1. Introduction

Wavelength division multiplexed-passive optical networks (WDM-PONs) have widely been considered as one of the most promising strategies for satisfying the exponentially increasing end-users’ demands for broadband services, as WDM-PONs are capable of offering a large number of excellent features including, for example, high quality data service with guaranteed wide bandwidth, large split ratio, extended transmission reach, aggregated traffic backhauling, simplified network architecture and enhanced end-user privacy [1,2]. For widespread deployment of WDM-PONs, the most critical challenges are cost-effectiveness and flexibility [3].

To achieve cost-effective WDM-PONs, the employment of reflective semiconductor optical amplifiers (RSOAs) in customer optical network units (ONUs) has been adopted, due to their salient advantages such as low component cost, compactness, low power dissipation, full coverage of the entire fibre transmission window and large-scale monolithic integration capability. Recently, use has already been made of RSOAs to achieve a wide range of key WDM-PON functionalities including, for example, intensity signal modulation [4,5], colorless network operation [6] and bidirectional transmission network architectures [7,8].

To enhance the flexibility of WDM-PONs with their compatibility still being preserved with existing time-division-multiplexed PONs (TDM-PONs) to transparently support legacy services, optical orthogonal frequency division multiplexing (OOFDM) [3] has been considered as one of the strongest contenders, since OOFDM has inherent and unique capabilities of providing, in both the frequency and time domains, dynamic bandwidth allocation (DBA) with sub-wavelength granularity to various synchronized end-users. In practice, DBA can be achieved by adaptively varying the total number of OFDM subcarriers assigned to different end-users. OOFDM can also reduce the network complexity due to its resistance to linear dispersion impairments and full use of mature digital signal processing (DSP). In particular, during 2009 we have made significant breakthrough in experimentally demonstrating a series of world-first real-time end-to-end off-the-shelf component-based OOFDM transceivers, which operate at record-breaking net signal bit rates of 1.5Gb/s [9], 3Gb/s [10,11], 5.25Gb/s [12] and 6Gb/s [13] in directly modulated DFB laser-based intensity-modulation and direct-detection (IMDD) transmission systems. These successful demonstrations strongly confirm the feasibility of the OOFDM technique for practical implementation in future optical networks of various architectures.

Compared to OOFDM, adaptively modulated OOFDM (AMOOFDM) can further improve, in a cost-effective manner, signal transmission capacity, network flexibility and performance robustness [14]. Such features are extremely valuable for cost-sensitive WDM-PONs. The key difference between AMOOFDM and OOFDM is that, in AMOOFDM, the modulation format taken on a specific subcarrier within a symbol can be adjusted according to the characteristics of a given transmission system, i.e., a high (low) modulation format is employed on a subcarrier experiencing a high (low) signal-to-noise ratio (SNR). Any subcarriers suffering very low SNRs may be dropped completely to avoid the occurrence of a large number of errors on these subcarriers [14].

Therefore, it is greatly beneficial if use can be made of the advanced AMOOFDM modems incorporating RSOAs as intensity modulators in IMDD single-mode fiber (SMF) transmission systems for WDM-PONs. Based on off-line DSP, experimental results have been reported of the transmission performance of RSOA intensity-modulated AMOOFDM signals over IMDD SMFs [15]. More recently, colorless real-time end-to-end OOFDM transmission at 7.5Gb/s over 25km SMFs has also been demonstrated experimentally using variable power loading and live-optimised RSOA intensity modulators with modulation bandwidths as narrow as 1GHz [16]. However, a number of crucial issues still remain unsolved, which are listed as followings:

Here it is also worth mentioning that, compared to semiconductor optical amplifiers (SOAs), the wide adoption of RSOAs as intensity modulators is mainly due to the fact that RSOAs have lower component cost, higher optical gain, smaller noise figure and larger optical signal extinction ratio [17,18]. However, SOAs exhibit better optical linearity as they have relatively higher input saturation powers. The feasibility of utilizing SOA intensity modulators in AMOOFDM modems have been explored in detail for WDM-PONs [19,20]. It has been shown [20] that colorless 30Gb/s over 60km SMF transmission of SOA intensity-modulated AMOOFDM signals is feasible over a wide wavelength range from 1510nm to 1590nm. Therefore, extensive performance comparisons between RSOA and SOA intensity modulators are also of great importance for practical network designs.

Addressing all the aforementioned challenges forms the main topic of the present paper. In this paper, detailed numerical simulations are undertaken to explore the transmission performance of RSOA intensity-modulated AMOOFDM signals in IMDD SMF systems without in-line optical amplification and chromatic dispersion compensation. The RSOA model adopted here is experimentally verified rigorously using very recently developed real-time OOFDM transceivers at 7.5Gb/s [16]. Special attention is also given to performance comparisons between RSOA and SOA intensity modulators. Optimum RSOA operating conditions are identified, which are independent of RSOA rear-facet reflectivity and very similar to those corresponding to SOAs. As presented in Section 4.3, under the identified optimum operating conditions, the RSOA/SOA intensity modulators support the identical AMOOFDM transmission performance of 30Gb/s over 60km SMFs. This indicates that, without sacrificing the system performance, use can be made of a great diversity of RSOA/SOA intensity modulators in IMDD AMOOFDM systems. As a direct result, the system flexibility and performance robustness can be improved considerably.

Whilst under low-cost optical component-enabled practical operating conditions, RSOA intensity modulators outperform considerably SOA intensity modulators in transmission performance, which decreases significantly with reducing RSOA rear-facet reflectivity and optical input power. For example, for an optical input power of 0dBm, a RSOA intensity modulator with a rear-facet reflectivity value of 0.9 supports 20Gb/s over 100km SMF transmission. The transmission distance is, however, reduced to 60km when an SOA intensity modulator is employed. In addition, simulation results also show that, for low optical input powers, use can be made of the RSOA/SOA intensity modulation-induced negative frequency chirp to improve the AMOOFDM transmission performance in IMDD SMF systems.

2. Transmission system models

2.1 Transmission system and AMOOFDM modems

In Fig. 1 , the transmission system considered here is illustrated, which consists of an AMOOFDM transmitter, an in-line optical amplification-free IMDD SMF link without incorporating chromatic dispersion compensation, and an AMOOFDM receiver.

 

Fig. 1 Transmission system diagram together with block diagrams of the AMOOFDM transmitter and receiver.

Download Full Size | PDF

The AMOOFDM transmitter is composed of an electrical OFDM modem [14,19,20], a RSOA/SOA intensity modulator subject to an injected CW optical wave at a desired optical wavelength and a specific optical power, an optical circulator and a variable optical attenuator. The use of the optical circulator is to separate the modulated AMOOFDM signal from the injected CW optical wave, and to prevent any backward propagating signals from re-entering the intensity modulator. The backward propagating signals may be produced by discrete optical reflection and Rayleigh backscattering in SMFs.

The generation, transmission and detection of the AMOOFDM signals are modeled following procedures similar to those reported in [14,19–22]. The major procedures associated with the transmitter modeling are outlined as followings: adaptive signal modulation format mapping, inverse fast Fourier transform (IFFT), cyclic prefix insertion, AMOOFDM symbol serialization, signal clipping and quantization, as well as Digital-to-Analog Conversion (DAC). The adaptive signal modulation format mapping is performed via negotiations between the transmitter and the receiver at the initial phase of establishing a transmission connection. Depending upon the characteristics of a given transmission system and following the procedure outlined in Section 1, the signal modulation format taken on each subcarrier may vary from differential binary phase shift keying (DBPSK), differential quadrature phase shift keying (DQPSK), 8-quadratic amplitude modulation (QAM) to 256-QAM.

The real-valued electrical signal emerging from the output of the electrical OFDM modem is up-shifted to ensure that each sample has a positive value. The up-shifted electrical signal is then attenuated as necessary and subsequently, together with a DC bias current, employed to directly drive the RSOA/SOA. The optical gain of the RSOA/SOA alters with the electrical current applied, the CW optical waveform injected into the RSOA/SOA is, therefore, modulated accordingly.

It should be noted that, the adjustment of the DC bias current and driving current is to set the RSOA/SOA at desired operating conditions. Such adjustment alters the output power of the modulated optical signal. An optical attenuator is therefore utilized to fix the optical power coupled into the SMF at a desired level.

At the AMOOFDM receiver end, the transmitted optical signal is detected using a square-law photo-detector. The received data is finally recovered following an inverse procedure of the electrical OFDM modem in the transmitter.

2.2 RSOA intensity modulator models

The schematic diagram of the RSOA intensity modulator of a cavity length of L is shown in Fig. 2 . A high reflective coating is applied at its rear-facet and a coating similar to that associated with a conventional SOA is applied at its front facet. Throughout this paper, the RSOA reflectivity, r, is denoted to as the RSOA rear-facet power reflectivity seen from the interior of the semiconductor waveguide. An optical signal is injected into the RSOA at z = 0, and reflected at the rear facet at z = L. $A_{in}(T)$ is the injected optical field propagating forward along the cavity. $A_{out}(T)$ is the modulated backward propagating optical field at z = 0.

 

Fig. 2 Schematic diagram of RSOA intensity modulator.

Download Full Size | PDF

The theoretical RSOA intensity modulator model adopted here is an extension of the theoretical SOA intensity modulator models presented in [19,20], due to the inclusion of rear-facet reflectivity and dynamic optical gain saturation induced by the following factors: 1) counter-propagating modulated AMOOFDM signals; 2) the CW optical input wave properties including powers and wavelengths, and 3) the total electrical currents [23]. In the RSOA intensity modulator model, an assumption made in [19,20] is still adopted, i.e., the impact of various ultrafast intraband dynamic processes including carrier heating, spectral hole-burning, two-photon absorption and ultrafast nonlinear refraction [24] are negligible. The amplified spontaneous emission (ASE) noise imposed by the RSOA is also included and simulated according to the approach presented in [20].

When a forward propagating optical signal is considered only, i.e, r = 0, the RSOA model has the forms identical to those presented in [19,20], and the simplified forms are used to simulate the dynamic characteristics of the SOA intensity modulator. For the transmission systems considered in this paper, the validity of the RSOA intensity modulator model is verified by excellent agreement between theoretical results obtained here and various experimental measurements at both device level [17,18] and system level [15,16]. It should also be noted that the RSOA model still holds well for ADCs/DACs with sampling rates as high as 50GS/s.

2.3 SMF and PIN models

A standard theoretical SMF model is adopted here, in which the effects of loss, chromatic dispersion, and optical power dependence of the refractive index are included. The effect of Kerr nonlinearity-induced phase noise to intensity noise conversion is also incorporated upon photon detection in the receiver.

In the receiver, a square-law photon detector is utilized to detect the optical signals emerging from the transmission system. Shot noise and thermal noise are considered and their effects are simulated following procedures similar to those presented in [25].

2.4 Simulation parameters

In simulating the AMOOFDM modems, the parameters presented in [20] are adopted: the total number of subcarriers is M = 64, in which 31 subcarriers locating in the positive frequency bins are used to carry original data, and one subcarrier close to the optical carrier frequency is dropped. Sampling rates of the DACs/ADCs are taken to be 12.5GS/s in both the transmitter and the receiver. The cyclic prefix parameter defined in [14] is 25%. The optimum bits of quantization and signal clipping ratio are 7-bits and 13dB, respectively [22]. The above-mentioned parameters give a signal bandwidth of 12.5/2 = 6.25GHz in the positive frequency bins, a bandwidth for each subcarrier of 6.25/32 ≈195.3MHz, and a cyclic prefix length of 1.28ns within each symbol having a total time duration of 6.4ns. 1500 AMOOFDM symbols are employed, which, prior to transmission over a SMF link, is oversampled to give a total number of sample points of 512488. It should be pointed out that the use of a relatively large cyclic prefix parameter can ease considerably the symbol synchronization process in real-time end-to-end OOFDM transceiver demonstrations [9–13,16]. As the cyclic prefix also occupies transmission bandwidth, throughout this paper, the net signal bit rate is, therefore, utilized, as defined in Eq. (3).

The parameters used in simulating the RSOA intensity modulator are representative for InGaAsP semiconductor materials at 1550nm, which are listed in Table 1 . For fair performance comparisons between RSOAs and SOAs, in simulating the performance of the SOA intensity modulators, the same set of parameters are also employed, except that r is set to zero for the SOA intensity modulators. In addition, in Table 1, the values of parameters of SMF and PIN detector are also listed, which are identical to those presented in [19, 20]. The optical power coupled into the transmission link is fixed at 6.3dBm [19, 20]. As in typical WDM-PONs nonlinear WDM impairments are negligible and the RSOA/SOA colorless operation capability has also been confirmed [15, 16, 20], a single wavelength of 1550nm is, therefore, considered throughout this paper.

Table 1. RSOA, SOA, SMF and PIN Parameters

View Table

It should be pointed out, in particular, that, to obtain the RSOA/SOA parameters and examine the validity of the developed RSOA/SOA intensity modulator model, as discussed in Section 5, fitting the experimental measurements [16] with the numerical results are undertaken prior to performing numerical simulations at both device and system levels. During the fitting procedures, all the RSOA/SOA parameter values obtained in the experiments are treated as constants, and all the parameter values which are not exactly known are initially taken from the literatures [19, 20] and subsequently adjusted within reasonable limits to obtain the best fit with all the experimental results [16].

3. Optical gain characteristics of RSOAs

To gain an in-depth understanding of the simulated results presented in all the following sections, here brief discussions are first made of the RSOA optical gain characteristics with special emphases being given to optical gain differences between RSOAs and SOAs. Throughout this paper, the optical gains for the RSOA and the SOA are defined as

 

Fig. 3 RSOA/SOA optical gain characteristics under different operating conditions. (a) Optical gain versus optical input power with the bias current being fixed at 100mA. (b)-(d) Optical gain versus bias current for different optical input powers: −10dBm for (b); 10dBm for (c) and 22.5dBm for (d).

Download Full Size | PDF

It can be seen from Fig. 3 that the RSOA and the SOA have similar optical gain evolution trends, except that considerable optical gain differences occur under specific operating conditions. As shown in Fig. 3(a), for optical input powers of <-10dBm, in comparison with the SOA, the RSOA has a much larger optical gain (smaller input saturation power), which increases (decreases) with increasing rear-facet reflectivity value. Such behaviors agree very well with experimental results reported in [17]. Whilst over the strongly saturated optical gain region corresponding to optical input powers of >10dBm, the optical gain differences between the RSOA and the SOA become very small and the RSOA exhibits a lower optical gain. This is because, under such cases, both devices have similar material optical gains of approximately 0dB, corresponding to which the RSOA rear-facet reflection-induced loss becomes pronounced.

Given the central role of the characteristics of optical gain versus electrical current in determining the quality of RSOA/SOA modulated AMOOFDM signals, the gain-current curves for various rear-facet reflectivity values are plotted in Fig. 3(b), Fig. 3(c) and Fig. 3(d) for three representative optical input powers of −10dBm, 10dBm and 22.5dBm. For the −10dBm case, as shown in Fig. 3(b), compared to the SOA, the RSOA exhibits a stiffer gain-current slope occurring in the vicinity of the transparency bias current of approximately 50mA, and the slope difference between the RSOA and the SOA decreases with increasing bias current. This shows excellent agreement with experimental measurements [18]. From the discussions in [19, 20], it is easy to understand that a sharp gain-current slope gives a high extinction ratio of the modulated AMOOFDM signal. Moreover, it can also be seen in Fig. 3(b) that, a small rear-facet reflectivity value brings about a broad linear gain-current region. This implies that, under such conditions, the modulated signals experience the relatively weak RSOA intensity modulation-induced signal clipping effect [19, 20].

On the other hand, as expected from Fig. 3(a), for an optical input power of 10dBm, almost identical gain-current curves are observed in Fig. 3(c) over the positive optical gain region for both the RSOA and the SOA. In particular, very similar gain-current curves between the RSOA and the SOA are observed over both the positive and negative optical gain regions when the optical input power is increased to 22.5dBm, as shown in Fig. 3(d). This suggests that, under the strongly saturated optical gain region, the quality of RSOA modulated AMOOFDM signals is similar to that modulated by SOAs.

4. Optimization of RSOA operating conditions

The aim of this section is two-fold: a) understanding various physical mechanisms underpinning the transmission performance of the RSOA-based AMOOFDM modems in IMDD SMF systems; b) exploring the maximum transmission performance of the modems without considering practical limitations set by cheap components that have been made commercially available.

4.1 Optical input power and bias current optimization

As the RSOA optical gain characteristics depend strongly upon its operation conditions, therefore, it is necessary to identify optimum RSOA operating conditions to maximize the transmission performance of the RSOA-modulated AMOOFDM signals. For a 60km IMDD SMF transmission system, Fig. 4 shows contour plots of signal line rate as a function of CW optical input power and bias current for different RSOA rear-facet reflectivity values. For comparisons, the corresponding performance for the SOA-modulated AMOOFDM signals is also plotted in Fig. 4(d). The driving current with a fixed PTP of 80mA is considered for both the RSOA and SOA cases. In numerical simulations, the signal line rate is calculated using the expression given below:

 

Fig. 4 Contour plots of signal line rate as a function of CW optical input power and bias current for RSOAs with different rear-facet reflectivity values of 0.3 in (a), 0.6 in (b) and 0.9 in (c) and SOAs in (d). An IMDD 60km SMF transmission system is considered.

Download Full Size | PDF

Figure 4 shows that, for a RSOA with any rear-facet reflectivity value, there exist an optimum bias current and an optimum optical input power, corresponding to which a maximum signal line rate is obtained. From discussions in Section 3, it is clear that, over the identified optimum conditions, the RSOA operates at a highly saturated optical gain region. The physical mechanisms behind the occurrence of the optimum operating conditions are the co-existed effects of effective carrier lifetime [26] and extinction ratio of modulated AMOOFDM signals: a large optical input power and/or a high bias current give rise to a short effective carrier lifetime, which corresponds to a large RSOA modulation bandwidth, thus leading to reduced spectral distortions imposed on the modulated AMOOFDM signals. On the other hand, an increase in optical input power and/or bias current also brings about a reduction in extinction ratio of modulated AMOOFDM signals, thus resulting in an increase in minimum optical signal-to-noise ratio (OSNR) required for achieving a specific BER. For optical input powers (bias currents) less than the identified optimum value, the improvement in transmission capacity for high optical input powers (bias currents) is mainly due to the reduction in effective carrier lifetime; whilst when optical powers (bias currents) exceed the optimum value, the reduction in signal extinction ratio becomes a major contributor to the AMOOFDM performance degradation observed in Fig. 4.

As the dependences of effective carrier lifetime and signal extinction ratio upon bias current are not as significant as those upon optical input power [19, 20], thus the optimum bias current occurs over a relatively wide bias current range, as seen in Fig. 4. For bias currents beyond the optimum current range, apart from the effects discussed above, the signal clipping effect associated with RSOA intensity modulation also contributes to the decrease in signal line rate, because the upper part of the electrical driving current applied to the RSOA experiences a flat RSOA optical gain, as seen in Fig. 3(b) and Fig. 3(c).

It is also very important to note in Fig. 4 that, for both the RSOAs and SOAs, the optimum operating conditions and the related maximum signal line rates are very similar, which are independent of RSOA rear-facet reflectivity. Such similarities are a direct result of the almost identical optical gains and gain-current slopes for these two components operating at highly saturated optical gain regions, as shown in Fig. 3(d).

4.2 Optimization of driving current PTP

The impact of driving current PTP defined explicitly in [20] on the maximum AMOOFDM transmission performance is explored in Fig. 5 for different rear-facet reflectivity RSOAs subject to various optical input powers including 22.5dBm (optimum), 10dBm and −10dBm. In obtaining Fig. 5, use is made of the identified optimum bias current of 100mA, and the transmission distance is fixed at 60km.

 

Fig. 5 Signal line rate versus PTP value of driving current for different RSOA rear-facet reflectivity and optical input powers.

Download Full Size | PDF

Figure 5 shows that, under the identified optimum optical input power and bias current, there exists an optimum driving current PTP value of 80mA, regardless of the variation in RSOA rear-facet reflectivity. The occurrence of the optimum driving current PTP is due to the co-existence of the PTP-dependent effects of signal extinction ratio and signal clipping: a high (small) PTP value produces a modulated AMOOFDM signal with a large (small) extinction ratio, which, however, suffers the strong (weak) signal clipping effect, as seen in Fig. 3. The observed rear-facet reflectivity independence of the optimum driving current PTP can be explained by considering Fig. 3(c) and Fig. 3(d), where almost identical gain-current curves under heavily saturated optical gain regions are shown for different rear-facet reflectivity values.

It can also be seen in Fig. 5 that the optimum driving current PTP value decreases with decreasing optical input power. As an example, when the optical input power drops from 22.5dBm to −10dBm, the optimum driving current PTP value reduces from 80mA to 40mA. Comparisons among Fig. 3(b), Fig. 3(c) and Fig. 3(d) indicate that, for a small optical input power, the gain-current curve in the vicinity of the adopted bias current has a stiff slope and corresponds to a relative short linear region. To maximize the transmission performance, a small driving current PTP is, therefore, essential to balance appropriately the effects of signal extinction ratio and signal clipping.

4.3 Capacity versus reach performance

First of all, it is worth mentioning that the previously identified optimum RSOA operating conditions are independent of transmission distance. Based on these optimum operating parameters including a CW optical input power of 22.5dBm, a bias current of 100mA and a driving current PTP of 80mA, the maximum transmission capacity versus reach performance of the RSOA modulated AMOOFDM signals is plotted in Fig. 6 for different rear-facet reflectivity values. It is very interesting to note in Fig. 6 that, under the above-mentioned optimum conditions, the transmission performance of RSOA-modulated AMOOFDM signals is, as expected from above discussions, almost identical to that corresponding to SOA-modulated AMOOFDM signals, and also independent of rear-facet reflectivity over the entire transmission distance range of interest of the present paper. This means that, without sacrificing the system performance, use can be made of a great diversity of RSOAs and/or SOAs as intensity modulators in IMDD AMOOFDM systems. Such a feature offers great opportunities for not only reducing significantly the cost in system installation and maintenance, but also improving considerably the system flexibility and performance robustness. In addition, Fig. 6 also shows that the AMOOFDM modems incorporating the RSOA intensity modulators are capable of supporting 30Gb/s over 60km transmission in IMDD SMF systems without in-line optical amplification and chromatic dispersion compensation.

 

Fig. 6 Signal line rate versus transmission distance for RSOA and SOA intensity modulators operating under identified optimum conditions.

Download Full Size | PDF

On the other hand, when use is made of the OOFDM technique, in which an identical signal modulation format is adopted across all the subcarriers for a given transmission distance, under the above-mentioned optimum RSOA/SOA operating conditions, the simulated maximum transmission capacity versus reach performance is also identical for both intensity modulators and independent of RSOA rear-facet reflectivity. Such behaviors can be easily understood by considering Fig. 3(d). However, owing to the OOFDM-induced imperfect compensation of the signal spectral distortions associated with the RSOA/SOA intensity modulators [14], the OOFDM transmission capacities are slightly lower than those presented in Fig. 6. For example, 29Gb/s over 60km transmission is observed for RSOA/SOA-based OOFDM signals modulated using 64-QAM. Furthermore, if the RSOA/SOA intensity modulator is replaced by an ideal intensity modulator, 38.8Gb/s over 60km transmission of OOFDM signals modulated using 256-QAM is feasible, indicating that the RSOA/SOA-induced nonlinear impairments can decrease the OOFDM transmission capacity by a factor of approximately 1.3. The nonlinear impairments are mainly due to the significant reduction in signal extinction ratio, as shown in Fig. 9(b) .

 

Fig. 9 Effective carrier lifetime (a) and signal extinction ratio (b) versus RSOA rear-facet reflectivity for different optical powers.

Download Full Size | PDF

5. Transmission performance under low-cost optical component-enabled practical operating conditions

The RSOA optimum operating conditions identified in Section 4 require a CW optical input power as large as 22.5dBm. To satisfy such a requirement, a typical semiconductor laser followed by an optical amplifier needs to be incorporated into the RSOA intensity modulator. Clearly, this approach may result in costly RSOA-based AMOOFDM modems.

To enable the cost-effective practical implementation of the RSOA-based AMOOFDM modems in ONUs, use can be made of two promising strategies: a) a single tunable semiconductor laser in each RSOA intensity modulator, and b) a central light source supplied by the central office. The first technical solution makes economic sense because of the availability of commercial tunable semiconductor lasers at a price of a few hundred U.S. dollars. The second technical strategy can enhance the wavelength control functionality of WDM-PONs. Given the fact that, without utilizing optical amplifiers to boost the CW light power in ONUs, these two low-cost solutions are just capable of providing the RSOAs with optical input powers of typically <10dBm, therefore, it is of great importance if detailed explorations of the practically achievable transmission performance of the RSOA-based AMOOFDM modems subject to the aforementioned optical powers provided by low-cost optical components.

5.1 Verification of the RSOA intensity modulator model at device and system levels

As experimental results are available for low optical powers only, here effort is first made to verify the validity of the developed RSOA intensity modulator model at both device and system levels, by comparing simulated results with experimental measurements using real-time end-to-end OOFDM transceivers in a system configuration similar to Fig. 1 [16]. To ensure fair comparisons, for this comparison only, the OOFDM modem parameters identical to those used in the experiments are adopted, which include 32 subcarriers in total, 25% cyclic prefix, 14.5dB signal clipping ratio, and 8-bit DAC/ADC operating at 4GS/s, as well as 16-QAM taken on all the 15 information-bearing subcarriers (whose powers are adjusted to compensate for the system frequency response roll-off effect [16]). The above parameters give a signal line rate of 6Gb/s (a raw signal line rate of 7.5Gb/s). In addition, the system parameters used in the numerical simulations are also identical to those adopted in the experiments [16]. These system parameters are listed as followings: an electrical OFDM signal with a PTP of 42mA, a DC bias current of 84mA, a CW optical input power of 5dBm, a 25km SMF, a received optical signal power of −5.5dBm and a receiver sensitivity of −17dBm. All other parameters that are not explicitly mentioned above are listed in Table 1.

RSOA frequency response comparisons between the numerical results and experimental measurements are made in Fig. 7(a) , where excellent agreements between these two cases are observed over the entire signal spectral region of 0-2GHz. It can also be found in Fig. 7(a) that, the RSOA intensity modulator has a 3dB modulation bandwidth of approximately 1.25GHz, which is mainly determined by the low CW optical input power-induced long effective carrier lifetime.

 

Fig. 7 Comparisons between simulations and real-time experimental measurements. (a) RSOA intensity modulator frequency response, and (b)-(g) constellations of representative subcarriers. (b)-(d) are simulated results and (e)-(g) are experimental results.

Download Full Size | PDF

By adopting variable subcarrier powers similar to those reported in [16], and after transmission through 25km SMF, the representative subcarrier constellations recorded prior to channel equalization in the receiver, are also presented in Fig. 7, where the simulated constellations are shown in Figs. 7(b)-7(d) and the corresponding experimentally measured constellations are shown in Figs. 7(e)-7(g). Once again, the simulated results agree very well with the experimental measurements. In addition, in comparison with the experimental measurements, for the same subcarrier the numerical simulations also give very similar constellation rotation, which increases with increasing subcarrier frequency. The constellation rotation occurs due to the phase shift induced by fibre chromatic dispersion. For high frequency subcarriers, the simulated constellation sizes are, however, larger than those measured in the experiments. This is due to the fact that the numerical simulations exclude the DAC/ADC-induced frequency response roll-off effect [13, 16], which introduces extra losses to high frequency subcarriers, for example, a loss of 8dB for the15-th subcarrier. As a direct result of the spurious points circled in Fig. 7(f) and Fig. 7(g), the simulated power penalty at a BER of 1.0 × 10⁻³ is slightly smaller than that measured in the experiments [16].

Comparisons are also made between the simulated results and experimental measurements using off-line DSP in a different system configuration based on adaptive modulation [15]. The parameters adopted in numerical simulations, for this comparison only, are identical to those used in the experimental measurements [15]. These parameters are: a CW optical input power of −10dBm at the input facet of the RSOA, a −1dBm optical power coupled into the transmission link, and a −19dBm optical power at the input facet of the photon detector. Our numerical simulations show that, a 10Gb/s AMOOFDM signal transmission over a 20km IMDD SMF link involving a RSOA intensity modulator is obtainable, which agrees very well with the experimental measurements [15].

5.2 Optical input power and rear-facet reflectivity dependent transmission performance

The dependence of the achievable signal line rate upon rear-facet reflectivity of a RSOA subject to three representative optical input powers of 10dBm, 0dBm and −10dBm is shown in Fig. 8 , where the bias current, the driving current PTP and the transmission distance are fixed at 100mA, 40mA (according to Fig. 5) and 60km, respectively. For performance comparisons between the optimum and practical operating conditions, the transmission performance achieved under the optimum operating conditions (22.5dBm optical input power), are also presented in Fig. 8, in which, once again, the corresponding transmission performance of the SOA-based intensity modulator is plotted for all the cases considered.

 

Fig. 8 Signal line rate versus rear-facet reflectivity value of RSOA subject to different optical input powers.

Download Full Size | PDF

It is shown in Fig. 8 that, as expected from Fig. 4, the transmission performance drops quickly with decreasing optical input power. More importantly, in comparison with the SOA intensity modulator, the RSOA intensity modulator improves the AMOOFDM transmission performance, and such performance enhancement is more pronounced for high RSOA rear-facet reflectivity values.

The physics underpinning the superiority of the RSOA intensity modulator to the SOA intensity modulator is the combination between the reduced effective carrier lifetime and the enhanced signal extinction ratio. This can be understood by considering Fig. 9. In obtaining Fig. 9, the simulation parameters identical to those adopted in Fig. 8 are considered. Figure 9(a) shows that, compared to the SOA, the effective carrier lifetime of the RSOA is much shorter, thus giving rise to a wider modulation bandwidth. This is because, for an optical input power of <10dBm, the RSOA has a higher optical gain (as seen from Fig. 3(a)), thus leads to a larger optical output power, which is inversely proportional to the effective carrier lifetime [19,20,26].

The RSOA-induced enhancement in signal extinction ratio shown in Fig. 9(b) is in good agreement with experimental measurements [17]. Such enhancement is due to a small optical input power-induced stiff slope of the gain-current curve, as shown in Fig. 3(b). Furthermore, for a specific optical input power, a high RSOA rear-facet reflectivity leads to a large optical gain [Fig. 3(a)] and a slightly reduced slope of the gain-current curve [Fig. 3(b)]. As a direct result, both the RSOA effective carrier lifetime and the signal extinction ratio decrease with increasing rear-facet reflectivity, as seen in Fig. 9. For the practical optical input power range, it can be worked out easily from Fig. 9(a) that the RSOA modulation bandwidth is smaller than (or comparable to) the transmitted signal bandwidth. This implies that, in comparison with the rear-facet reflectivity dependent signal extinction ratio presented in Fig. 9(b), the large rear-facet reflectivity-induced reduction in effective carrier lifetime, as seen in Fig. 9(a), is the dominant physical mechanism underpinning the improved signal capacity for high rear-facet reflectivity values, as seen in Fig. 8. This is because a short effective carrier lifetime corresponds to a large RSOA modulation bandwidth and thus an increase in signal line rate.

It can also be seen in Fig. 9 that, for an optical input power of −10dBm and a RSOA rear-facet reflectivity value of 0.3, the RSOA outperforms the SOA in terms of both effective carrier lifetime and signal extinction ratio, but Fig. 8 shows that the RSOA intensity modulator supports the transmission performance slightly worse than that corresponding to the SOA intensity modulator. This can be explained by considering the fact that, in comparison with the SOA running at such an optical input power, the RSOA has a significantly higher optical gain, as shown in Fig. 3(a), thus resulting in a dramatically larger frequency chirp.

5.3 Capacity versus reach performance

Figure 10 shows the signal line rate versus reach performance of the RSOA/SOA intensity-modulated AMOOFDM signals for different practical optical input powers and rear-facet reflectivity values. In simulating Fig. 10, the parameters identical to those used in Fig. 9 are adopted. It is very interesting to note in Fig. 10 that, in comparison with the SOA intensity modulators, the RSOA intensity modulators subject to injected optical powers of >-10dBm are capable of improving the signal line rate over the entire transmission distance range including both the chromatic dispersion-dominant performance region (<100km) and the loss-dominant performance region (>100km) [14]. In addition, it can also be seen in Fig. 10 that, the use of RSOA intensity modulators is more effective in the loss-dominant performance region, as the RSOA enhanced signal extinction ratio can offset, to some extent, the transmission link loss.

 

Fig. 10 Signal line rate versus reach performance for different optical input powers and rear-facet reflectivity values.

Download Full Size | PDF

Moreover, Fig. 10 also indicates that a RSOA having a large rear-facet reflectivity is always preferred, especially in the dispersion-dominant performance region. As an example, for an optical input power of 0dBm, an increase in RSOA rear-facet reflectivity from 0.3 to 0.9 can extend 20Gb/s AMOOFDM signal transmission distance from 60km to 100km. More specifically, for a 5dBm optical input power adopted in a typical WDM-PON, RSOA rear-facet reflectivity values of 0, 0.3, 0.6 and 0.9 support AMOOFDM signal capacities of 22.0Gb/s, 22.2Gb/s, 23.4Gb/s and 24.4Gb/s, respectively, in IMDD 60km SMF systems. This confirms strongly the importance of the RSOA rear-facet reflectivity in determining the quality of the RSOA intensity modulated AMOOFDM signals under the practical operating conditions.

In comparison with the AMOOFDM case, the effectiveness of utilizing RSOAs with high rear-facet reflectivity values in improving the system performance is more pronounced when OOFDM is adopted. For example, numerical simulations indicate that, for a 5dBm optical input power and a 60km transmission distance, a RSOA rear-facet reflectivity variation from 0.3 to 0.6 results in an increase in OOFDM signal capacity from 14.5Gb/s (8-QAM) to 19.4Gb/s (16-QAM). From the above discussions, it is clear that, in comparison with the SOA intensity modulators, for the practical operating conditions, the RSOA intensity modulators are capable of offering reduced performance penalties associated with intensity modulation.

5.4 Impact of negative frequency chirp

Apart from cost-effectiveness, another benefit of employing a RSOA operating at low optical input powers is that it can produce a fair amount of controllable negative frequency chirp, which has an opposite sign compared to the dispersion parameter of a standard SMF. Therefore, use can be made of such property to improve either the transmission capacity for a fixed link power budget, or the link power budget for a fixed transmission capacity.

OOFDM has strong resilience to chromatic dispersion in both coherent and IMDD transmission systems. In addition to that, the above-mentioned “dispersion compensation” approach can further improve the AMOOFDM transmission performance in IMDD transmission systems. This is because of square-law photon detection in the receiver, which takes effects via the following two physical mechanisms: 1) Square-law photon detection cannot preserve perfectly the chromatic dispersion-induced optical phase shift in the electrical domain. The “dispersion compensation” approach can reduce the phase variation, thus leading to the improved AMOOFDM transmission performance. 2) A small optical signal phase also reduces the unwanted subcarrier intermixing effect, which occurs upon square-law photon detection in the receiver [21]. The effectiveness of the “dispersion compensation” approach has been confirmed experimentally in [10, 12], where a negative power penalty of −2dB (−0.5dB) has been observed when a 3Gb/s (5.25Gb/s) 16-QAM (128-QAM)-encoded real-time OOFDM signal is transmitted over a 75km (25km) MetroCor SMF IMDD system involving a directly modulated DFB laser (DML). In such system configurations, the DMLs impose positive frequency chirps onto the modulated optical signals, which can be compensated by negative dispersion MetroCor fibres. Here it is also worth pointing out that the resulting negative power penalty is independent upon the length of the cyclic prefix adopted.

To demonstrate the effectiveness of the aforementioned dispersion compensation approach for RSOA/SOA intensity modulated AMOOFDM signals in IMDD standard SMF transmission systems, in Fig. 11 performance comparisons are made for the cases of including and excluding chromatic dispersion. In Fig. 11 the optical input power is taken to be −10dBm. It can be seen in Fig. 11 that the “dispersion compensation” approach is capable of enhancing the transmission capacity over standard SMFs of up to 100km. It should be pointed out that, the RSOA negative frequency chirp is a function of operating conditions, suggesting that such dispersion compensation is dynamically controllable.

 

Fig. 11 Signal line rate versus reach performance for the cases of including and excluding chromatic dispersion. Optical input power is fixed at −10dBm.

Download Full Size | PDF

6. Conclusions

Extensive numerical simulations have been undertaken of exploring the transmission performance of RSOA modulated AMOOFDM signals over IMDD SMF transmission systems without in-line optical amplification and chromatic dispersion compensation for WDM-PONs. The RSOA theoretical model adopted in the paper for modulating AMOOFDM signals has been verified rigorously at both device and system levels. Detailed performance comparisons have also been made between RSOA and SOA intensity modulators. Optimum RSOA operating conditions have been identified, which are independent of RSOA rear-facet reflectivity and very similar to those corresponding to SOAs.

Under the identified optimum operating conditions, both RSOA and SOA intensity modulators support identical AMOOFDM transmission performances of 30Gb/s over 60km SMFs. Under low-cost optical component-enabled practical operating conditions, the RSOA intensity modulators with rear-facet reflectivity values of >0.3 outperform considerably the SOA intensity modulators in transmission performance, which increases significantly with increasing RSOA rear-facet reflectivity and optical input power. In addition, simulations also show that, for low optical input powers, RSOA/SOA intensity modulation-induced negative frequency chirp can be used to improve the AMOOFDM transmission performance in IMDD SMF systems.