Δλ defines as a wavelength difference between master (λ_{M}) and slave (λ_{S}) CLDs with Δλ≣λ_{M}-λ_{S}) on RIN and BER performances of our device is analyzed. The RIN vs. Δλ plot of the slave CLD injection-locked at 0 dBm under different biases is shown in Fig. 5(a). To stabilize the injection-locking of our device biased at 2 I_{th}, the Δλ must be enlarged from 0.13 to 0.17 nm for suppressing the RIN from −102.2 to −104.1 dBc/Hz [38]. This concurrently reduces the received BER of carried 256-QAM OFDM data from from 1.1 × 10^{−2} to 8.5 × 10^{−3}, as shown in Fig. 5(b). However, it still fails to meet the FEC criterion of 3.8 × 10^{−3}. The continuously enlarged Δλ to 0.19 nm deteriorates the injection-locking such that the RIN conversely degrades to −97.1 dBc/Hz with corresponding BER as high as 5.1 × 10^{−2}. Further improvement relies on raising the bias to 3I_{th}. At Δλ of 0.17 nm, the lowest RIN of −104.9 dBc/Hz is observed with optimized BER of 1.2 × 10^{−3}. Hence, at a desired injection power of 0 dBm, the optimized detuning wavelength for the master is Δλ = 0.17 nm for the slave CLD biased at 3I_{th}.

Figure 6(a) shows the RIN spectra of the free-running or the injection locked CLD at different biases, which is measured by a commercial lightwave signal analyzer (HP, 71400C). By enlarging the bias point from 2I_{th} to 3I_{th}, the RIN peak of the free-running CLD is significantly up-shifted from 6.2 to 9.1 GHz and the RIN level can also be suppressed from −98.2 to −101.2 dBc/Hz. After injection-locking at 0 dBm, the RIN peak of IL-CLD can be suppressed to −104.9 dBc/Hz and buried in the noise background with enlarging the bias point from 2I_{th} to 3I_{th}. To verify, the RF spectrum of the 256-QAM OFDM data delivered by the IL-CLD is shown in Fig. 6(b). With increasing the bias point of the IL-CLD from 2I_{th} to 3I_{th}, the RIN related noise floor can be effectively reduced from −44.8 to −50.2 dBm, which is the dominated effect to improve the transmission performance of the 256-QAM OFDM data.

The received waveform of the IL-CLD carried 256-QAM OFDM data is shown in Fig. 7(a). Since the RIN introduces high-frequency noise to affect the time-domain data waveform, the peak-to-peak amplitude of the received 256-QAM OFDM waveform is suppressed from 3.6 to 2.8 volts when enlarging the bias point from 2.0I_{th} to 3.0I_{th}. To obtain the direct evidence of reduced RIN, the decoded SNRs of the back-to-back (BtB) transmitted OFDM data carried by the IL-CLD at different bias conditions is shown in Fig. 7(b). The increased bias from 2I_{th} to 3I_{th} apparently enlarges the received SNR of the high-frequency 256-QAM OFDM data by 4.5 dB at least, and the average SNR is concurrently improved from 25.6 to 28.7 dB. This observation interprets the fact of the enlarged bias level can effectively compensate the injection-locking induced throughput power degradation and suppress the RIN to improve the OFDM transmission performance.

#### 3.2 The fiber dispersion induced power fading effect on the transmission distortion of the 256-QAM OFDM data carried by IL-CLD

During SMF transmission, the fiber dispersion degrades the transmission performance by distorting the temporal waveform of the delivered 256-QAM OFDM data due to the power fading effect, which leads to a notched dip on the transmitted data spectrum and depends on the fiber length and laser linewidth enhancement factor. In principle, the output power of the IL-CLD is expressed as [39]

*η*denotes the differential quantum efficiency,

_{d}*h*the Planck constant,

*ν*the optical frequency,

*q*the electron charge,

*I(t)*the driving current,

*I*the threshold current of the slave CLD,

_{th}*V*the volume of the active region for the CLD, Γ the optical confinement factor,

*v*the group velocity,

_{g}*g’*the differential gain coefficient,

*η*the internal quantum efficiency,

_{i}*τ*the spontaneous carrier lifetime,

_{s}*κ*the injection coupling coefficient which is given by

*κ*=

*v*

_{g}/2

*L*(1-

_{cavity}*R*)/(

_{f}*R*)

_{f}^{0.5}[40] with

*R*defining the front-facet reflectance of the CLD and

_{f}*L*the cavity length,

_{cavity}*α*the linewidth enhancement factor,

*S*the externally injected photon number,

_{inj}*S*the output photon number and

_{B}*I*the threshold current of the IL-CLD. In Eq. (1), the differential quantum efficiency can be expressed as

_{th}’*η*=

_{d}*η*/(

_{i}α_{m}*<α*) with

_{i}> + α_{m}*α*denoting the internal loss and

_{i}*α*the mirror loss which can be simply written as

_{m}*α*/

_{m}= (1*2L*/

_{cavity})ln(1*R*by knowing

_{f}R_{r})*R*/

_{f}*R*as the front/rear-facet reflectance of the CLD. After combining a DC bias current (

_{r}*I*) and an OFDM data to substitute the driving current, the output powers of different OFDM subcarriers can be rewritten as

_{DC}*ν*,

_{n}, i_{n}*f*,

_{subcarrier}*θ*and

_{n}*λ*represent the optical frequency, the modulating amplitude, the frequency, the phase and the optical wavelength of the

_{n}*n*th OFDM subcarrier, respectively. After propagating through a SMF fiber, the fiber dispersion induced power fading effect is introduced to affect the received waveform as well as spectrum of the 256-QAM OFDM data. Without considering the fiber loss, the received power of each OFDM subcarrier in frequency domain can be expressed as [28, 30, 41]

*β*denotes the group velocity dispersion,

_{2}*L*the length of the SMF,

*F*{.} the Fourier transform and

*P*the output powers of the IL-CLD carried OFDM subcarriers in frequency domain. Once the power fading dependent term of “|2

_{n}(f_{subcarrier})*π*|” is equivalent to

^{2}β_{2}Lf^{2}_{subcarrier}-tan^{−1}α*π*/

*2*, the power of OFDM subcarrier would rapidly decline to attenuate the received power. Furthermore, the linewidth enhancement factor

*α*is described as the carrier density induced variation on the refractive index (

*δμ/δn*) divided by the differential gain coefficient (

*g'*), which can be simply expressed as

*α =*(

*-4π/λ*)[(

*δμ/δn*)/

*g'*] [29]. In this case, the differential gain coefficient is given by

*δg*/

_{0}*δn*=

*Vg*/

_{e}*N*(

*t*), where

*g*denotes the empirical gain coefficient and

_{e}*N(t)*the carrier number of the CLD which can be represented as

*N*(

*t*) =

*η*(

_{i}I*t*)

*τ*/

_{s}*q*[38]. Therefore, the

*α*can be rewritten as

Equation (4) clearly interprets that the *α* is positively proportional to the laser driving current. The higher driving current results in the larger the linewidth enhancement factor is. In view of previous works, the *δμ*/*δn* is found to be negative to lead the positive *α* [42]. By substituting Eq. (2) and Eq. (4) into Eq. (3), the received power of each OFDM subcarrier in frequency domain can be derived as

For numerical simulation, all parameters of the IL-CLD and the carried 256-QAM OFDM data are summarized in Table 1 [25]. According to Eq. (4), the bias point dependent power fading response (*P _{fading}*) after 25-km SMF transmission is illustrated in Fig. 8(a).Raising the bias of the CLD from 2I

_{th}to 3I

_{th}strengthens the power fading by down-shifting the notched dip frequency from 9.6 to 8.8 GHz. If such a notched dip frequency moves into the data bandwidth, the received power of the 25-km SMF transmitted-OFDM data at high-frequency region would rapidly drop by the declined power fading response. The RF spectrum of the 25-km SMF transmitted OFDM data at different biases is simulated by Eq. (5) and shown in Fig. 8(b). The RF power degradation of the 25-km transmitted OFDM data at 5 GHz is enlarged from −1.9 to −2.9 dB with enlarging the laser diode bias.

When comparing with the IL-CLD at BtB transmission case, the strong injection induced throughput intensity degradation can be greatly released in the highly biased laser diode, which provides suppressed RIN and enhanced modulation. Unfortunately, the seriously dropped power of high-frequency OFDM subcarriers suffers from the fiber dispersion induced power fading effect, which undoubtedly sets a trade-off between the enlarged biasing and the enhanced power fading for the IL-CLD. To confirm this discussion in experiment, the RF spectra of the IL-CLD carried OFDM data before and after 25-km SMF transmissions at different bias points are compared in Fig. 9.

Despite the simulated power fading distortion, an additional propagating loss of 5.5 dB is introduced for a 25-km SMF transmission. With increasing the bias form 2I_{th} to 3I_{th}, the difference of throughput power loss (ΔP_{throughput}) is illustrates in Table 2. Note that the modulation throughput power loss of 2.4 dB at 4.5 GHz is higher than that of 2 dB at 0.5 GHz, which mainly results from the fiber transmission induced power fading effect.

Owing to the serious power fading, the degraded SNR spectrum of the 25-km SMF transmitted OFDM data delivered by the IL-CLD with enlarging bias is shown in Fig. 10(a). The increasing bias from 2I_{th} to 3I_{th} not only suppresses the RIN but also mitigates the strong injection declined throughput, which eventually enhances the average SNR from 24.8 to 26.1 dB and reduces the corresponded BER from 1.3 × 10^{−2} to 6.5 × 10^{−3} for the 25-km transmitted OFDM data. Even though, the obtained SNR and BER of the 25-km transmitted OFDM data still fails to meet the FEC criterion of 26.9 dB and 3.8 × 10^{−3} [43], respectively. That is, the performance improvement of the 25-km transmitted OFDM data at high-bias condition is still compromised by the fiber transmission induced power fading effect.

When comparing with the BtB case, the power fading induced SNR degradations on the OFDM data after transmitting a 25-km SMF at bias point ranged between 2I_{th} and 3I_{th} are shown in Fig. 10(b). Increasing the bias inevitably induces strong power fading effect after 25-km SMF transmission so as to degrade the SNR from 1 to 4 dB for the OFDM data at high-frequency region. The experimental results are in agreement with the numerical simulations on the performance degradation with the power fading effect. This has set an upper limitation on the I_{bias} of the IL-CLD for carrying the OFDM data even though it must be large enough to achieve sufficient power and bandwidth, the over-bias should be avoided to suppress the strong power fading induced notch within OFDM bandwidth.

#### 3.3 The transmission performance of the IL-CLD carried 256-QAM OFDM data without and with OFDM subcarrier pre-leveling

In more detail, the effect of I_{bias} for the IL-CLD on the SNR of delivered 256-QAM OFDM data is shown in Fig. 11. The BtB transmitted SNR can be improved by simply enlarging the I_{bias}, whereas the 25-km SMF transmitted SNR is rather degraded owing to the high-bias operation enhanced modal dispersion as well as power fading. To release such a power fading effect induced SNR degradation, the OFDM subcarrier pre-leveling operation is employed to pre-emphasize the amplitudes of high-frequency OFDM subcarriers. The suitably high-bias operation can provide sufficient output power to implement the OFDM subcarrier pre-leveling without inducing significant power fading after 25-km SMF transmission. To compare, the SNR spectra of the BtB transmitted 256-QAM OFDM data with the IL-CLD biased at 3I_{th} without and with pre-leveling are shown in Fig. 11(a).

By pre-leveling with a positive slope of *C* = 1.5 dB/GHz (*C = dP/df* in an unit of dB/GHz), the SNR of OFDM subcarriers at high-frequency band (4-5 GHz) can be enlarged by 4.1 dB without affecting that at low-frequency band. The corresponded average SNR and BER performances of the BtB transmitted OFDM data without and with pre-leveling are illustrated in Table 3. In particular, the maintained SNR performance at low-frequency band (DC-1 GHz) is because that the output power of the IL-CLD biased at 3I_{th} can compensate the pre-leveling operation caused OFDM subcarrier amplitude degradation at low-frequency region.

In comparison, the Fig. 11(b) illustrates the 25-km SMF transmitted SNR of the 256-QAM OFDM data carried by the IL-CLD biased at 3I_{th}. Due to the power fading effect, the OFDM subcarrier pre-leveling slope must be modified from 1.5 to 3.2 dB/GHz, which enhances the SNR of OFDM subcarriers at high-frequency region by 3 dB at least, and the average SNR and related BER are shown in Table 3. On the other hand, the SNR at low-frequency region inevitably sacrifices even though the IL-CLD is DC biased as high as 3.0I_{th} to provide sufficient output power. As evidence, the Fig. 12(a) illustrates the BtB and 25-km SMF transmitted constellation plots of the OFDM data carried by the IL-CLD biased at 3I_{th}. By pre-leveling the amplitude of the OFDM subcarriers with a positive slope of 1.5 dB/GHz, the constellation plot of the BtB transmitted OFDM data becomes clearer than that without pre-leveling, which effectively suppresses the EVM from 3.7% to 3.1%. After 25-km SMF transmission, further enlarging the pre-leveling slope from 1.5 to 3.2 dB/GHz also helps to suppress the EVM from 4.9% to 4.2%, which has already passed the FEC criterion of 4.5%.

After all optimizations, the Fig. 12(b) shows the BER of BtB and 25-km SMF transmitted 40-Gbit/s 256-QAM OFDM data without and with OFDM subcarrier pre-leveling. Under BtB transmission without pre-leveling, the lowest BER of 9.0 × 10^{−4} is observed at a receiving power of −2 dBm, which can be further optimized to 1.5 × 10^{−4} by pre-leveling with a positive slope of 1.5 dB/GHz. After 25-km SMF transmission without pre-leveling, the lowest BER is inevitably increased from 9.0 × 10^{−4} to 6.5 × 10^{−3} at the same receiving power of −2 dBm because of the fiber dispersion induced power fading effect, which fails to meet the FEC required BER of 3.8 × 10^{−3}. Nevertheless, such a degraded BER can be improved to 2.6 × 10^{−3} by pre-leveling the amplitude of OFDM subcarriers with a positive slope of 3.2 dB/GHz. Furthermore, to fit the FEC criterion for the case of without pre-leveling, the receiving power sensitivity of the BtB transmitted OFDM data is only −6.4 dBm, which reveals a large receiving power penalty of >5 dB after 25-km SMF transmission. With the aid of OFDM subcarrier pre-leveling, the receiving power sensitivity can be improved from −6.4 to −7.2 dBm after BtB transmission and from >-2 to −3 dBm after 25-km SMF transmission case, which further suppresses the receiving power penalty from >5 to 4.2 dB. Apparently, the OFDM subcarrier pre-leveling technology can help the IL-CLD to deliver the 256-QAM OFDM data by overcoming the power fading induced SNR degradation during 25-km SMF transmission, which successfully extends the usable QAM OFDM data bandwidth to magnify the transmission capacity to 40 Gbit/s and enlarge the spectral usage efficiency to 8 bit/sec/Hz.

## 4. Conclusion

The 40-Gbit/s 256-QAM OFDM transmission with the DWDM-PON demanded IL-CLD is successfully demonstrated by properly adjusting the I_{bias} and pre-leveling the modulation throughput to pre-compensate the fiber dispersion induced power fading in a 25-km SMF. After increasing the bias point to 3I_{th} with an injection-locking power of 0 dBm, the IL-CLD not only enhances its inherent cavity resonance to promote the carrier coherence, but also suppresses the strong injection induced throughput power declination to mitigate its modulation throughput degradation to 4.9 dB within 5 GHz bandwidth. The increased bias current concurrently suppresses the RIN related noise floor to −50.2 dBm to improve the SNR of the OFDM data at high-frequency region. An analytical model for simulating the receiving power of the IL-CLD carried OFDM subcarriers is established to compromise the bias and power fading after 25-km SMF transmission, as the raising bias of the IL-CLD to 3I_{th} inevitably strengthens the power fading effect by down-shifting its notched dip frequency to 8.8 GHz. This enlarges the subcarrier peak power degradation of the 25-km transmitted OFDM data to −2.9 dB at 5 GHz.

With enlarging I_{bias} to 3I_{th}, the average SNR and corresponded BER of the 25-km transmitted OFDM data carried by the IL-CLD can be slightly improved to 26.1 dB and 6.5 × 10^{−3}, respectively; but the over-bias operation induced strong power fading effect still sets an upper limitation on the I_{bias} of the IL-CLD. The OFDM subcarrier pre-leveling operation is thus employed to release the power fading effect by pre-emphasizing the amplitudes of high-frequency OFDM subcarriers. For optimization, the IL-CLD biased at 3I_{th} can provide sufficient output power to implement the 1.5-dB/GHz pre-leveling operation, which delivers the BtB OFDM data with an SNR of 29.9 dB, an EVM of 3.1% and a BER of 1.5 × 10^{−4}. To mitigate the power fading effect, an increased pre-leveling slope of 3.2 dB/GHz is set to enhance the 25-km transmitted OFDM data throughput with an average SNR of 27.4 dB, an EVM of 4.2% and a BER of 2.6 × 10^{−3}. With the aid of the OFDM subcarrier pre-leveling technique, the receiving power sensitivity of BtB and 25-km transmitted OFDM data can be improved to −7.2 and −3 dBm, respectively, thus providing a low receiving power penalty of 4.2 dB in between. This work explores a designer’s rule on selecting the I_{bias} and pre-leveling slope for the DWDM-PON demanded IL-CLD transmitter to compensate the power fading of the carried 256-QAM OFDM data after fiber transmission.

## Acknowledgment

This work was supported by the Ministry of Science and Technology, Taiwan, R.O.C., under grants MOST 101-2221-E-002-071-MY3, MOST 103-2221-E-002-042-MY3 and MOST 103-2218-E-002-017-MY3.

## References and links

**1. **C. A. Brackett, “Dense wavelength division multiplexing networks: principles and applications,” IEEE J. Sel. Areas Comm. **8**(6), 948–964 (1990). [CrossRef]

**2. **L. G. Kazovsky, W.-T. Shaw, D. Gutierrez, N. Cheng, and S.-W. Wong, “Next-generation optical access networks,” J. Lightwave Technol. **25**(11), 3428–3442 (2007). [CrossRef]

**3. **G.-K. Chang, A. Chowdhury, Z. Jia, H.-C. Chien, M.-F. Huang, J. Yu, and G. Ellinas, “Key technologies of WDM-PON for future converged optical broadband access networks [invited],” IEEE/OSA J. Opt. Commun. Netw. **1**(4), 737–758 (2005).

**4. **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. **16**(10), 2380–2382 (2004). [CrossRef]

**5. **E. Wong, K. L. Lee, and T. B. Anderson, “Directly modulated self-seeding reflective semiconductor optical amplifiers as colorless transmitters in wavelength division multiplexed passive optical networks,” J. Lightwave Technol. **25**(1), 67–74 (2007). [CrossRef]

**6. **J. M. Kang and S. K. Han, “A novel hybrid WDM/SCM-PON sharing wavelength for up- and down-link using reflective semiconductor optical amplifier,” IEEE Photon. Technol. Lett. **18**(3), 502–504 (2006). [CrossRef]

**7. **H. Kim, “10-Gb/s operation of RSOA using a delay interferometer,” IEEE Photon. Technol. Lett. **22**(18), 1379–1381 (2010). [CrossRef]

**8. **S.-M. Lee, M.-H. Kim, and C.-H. Lee, “Demonstration of a bidirectional 80-km-reach DWDM-PON with 8-Gb/s capacity,” IEEE Photon. Technol. Lett. **19**(6), 405–407 (2007). [CrossRef]

**9. **J. H. Lee, K. Lee, S. B. Lee, and C. H. Kim, “Extended-reach WDM-PON based on CW supercontinuum light source for colorless FP-LD based OLT and RSOA-based ONUs,” Opt. Fiber Technol. **15**(3), 310–319 (2009). [CrossRef]

**10. **S.-B. Park, D. K. Jung, D. J. Shin, H. S. Shin, I. K. Yun, J. S. Lee, Y. K. Oh, and Y. J. Oh, “Colorless operation of WDM-PON employing uncooled spectrum-sliced reflective semiconductor optical amplifiers,” IEEE Photon. Technol. Lett. **19**(4), 248–250 (2007). [CrossRef]

**11. **Z. Al-Qazwini and H. Kim, “Symmetric 10-Gb/s WDM-PON using directly modulated lasers for downlink and RSOAs for uplink,” J. Lightwave Technol. **30**(12), 1891–1899 (2012). [CrossRef]

**12. **K. Y. Cho, Y. Takushima, and Y. C. Chung, “10-Gb/s operation of RSOA for WDM PON,” IEEE Photon. Technol. Lett. **20**(18), 1533–1535 (2008). [CrossRef]

**13. **H. Kim, “Transmission of 10-Gb/s directly modulated RSOA signals in single-fiber loopback WDM PONs,” IEEE Photon. Technol. Lett. **23**(14), 965–967 (2011). [CrossRef]

**14. **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. **12**(8), 1067–1069 (2000). [CrossRef]

**15. **L. Y. Chan, C. K. Chan, D. T. K. Tong, F. Tong, and L. K. Chen, “Upstream traffic transmitter using injection-locked Fabry-Perot laser diode as modulaotr for WDM access networks,” Electron. Lett. **38**(1), 43–45 (2002). [CrossRef]

**16. **W. Hung, C.-K. Chan, L.-K. Chen, and F. Tong, “An optical network unit for WDM access networks with downstream DPSK and upstream re-modulated OOK data using injection-locked FP laser,” IEEE Photon. Technol. Lett. **15**(10), 1476–1478 (2003). [CrossRef]

**17. **G.-R. Lin, Y.-C. Chi, Y.-C. Li, and J. Chen, “Using a L-band weak-resonant-cavity FPLD for subcarrier amplitude pre-leveled 16-QAM-OFDM transmission at 20 Gbit/s,” J. Lightwave Technol. **31**(7), 1079–1087 (2013). [CrossRef]

**18. **S.-G. Mun, J.-H. Moon, H.-K. Lee, J.-Y. Kim, and C.-H. Lee, “A WDM-PON with a 40 Gb/s (32 x 1.25 Gb/s) capacity based on wavelength-locked Fabry-Perot laser diodes,” Opt. Express **16**(15), 11361–11368 (2008). [CrossRef] [PubMed]

**19. **G.-R. Lin, H.-L. Wang, G.-C. Lin, Y.-H. Huang, Y.-H. Lin, and T.-K. Chen, “Comparison on injection-locked Fabry–Perot laser diode with front-facet reflectivity of 1% and 30% for optical data transmission in WDM-PON system,” J. Lightwave Tehcnol. **27**(14), 2779–2785 (2009). [CrossRef]

**20. **C.-H. Yeh, C.-W. Chow, Y.-F. Wu, S.-P. Huang, Y.-L. Liu, and C.-L. Pan, “Performance of long-reach passive access networks using injection-locked Fabry–Perot laser diodes with finite front-facet reflectivities,” J. Lightwave Technol. **31**(12), 1929–1934 (2013). [CrossRef]

**21. **Y.-C. Chi, Y.-C. Li, and G.-R. Lin, “Specific jacket SMA connected TO-can package FPLD transmitter with direct modulation bandwidth beyond 6 GHz for 256-QAM single or multi-subcarrier OOFDM up to 15Gbit/s,” J. Lightwave Technol. **31**(1), 28–35 (2013). [CrossRef]

**22. **L. J. Cimini Jr., “Analysis and simulation of a digital mobile channel using orthogonal frequency division multiplexing,” IEEE Trans. Commun. **33**(7), 665–675 (1985). [CrossRef]

**23. **N. Cvijetic, D. Qian, J. Hu, and T. Wang, “Orthogonal frequency division multiple access PON (OFDMA-PON) for colorless upstream transmission beyond 10 Gb/s,” IEEE Sel. Areas Commun. **28**(6), 781–790 (2010). [CrossRef]

**24. **M.-C. Cheng, Y.-C. Chi, Y.-C. Li, C.-T. Tsai, and G.-R. Lin, “Suppressing the relaxation oscillation noise of injection-locked WRC-FPLD for directly modulated OFDM transmission,” Opt. Express **22**(13), 15724–15736 (2014). [CrossRef] [PubMed]

**25. **D.-Z. Hsu, C.-C. Wei, H.-Y. Chen, W.-Y. Li, and J. Chen, “Cost-effective 33-Gbps intensity modulation direct detection multi-band OFDM LR-PON system employing a 10-GHz-based transceiver,” Opt. Express **19**(18), 17546–17556 (2011). [CrossRef] [PubMed]

**26. **C. Harder, K. Vahala, and A. Yariv, “Measurement of the linewidth enhancement factor α of semiconductor lasers,” Appl. Phys. Lett. **42**(4), 328–330 (1983). [CrossRef]

**27. **G. P. Agrawal and M. J. Potasek, “Effect of frequency chirping on the performance of optical communication systems,” Opt. Lett. **11**(5), 318–320 (1986). [CrossRef] [PubMed]

**28. **F. Devaux, Y. Sorel, and J. F. Kerdiles, “Simple measurement of fiber dispersion and of chirp parameter of intensity modulated light emitter,” J. Lightwave Technol. **11**(12), 1937–1940 (1993). [CrossRef]

**29. **C. C. Wei, “Small-signal analysis of OOFDM signal transmission with directly modulated laser and direct detection,” Opt. Lett. **36**(2), 151–153 (2011). [CrossRef] [PubMed]

**30. **M. Osinski and J. Buus, “Linewidth broadening factor in semiconductor lasers–an overview,” IEEE J. Quantum Electron. **23**(1), 9–29 (1987). [CrossRef]

**31. **C.-T. Tsai, M.-C. Cheng, Y.-C. Chi, and G.-R. Lin, “A novel colorless FPLD packaged with TO-Can for 30-Gbit/s preamplified 64-QAM-OFDM transmission,” IEEE J. Sel. Top. Quantum Electron. **21**(6), 1500313 (2015). [CrossRef]

**32. **E. K. Lau, X. Zhao, H.-K. Sung, D. Parekh, C. Chang-Hasnain, and M. C. Wu, “Strong optical injection-locked semiconductor lasers demonstrating > 100-GHz resonance frequencies and 80-GHz intrinsic bandwidths,” Opt. Express **16**(9), 6609–6618 (2008). [CrossRef] [PubMed]

**33. **M.-C. Cheng, C.-T. Tsai, Y.-C. Chi, and G.-R. Lin, “Direct QAM-OFDM encoding of an L-band master-to-slave injection-locked WRC-FPLD pair for 28 × 20 Gb/s DWDM-PON transmission,” J. Lightwave Technol. **32**(17), 2981–2988 (2014). [CrossRef]

**34. **E. K. Lau, H. K. Sung, and M. C. Wu, “Frequency response enhancement of optical injection-locked lasers,” IEEE J. Quantum Electron. **44**(1), 90–99 (2008). [CrossRef]

**35. **S.-Y. Lin, Y.-C. Su, Y.-C. Li, H.-L. Wang, G.-C. Lin, S.-M. Chen, and G.-R. Lin, “10-Gbit/s direct modulation of a TO-56-can packed 600-μm long laser diode with 2% front-facet reflectance,” Opt. Express **21**(21), 25197–25209 (2013). [CrossRef] [PubMed]

**36. **Y.-C. Li, Y.-C. Chi, M.-C. Cheng, I.-C. Lu, J. Chen, and G.-R. Lin, “Coherently wavelength injection-locking a 600-μm long cavity colorless laser diode for 16-QAM OFDM at 12 Gbit/s over 25-km SMF,” Opt. Express **21**(14), 16722–16735 (2013). [CrossRef] [PubMed]

**37. **E. K. Lau, L. J. Wong, and M. C. Wu, “Enhanced modulation characteristics of optical injection-locked lasers: a tutorial,” IEEE J. Sel. Top. Quantum Electron. **15**(3), 618–633 (2009). [CrossRef]

**38. **N. Schunk and K. Petermann, “Noise analysis of injection-locked semiconductor injection lasers,” IEEE J. Quantum Electron. **22**(5), 642–650 (1986). [CrossRef]

**39. **L. A. Coldren and S. W. Corzine, *Diode Lasers and Photonic Integrated Circuits* (Wiley, 1995).

**40. **A. Murakam, “Phase locking and chaos synchronization in injection-locked semiconductor lasers,” IEEE J. Quantum Electron. **39**(3), 438–447 (2003). [CrossRef]

**41. **F. Devaux, Y. Sorel, and J. F. Kerdiles, “Simple measurement of fiber dispersion of chirp parameter of intensitymodulated light emitter,” J. Lightwave Technol. **11**(12), 1937–1940 (1993). [CrossRef]

**42. **S. Gozu, T. Mozume, and H. Ishikawa, “Refractive index of Si-doped n-InGaAs,” J. Appl. Phys. **104**(7), 073507 (2008). [CrossRef]

**43. **I. B. Djordjevic, M. Arabaci, and L. L. Minkov, “Next generation FEC for high-capacity communication in optical transport networks,” J. Lightwave Technol. **27**(16), 3518–3530 (2009). [CrossRef]