Open Access
Add to BibSonomyAbstract
Detailed numerical investigations are undertaken of wavelength reused bidirectional transmission of adaptively modulated optical OFDM (AMOOFDM) signals over a single SMF in a colorless WDM-PON incorporating a semiconductor optical amplifier (SOA) intensity modulator and a reflective SOA (RSOA) intensity modulator in the optical line termination and optical network unit, respectively. A comprehensive theoretical model describing the performance of such network scenarios is, for the first time, developed, taking into account dynamic optical characteristics of SOA and RSOA intensity modulators as well as the effects of Rayleigh backscattering (RB) and residual downstream signal-induced crosstalk. The developed model is rigorously verified experimentally in RSOA-based real-time end-to-end OOFDM systems at 7.5Gb/s. It is shown that the RB noise and crosstalk effects are dominant factors limiting the maximum achievable downstream and upstream transmission performance. Under optimum SOA and RSOA operating conditions as well as practical downstream and upstream optical launch powers, 10Gb/s downstream and 6Gb/s upstream over 40km SMF transmissions of conventional double sideband AMOOFDM signals are feasible without utilizing in-line optical amplification and chromatic dispersion compensation. In particular, the aforementioned transmission performance can be improved to 23Gb/s downstream and 8Gb/s upstream over 40 km SMFs when single sideband subcarrier modulation is adopted in the downstream systems.
©2010 Optical Society of America
1. Introduction
Wavelength division multiplexed-passive optical networks (WDM-PONs) have been widely considered as a promising “future-proof” strategy for deployment in broadband access networks, since WDM-PONs have a 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, enhanced end-user privacy and protocol transparency [1]. However, at present WDM-PONs are not regarded as a cost-effective solution, as the optical transmitter in the customer optical network unit (ONU) requires an optical carrier source precisely aligned with a specifically allocated WDM grid.
To make WDM-PONs commercially viable, spectrally-sliced centralized low-cost light sources of various types in the optical line termination (OLT) have been utilized to convey upstream data for achieving colorless ONUs. Those light sources include, for example, light-emitting diodes (LEDs) [2], Fabry-Pérot (FP) lasers [3] and amplified spontaneous emission (ASE) generated by an erbium-doped fiber amplifier (EDFA) [4]. More importantly, downstream wavelength reuse [5–8] has also attracted overwhelming research interest, as it does not require any extra light source in the ONU except the downstream optical signal itself. In wavelength reused WDM-PONs, a fraction of the downstream optical signal is fed into a reflective semiconductor optical amplifier (RSOA) intensity modulator for upstream data re-modulation in the ONU. Clearly, wavelength reuse improves both the cost-effectiveness and wavelength control functionalities of WDM-PONs.
Apart from the well known Rayleigh backscattering (RB) effect, the major challenge of wavelength reused WDM-PONs is how to suppress effectively crosstalk due to residual downstream optical signal-induced upstream signal fluctuations. To reduce the crosstalk effect, use can be made of a number of approaches listed as followings: a) an optical gain saturated SOA-based data eraser [5] to suppress the downstream optical signal prior to being re-modulated for transmitting upstream data; b) feed-forward current injection [6] in a RSOA intensity modulator to smooth the residual downstream optical signal waveform. Alternatively, use can also be made of different signal modulations for downstream and upstream signals. For example, downstream/upstream signal modulations can be frequency-shift keying (FSK)/on-off keying (OOK) [1], differential phase-shift keying (DPSK)/OOK [7], inverse return-to-zero (IRZ)/OOK [8], respectively. In all the above-mentioned signal modulation techniques, the constant downstream FSK and DPSK signal waveforms can reduce the crosstalk effect imposed on the RSOA intensity modulated upstream signals.
It is well known that optical orthogonal frequency division multiplexing (OOFDM) [9–19] has intrinsic nature, i.e., a “noise-like” time domain signal waveform [9] and a small signal extinction ratio [10], and that OFDM has inherent and unique advantages including, for example, dynamic provision of hybrid bandwidth allocation in both the frequency and time domains, significant reduction in network complexity and great potential for cost-effective implementation. In addition, compared to OOFDM, adaptively modulated OOFDM (AMOOFDM) can further improve signal transmission capacity, network flexibility and performance robustness [9]. Therefore, it is greatly beneficial if use is made of AMOOFDM in wavelength reused bidirectional transmission over a single single-mode fibre (SMF) for WDM-PONs. Unfortunately, in all the previously published works, AMOOFDM (OOFDM) has been utilized separately in upstream [15,17] (downstream [16,19]) transmission only. Apart from that, a key network design issue still remains unsolved, i.e., the impact of the effects of RB noise and crosstalk on the transmission performance of downstream and upstream AMOOFDM signals in wavelength reused bidirectional transmission WDM-PONs.
Moreover, given the fact that SOAs have many merits such as compactness, monolithic integration capabilities and colorless operation over the entire optical fiber communication window, in addition to the RSOA intensity modulator in the ONU, the use of SOA intensity modulators in the OLT [10] is also preferred to expensive external modulators. Such a SOA-based OLT and RSOA-based ONU configuration is capable of providing colorless WDM-PON operation (rather than just colorless ONUs), thus leading to a significant reduction in the installation and maintenance cost.
In this paper, detailed numerical investigations are undertaken of wavelength reused bidirectional transmission of AMOOFDM signals in WDM-PONs incorporating a SOA and RSOA intensity modulator in the OLT and ONU, respectively. A comprehensive theoretical model describing the performance of such network architecture is, for the first time, developed, taking into account the dynamic optical characteristics of the SOA and RSOA intensity modulators as well as the effects of RB noise and crosstalk. The developed model is verified experimentally in RSOA-based end-to-end real-time OOFDM intensity modulation and direct detection (IMDD) systems at 7.5Gb/s. It is shown that the RB noise effect and the residual downstream optical signal-induced crosstalk are the dominant factors limiting the maximum achievable downstream and upstream transmission performances. Under optimum SOA and RSOA operating conditions as well as practical downstream and upstream optical launch powers, 10Gb/s downstream and 6Gb/s upstream over 40km SMF transmission of conventional double sideband (DSB) AMOOFDM signals are feasible. In particular, the aforementioned transmission performances can be improved to 23Gb/s downstream and 8Gb/s upstream over 40 km SMF when single sideband (SSB) subcarrier modulation (SSB-SCM) is introduced in the downstream systems.
2. Theoretical models
2.1 Wavelength-reused bidirectional transmission WDM-PON architecture and AMOOFDM modems incorporating SOA and RSOA intensity modulators
Figure 1 illustrates the wavelength-reused bidirectional transmission colorless WDM-PON architecture considered here, which consists of a SOA intensity modulator-based AMOOFDM downstream link and a RSOA intensity modulator-based AMOOFDM upstream link, both of which share the same optical wavelength and a single optical amplification- and dispersion compensation-free IMDD SMF system. The downstream link is composed of an AMOOFDM transmitter, a SMF, a square-law photon detector and an AMOOFDM receiver. The downstream AMOOFDM transmitter includes an OFDM downstream transmitter, a laser diode (LD), an SOA intensity modulator and a variable optical attenuator (VOA). In the downstream OFDM transmitter, the generation of a real-valued electrical OFDM signal is modeled following a procedure similar to that described in [9–11]. The major procedure includes signal modulation format mapping, inverse fast Fourier transform (IFFT), cyclic prefix insertion, OFDM symbol serialization and digital-to-analog conversion (DAC). The modulation format taken on each subcarrier within a symbol varies from differential binary phase shift keying (DBPSK), differential quadrature phase shift keying (DQPSK), 8-quadrature amplitude modulation (QAM) to 256-QAM. Generally speaking, through negotiations between the downstream OFDM transmitter and downstream OFDM receiver, a high (low) modulation format is used on a subcarrier suffering a high (low) signal to noise ratio (SNR). Any subcarrier experiencing a very low SNR may be dropped completely if there are still a large number of errors occurred even when the lowest modulation format is employed. By combining an optimized DC bias current, the real-valued electrical OFDM signal from the downstream OFDM transmitter is up-shifted to ensure that each sample has a positive value. Then the up-shifted electrical OFDM signal drives directly the SOA to modulate an injected optical CW wave by varying the SOA optical gain [10,11]. Finally, the SOA intensity modulated AMOOFDM downstream signal is coupled and multiplexed, via a variable optical attenuator, an optical circulator and a multiplexer, into a standard SMF.
Fig. 1 Wavelength-reused bidirectional transmission WDM-PON architecture with SOA intensity modulated downstream AMOOFDM signals and RSOA intensity modulated upstream AMOOFDM signals.
After transmitting through the SMF, the received downstream AMOOFDM signal is de-multiplexed and subsequently split into two optical beams: the first one is detected using a square-law photon detector, and data is recovered in the downstream OFDM receiver having an inverse procedure of the downstream OFDM transmitter.
On the other hand, the second optical beam is utilized directly as an optical carrier of the upstream AMOOFDM signal. Via an optical circulator, the second optical beam is injected into a RSOA intensity modulator, which is driven by a real-valued up-shifted and electrically amplified OFDM signal generated by the upstream OFDM transmitter. In the RSOA intensity modulator, re-modulation of the injected downstream optical signal takes place via the variation of the RSOA optical gain by the upstream electrical OFDM signal [13,15]. The produced upstream AMOOFDM signal is coupled into the same SMF link compared to the downstream AMOOFDM signal. Here, an optical attenuator is employed to fix the coupled upstream optical power at a specific level. After upstream transmission through the SMF, the received upstream AMOOFDM signal is detected by a square-law photon detector and recovered by an upstream OFDM receiver having an inverse procedure of the corresponding upstream OFDM transmitter.
From the above description, it is clear that, the residual downstream optical signal introduces waveform fluctuations into the RSOA modulated upstream AMOOFDM signal. Throughout this paper, such a residual downstream signal-induced waveform distortion to the upstream optical signal is referred to as crosstalk, which degrades the upstream AMOOFDM signal transmission performance. In addition, due to wavelength re-used bidirectional transmission over the same SMF link, the RB noise effect also affects considerably both the upstream and downstream transmission performances.
2.2 SOA/RSOA intensity modulators
In this paper, comprehensive SOA and RSOA intensity modulator models developed in [10,11] and [13], respectively, are employed to describe the dynamic optical characteristics of the SOA/RSOA intensity modulated AMOOFDM signals. Given the fact that the SOA intensity modulator model has already been verified and successfully applied in various system architectures [10,11], here, special attention is focused on examining the validity of the RSOA intensity modulator model only. In Section 3.1, extensive comparisons of the transmission performance of RSOA intensity modulated optical signals between simulated results and experimental measurements using real-time end-to-end OOFDM transceivers are made at both device and system levels, and excellent agreements between these two cases are observed.
2.3 RB noise
It is well known [20–22] that, both RB noise and discrete reflection induced by passive components such as optical connectors may affect the system performance in bidirectional transmission systems. However, discrete reflection can be reduced to a level of 55 dB lower than that corresponding to the signal power, provided that the optical connectors with oblique end faces are applied [22]. For simplicity, in this paper, the discrete reflection effect is not considered. In the WDM-PON architecture of interest of the present paper, the RB noise effect is, however, unavoidable, which mainly stems from two major sources:
- • Downstream AMOOFDM signals. The associated RB noise affects upstream AMOOFDM signals.
- • Upstream AMOOFDM signals. A fraction of the associated RB noise joins the downstream AMOOFDM signals, and the remaining part can be amplified and reflected by the RSOA intensity modulator, and finally joins the upstream signals.
Moreover, in simulations, attenuation of downstream and upstream signals due to Rayleigh scattering is also included into the SMF attenuation coefficient.
First, we analytically derive the total RB noise power imposed on the upstream optical signal. For a downstream optical signal power launched into the SMF, , which generates a upstream RB noise power at the output facet of the SMF, , is given by [20]
where with [km−1] being the fiber scattering coefficient, S being the fiber recapture coefficient, and μ [km−1] being the SMF attenuation coefficient. L is the fiber length. Depending upon the type of fibre and the selected wavelength, for fiber links longer than 20km, converges towards an almost constant value [21,22].Given the importance of ζ in determining the RB noise effect and for identifying an optical power threshold below which the stimulated Brillouin scattering (SBS) effect is negligible, at 1550nm experimental measurements are undertaken in a system setup illustrated in Fig. 2(a) . The measured backscattered optical power and corresponding ζ as a function of CW launch power are plotted in Fig. 2(b) for different standard SMF lengths ranging from 25km to 125km. It can be seen from Fig. 2(b) that, for optical launch powers of approximately <7dBm, regardless of fibre length, ζ can be taken to be −34dB, which is adopted in numerical simulations presented in this paper. Whilst when CW launch powers exceed 7dBm, the SBS effect plays a dominant role in determining the backscattered optical power measured. Therefore, throughout this paper, both downstream and upstream launch powers are limited to values of <7dBm.
Fig. 2 (a) Experimental setup; and (b) Measured backscattered optical power and ζ versus optical launch power for different SMF lengths.
The RB noise power contributed by the upstream optical signal and imposed on the upstream signal can be expressed as
where is the upstream optical signal power launched into the SMF. is the total optical gain of the upstream AMOOFDM transmitter, and is defined asConsidering Fig. 1 and Eq. (3), it is clear that, includes RSOA optical gain, splitter loss (−3dB), optical circulator insertion loss and optical attenuator loss. Based on Eqs. (1)-(3), the optical signal to RB noise ratio (OSRNR) of the upstream signal at the output facet of the upstream link can be expressed asIt can be derived easily from Eq. (4) that, a maximum OSRNR value occurs when , corresponding to which, according to Eq. (3), the downstream and upstream launch powers are identical.Similarly, the RB noise power imposed on the downstream signal is given by
Thus the OSRNR of the downstream signal at the output facet of the downstream link can be written asIt is worth mentioning that the RB noise has a colored power spectral density (PSD), which is proportional to the PSD of the corresponding optical signal, as shown in Fig. 3 . In numerical simulations, the RB noise spectrum is discretised at a frequency resolution, and for a given transmission link, the RB noise power in each frequency interval is calculated, in the frequency domain, by using Eqs. (1), (2) and (5), and taking into account the corresponding spectra of the involved optical signals. The resulting frequency dependent RB noise power is shown in Fig. 3 by the red triangle arrows. A white Gaussian noise within each frequency interval is then generated, as shown in Fig. 3(blue graph), and the generated RB noise is finally added into the received optical signals in the frequency domain. As the RB noise PSD varies with both the downstream and/or upstream optical signals, in our numerical simulations, the RB noise PSD is, therefore, updated when a different transmission link involving different downstream and upstream signals is considered. The above-mentioned treatment has been verified by good agreements between numerical simulation results and experimental measurements.2.4 SMF and PIN detector
A theoretical SMF model successfully used in [9–11,13,14] is adopted here, in which the effects of loss, chromatic dispersion, and optical power dependence of refractive index are taken into account. The effect of fiber nonlinearity-induced phase noise to intensity noise conversion is also considered.
As already mentioned, in the downstream and upstream AMOOFDM receivers, the received optical signals are detected using the square-law photon detectors, in which shot noise and thermal noise are included. The shot and thermal noises are simulated following procedures similar to those presented in [23].
2.5 Simulation parameters
In simulating all the OFDM transceivers shown in Fig. 1, parameters identical to those presented in [10] are adopted: the total number of subcarriers is M = 64, in which 32 subcarriers in the positive frequency bins are used to carry original data with one subcarrier close to the optical carrier frequency being dropped completely. The sampling rates of the DAC/ADC are taken to be 12.5GS/s. The cyclic prefix parameter defined in [9] is 25%. The optimum number of quantization bits and the optimum signal clipping ratio is 7-bits and 13dB, respectively [24]. The above-mentioned parameters give a signal bandwidth in the positive frequency bins of 12.5/2 = 6.25GHz, a bandwidth for each subcarrier of 6.25/32 ≈195.3MHz, and a cyclic prefix length of 1.28ns within each OFDM symbol having a total time duration of 6.4ns. In numerical simulations, 1500 symbols are employed. It should be noted, in particular, that different random bit sources are employed for the downstream and upstream links.
The parameters used in simulating the SOA and RSOA intensity modulators are representative for InGaAsP semiconductor materials operating at a wavelength of ~1550nm, which are listed in Table 1 . It should be pointed out that, to obtain the SOA/RSOA parameters and examine the validity of the RSOA intensity modulator models, fitting of experimental measurements [15] with numerical results are undertaken prior to performing numerical simulations. During the fitting procedure, all the RSOA parameter values obtained in the experiments are treated as constant, and all the parameter values which are not exactly known are initially taken from the literatures [10,11] and subsequently adjusted within reasonable limits to obtain the best fit with all the experimental results [15]. As the SOA intensity modulator is used in the relatively cost-insensitive OLT only, for all the simulations, optimum operating conditions identified in [10] are adopted, which are a CW optical input power of 20dBm, a bias current of 100mA and a driving current with a peak-to-peak (PTP) value of 80mA.
Table 1. RSOA, SOA, RB, SMF and PIN Parameters
In Table 1, the parameters adopted to simulate RB noise are also shown, which are based on the experiment measurement presented in Section 2.3. The simulation parameters for SMFs [9–11] and PIN detectors [10,11] are also listed in Table 1. The transmission distance is fixed at 40km. It is worth mentioning, once again, that attenuation of optical signals due to Rayleigh scattering is included in the SMF attenuation coefficient.
3. Transmission performance
In numerical simulations, the signal line rate for both downstream and upstream transmission is calculated using the expression given below:
where is the total number of original data-carrying subcarriers in the positive frequency bins, Sk is the signal bit rate corresponding to the k-th subcarrier, nk is the total number of binary bits conveyed by the k-th subcarrier within one symbol period Tb, fs is the ADC/DAC sampling rate, and η is the cyclic prefix parameter [9]. The total channel bit error rate (BER), , is defined ashere is the total number of detected errors and is the total number of transmitted binary bits. Both and are for the k-th subcarrier, whose sub-channel BER, is given by . Based on and, the maximum modulation format adopted on each of the subcarriers within a symbol can be identified through negotiations between the transmitter and the corresponding receiver. It is also worth addressing that, the signal line rate computed using Eq. (7) is considered to be valid only when the condition of BERT ≤ 1.0 × 10−3 is satisfied.It is well known [10,11,13] that, the downstream and upstream transmission performance of SOA and RSOA modulated AMOOFDM signals is strongly dependent upon modulator’s operating conditions including, for example, electrical driving current PTP (defined explicitly in [11]), bias current and optical power injected into the SOA and RSOA, therefore these operating conditions affect significantly SOA/RSOA effective carrier lifetime, SOA/RSOA dynamic gain characteristics, and thus extinction ratio and clipping of the modulated AMOOFDM signals. As already mentioned in Section 2.5, the SOA intensity modulator always operates at optimum operating conditions, in this paper, discussions are made of the influence of RSOA intensity modulator’s operating conditions on the network performance only. Here, the RSOA modulated AMOOFDM signal extinction ratio, is defined as [10,11]
where is the average optical power, () is the i-th (j-th) signal sample, is the sampling duration, K1 (K2) is the number of samples satisfying () within the entire AMOOFDM signal considered, and K1 + K2 is the total number of samples. The statistical definition of the RSOA modulated AMOOFDM signal extinction ratio takes into account the fact that the time-domain signal waveform has an approximated Gaussian probability density function. According to the definition, the resulting signal extinction ratio is a function of electrical driving current PTP, bias current and injected power of the optical carrier.3.1 Verification of the RSOA intensity modulator model
Given the central role of the RSOA intensity modulator and RB noise in determining the network performance, effort is first made to verify, at both the device and system levels, the validity of the theoretical models of the RSOA intensity modulator and RB noise.
Experimental measurements of RSOA intensity modulated real-time end-to-end OOFDM transmission have been undertaken in a unidirectional system with a CW optical carrier wave being supplied locally [15]. In such a system, the effects of RB noise and aforementioned crosstalk are absent. To ensure fair comparisons between numerically simulated results and the experimental measurements, for this comparison only, the OOFDM modem parameters identical to those used in the experiments are adopted, which include 32 subcarriers, a 25% cyclic prefix, a 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 transmission system roll-off effect [12,15]). The above parameters give a net signal line rate of 6Gb/s (a raw signal line rate of 7.5Gb/s). In addition, system parameters used in the numerical simulations are also identical to those employed in the experiments [15]. These system parameters are listed as followings: an electrical OFDM driving current PTP of 42mA, a DC bias current of 84mA, a 5dBm CW optical power injected into the RSOA, 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 shown in Fig. 4(a) , from which excellent agreements are observed across the entire OOFDM signal spectral region. It can also be found from Fig. 4(a) that, the RSOA intensity modulator has a 3-dB modulation bandwidth of approximately 1.25GHz, which is mainly determined by the low optical input power-induced long RSOA effective carrier lifetime [10,11,15]. As seen from Fig. 4(b), the simulated optical OOFDM signal spectrum at the output of the RSOA intensity modulator also agrees very well with the experimental measurements. In comparison with Fig. 4(a), the occurrence of a severer spectral roll-off in Fig. 4(b) results from the following two reasons: a) output filtering and inherent sin(x)/x response in the DAC, b) signal spectral distortion induced by the dynamic RSOA frequency chirp effect [15].
Fig. 4 Comparisons between simulation and experiment cases: (a) RSOA intensity modulator frequency response, and (b) RSOA modulated output signal spectrum.
Making use of variable subcarrier powers similar to those reported in [15], after transmission through a 25km SMF, the representative subcarrier constellations recorded prior to channel equalization in the receiver, are presented in Fig. 5 , where the simulated constellations are shown in Figs. 5(a)-5(c) and the corresponding experimentally measured constellations are shown in Figs. 5(d)-5(f). Once again, the simulated results fit extremely well with the experimental measurements. The constellation rotation increases with increasing subcarrier frequency, originating from the phase shift induced by fibre chromatic dispersion. As a direct result of the spurious points circled in Figs. 5(e) and 5(f), the simulated power penalty at a BER of 1.0 × 10−3 is slightly smaller than that measured in the experiments [15].
Fig. 5 Comparisons between simulation and experiment cases: (a)-(f) constellations for representative subcarriers. (a)-(c) are simulated results and (d)-(f) are experimental results.
Furthermore, comparisons of the transmission performance of RSOA intensity modulated AMOOFDM signals between numerical simulations and experimental measurements using off-line DSP are also made in a more complicated transmission link, into which a CW optical wave is injected from the OLT and propagates in an opposite direction with respect to the modulated AMOOFDM signal over the same SMF [17]. In such a system, the RB noise effect is present. For this comparison only, the parameters adopted in numerical simulations are identical to those used in the experimental measurements [17]. These parameters are: a −1dBm CW optical power coupled into the SMF link in the OLT, an optical input power of −10dBm at the input facet of the RSOA, and a −14dBm 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 is feasible at a BER of 1 × 10−4, which agrees very well with the experimental measurements [17].
3.2 Downstream AMOOFDM transmission performance
Compared to the upstream signal, the downstream AMOOFDM signal suffers from the RB noise effect only. To demonstrate explicitly the impact of such an effect on the transmission performance of the downstream AMOOFDM signal, the downstream AMOOFDM signal line rate versus upstream launch power is plotted in Fig. 6 . In obtaining Fig. 6, the RSOA bias current and driving current PTP are fixed at 100mA and 280mA, respectively, and the downstream launch power is set to 6.3dBm, which gives rise to an optical power of −5.5dBm coupled into the RSOA. The alteration of the upstream launch power is realized by adjusting the VOA at the output of the RSOA intensity modulator, as shown in Fig. 1. It should be noted that the SOA operating conditions are given in Section 2.5.
Fig. 6 Signal line rate and OSRNR of downstream AMOOFDM signals as a function of upstream launch power. The downstream launch power is fixed at 6.3dBm.
It can be seen from Fig. 6 that, 17Gb/s over 40km downstream transmission is feasible for upstream launch powers of <-6dBm. More importantly, the downstream signal line rate decreases almost linearly with increasing upstream launch power expressed in dBm. When the upstream launch power is increased to 6dBm, the downstream signal line rate is reduced to a value as small as 4Gb/s. Such a significant variation in downstream signal line rate is due to the RB noise effect induced by the upstream AMOOFDM signal. From Eqs. (5) and (6), it can be understood that a high upstream launch power corresponds a large RB noise power and high ONU gain, thus resulting in a reduction in downstream OSRNR, as seen in Fig. 6. The above discussions imply that the RB effect plays a dominant role in determining the maximum achievable downstream transmission performance, and that a small upstream launch power is preferred for maximizing the downstream transmission performance.
3.3 Upstream AMOOFDM transmission performance and optimization of RSOA intensity modulator operating conditions
In the previously discussed downstream transmission case, the upstream signal-induced RB noise is the only dominant factor limiting the achievable downstream signal bit rate, whilst the upstream transmission performance is mainly attributed to three factors listed as followings: 1) RB noise induced by both the downstream and upstream AMOOFDM signals, 2) crosstalk due to the residual downstream signal-induced fluctuation of the upstream signal, and 3) dynamic optical characteristics of the RSOA intensity modulator. In addition, these three factors are also strongly dependent upon each other.
To explore the dependence of the upstream AMOOFDM signal transmission performance on RSOA operating conditions, Fig. 7 is plotted to show the corresponding signal line rate versus electrical driving current PTP for different RSOA bias currents. In obtaining Fig. 7, both the downstream and upstream launch powers are fixed at 6.3dBm, which results in an optical power of −5.5dBm injected into the RSOA intensity modulator. As shown in Fig. 7, the driving current PTP affects significantly the achievable upstream AMOOFDM signal bit rate. It is also very interesting to note that, for various bias currents, there exists an almost identical optimum driving current PTP of 280mA, corresponding to which the maximum upstream signal line rate is obtained. Compared to the strong driving current PTP-dependent upstream transmission performance, the dependence of the upstream transmission performance on RSOA bias current is not pronounced considerably, as shown in Fig. 7, and an optimum bias current of 100mA is identified.
To gain an in-depth understanding of the physical mechanisms underpinning the upstream signal line rate evaluation trends shown in Fig. 7, in Fig. 8 performance comparisons are made between two cases of modulating a centrally-supplied downstream CW optical wave (here referred to as CASE I) and re-modulating the downstream AMOOFDM optical signal (referred to as CASE II). For each of the above-mentioned cases, simulated results are also presented for conditions of including and excluding the RB noise effect. In obtaining Fig. 8, the bias current is set to 100mA and all other simulation parameters are identical to those adopted in simulating Fig. 7.
Fig. 8 Upstream signal line rate versus driving current PTP for cases of using a centrally-supplied downstream CW optical wave and re-modulating a downstream AMOOFDM signal. For each of these two cases, numerical results obtained under conditions of including and excluding the RB noise effect are also plotted.
In Fig. 8, the signal line rate difference between two red curves results from the RB noise effect, which, as expected from Eq. (4), is independent of the driving current PTP. In addition, Fig. 8 also indicates that the RB noise effect can halve the upstream signal line rate. On the other hand, for the case of excluding the RB noise effect, the upstream performance difference between the red (CASE I) and blue (CASE II) curves is due to the crosstalk effect, which, as seen in Fig. 8, plays a dominant role in determining the achievable upstream signal line rate for driving current PTPs of <200mA, beyond this value the effect becomes relatively weak.
The above discussions imply that a high driving current PTP is capable of reducing effectively the crosstalk effect. This statement is confirmed in Fig. 9 , where normalized RSOA modulated upstream optical waveform comparisons between CASE I and CASE II are made for different driving current PTP values of 80mA and 280mA. For fair comparison, in calculating Fig. 9, use is also made of identical random bit sequences and the same subcarrier modulation format distribution across all the subcarriers. It can be seen from Fig. 9 (a) that, for the low PTP value of 80mA, there exist very strong unwanted waveform fluctuations for CASE II. However, such optical waveform fluctuations almost disappear when the PTP value is increased to 280mA, as seen in Fig. 9(b). The high driving current PTP-induced reduction in the crosstalk effect can be explained by considering the fact that, a high driving current PTP gives rise to a large extinction ratio of the RSOA modulated AMOOFDM signal [13], as shown in the upper x-axis of Fig. 8. The large upstream signal extinction ratio can drown out the residual downstream signal waveform fluctuations in the RSOA intensity modulator.
Fig. 9 RSOA-modulated optical output waveforms for cases of using a centrally-supplied downstream CW optical wave (CASE I) and re-modulating a downstream AMOOFDM signal (CASE II).
From the above analysis, it is clear that both RB noise and crosstalk are dominant in determining the achievable upstream AMOOFDM transmission performance. Based on such an understanding, the driving current PTP dependent signal line rate behaviors shown in Fig. 7 can be explained easily: for driving current PTPs less than the optimum value of 280mA, the sharp increase in upstream signal line rate with increasing driving current PTP is a direct result of the rapid reduction in the crosstalk effect. While a further increase in driving current PTP introduces the severe signal clipping effect owing to the strong RSOA optical gain saturation [13], giving rise to a small variation in signal extinction ratio (shown in the upper x-axis of Fig. 8), thus a flattened signal line rate developing trend is observed in Fig. 7. In addition, the signal clipping effect also contributes to the occurrence of the optimum bias current of approximately 100mA in Fig. 7 [10,11]: for bias currents below the optimum value, the decrease in signal line rate is because the lower part of the driving current is clipped, as this part penetrates into the negative optical gain region of the RSOA. Whilst for bias currents above the optimum bias current, the transmission performance degradation originates mainly from decreased signal extinction ratio and increased signal clipping, as the upper part of the driving current experiences an almost flat optical gain of the RSOA [10,11].
The above-mentioned numerical simulations are performed based on an assumption that both the downstream and upstream launch powers are fixed at 6.3dBm. However, discussions in Section 2.3 indicate that the RB noise effect experienced by the upstream AMOOFDM signal depends upon both downstream and upstream launch powers. To explore such an interesting issue, contour plots of upstream OSRNR and corresponding signal line rate as a function of downstream and upstream launch powers are plotted in Fig. 10 . In computing Fig. 10, the RSOA intensity modulator is assumed to operate at the optimum operating conditions identified in Fig. 7, and the variations in downstream and upstream launch powers are realized by adjusting the VOAs in the OLT and ONU. As predicted in Eq. (4), Fig. 10(a) shows that the maximum upstream OSRNR is obtainable only when the downstream and upstream launch powers are very similar. As a direct result, the maximum upstream AMOOFDM signal line rate of 8Gb/s occurs when the difference between upstream and downstream launch powers is less than 3dB, as seen in Fig. 10(b).
Fig. 10 Contour plot of upstream OSRNR (a) and upstream signal line rate (b) as a function of upstream and downstream launch powers.
Comparisons between Fig. 6 and Fig. 10 indicate that, there is a trade-off between the achievable upstream and downstream transmission performances. In practical system design, the specific downstream and upstream launch powers can be chosen according to specific requirements. For example, under optimum SOA and RSOA operating conditions as well as the downstream and upstream launch powers of 6.3dBm and 0dBm, respectively, the system performance of 10Gb/s in downstream and 6Gb/s in upstream over a single 40km SMF are feasible.
4. SSB-SCM for improving the downstream and upstream transmission performance
Having investigated the significant impacts of the effects of RB noise and crosstalk on the achievable AMOOFDM transmission performance of the WDM-PON architecture of interest of this paper, in this section, the feasibility of utilizing a novel SSB-SCM technique [14] in the downstream is explored, for the first time, to alleviate effectively these two critical effects to simultaneously improve considerably the downstream and upstream AMOOFDM transmission performance.
In the SSB-SCM technique, use is made of a RF modulator utilizing a Hilbert transform-based phase-shift approach to produce a real-valued SSB downstream electrical signal in the OLT. The generated SSB-SCM signal can be expressed as
where A DSB(t) is the real-valued double sideband (DSB) downstream electrical signal created by the conventional OFDM transmitter [9–11]. H{A DSB (t)} presents the Hilbert transform of A DSB (t). ωRF is the intermediate RF carrier frequency, which is taken to be 15GHz [14]. is utilized to drive directly the SOA intensity modulator in the OLT. The resulting downstream optical signal spectrum at the output of the SOA intensity modulator is shown in Fig. 11(a) . In the corresponding downstream optical receiver, an optical filter is inserted before square-law photon detection to remove the lower sideband of the received optical signal. RF down-conversion is performed to the converted electrical signal, prior to data recovery in the downstream OFDM receiver; whilst the upstream AMOOFDM transmitter has a configuration identical to that illustrated in Fig. 1, where a conventional real-valued DSB upstream electrical signal is used to drive the RSOA intensity modulator with the downstream optical signal being injected as an optical carrier wave. The RSOA-modulated upstream optical signal spectrum is shown in Fig. 11(b), from which it can be found that the upstream and downstream optical signal spectra are separated clearly. In the upstream optical receiver in the OLT, before performing square-law photon detection, an optical filter is employed to remove the residual downstream optical signal spectrum and a part of the RB noise power.From the above analysis, it is clear that the SSB-SCM technique is capable of preventing the downstream optical spectrum from overlapping with the upstream optical spectrum, thus resulting in a considerable reduction in the effects of RB noise and crosstalk, provided that appropriate optical filters are employed. The above-mentioned design also makes economic sense, as the simplicity of both the upstream optical transmitter and downstream optical receiver is still being preserved to satisfy the demands for low-cost transceivers in the ONU. Here, it should be pointed out that, in comparison with the schemes discussed in Section 3, the SSB-SCM scheme requires a wider bandwidth photodiode and a RF down-converter in each ONU. It is envisaged that such requirements may not increase significantly the cost of the ONUs, because of the potentially large volume of the ONUs and the fast development of highly integrated wide bandwidth photon detectors [25].
The effectiveness of the SSB-SCM technique in improving the downstream AMOOFDM transmission capacity is shown in Fig. 12 , where the transmission performance of a conventional DSB downstream AMOOFDM signal is also plotted for comparison. In obtaining Fig. 12, simulation parameters identical to those considered in Fig. 6 are adopted. It is shown in Fig. 12 that, in comparison with the conventional case and for upstream launch powers of <0dBm, the SSB-SCM technique can improve significantly the downstream transmission capacity by a factor of approximately 2, which can be further enhanced for upstream launch powers of >0dBm. The downstream signal capacity decay with increasing upstream launch power is because of a part of the RB noise induced by the residual upstream waveform still remains.
Fig. 12 Downstream signal line rate versus upstream launch power for SSB-SCM and conventional DSB AMOOFDM signals. The downstream launch power is fixed at 6.3dBm.
Compared to the conventional DSB baseband technique, the SSB-SCM technique can also enhance simultaneously the upstream transmission performance, as shown in Fig. 13 . In obtaining Fig. 13, simulation parameters identical to those adopted in Fig. 10 are considered, except that an optimum driving current PTP value of 80mA (rather than 280mA) is used in the RSOA intensity modulator. Such a reduction in optimum driving current PTP arises due to the elimination of the crosstalk effect. In comparison with the conventional case, the SSB-SCM technique can improve the upstream AMOOFDM signal capacity by about 30%. Such improvement is, however, much lower than that corresponding to downstream transmission, this is mainly due to the fact that a reflected part of the RB noise associated with the upstream data signal still exists.
Fig. 13 Upstream signal line rate versus upstream launch power for both SSB-SCM and conventional DSB AMOOFDM signals. The downstream launch power is fixed at 6.3dBm.
5. Conclusions
Detailed numerical investigations have been undertaken of wavelength reused bidirectional transmission of AMOOFDM signals in colorless WDM-PONs incorporating a SOA intensity modulator and a RSOA intensity modulator in the OLT and ONU, respectively. A comprehensive theoretical model describing the performance of such networks has, for the first time, been developed, taking into account the dynamic optical characteristics of the SOA and RSOA intensity modulators as well as the effects of RB noise and crosstalk. The developed model has been rigorously verified experimentally in RSOA-based real-time end-to-end OOFDM transmission systems at 7.5Gb/s. It has been shown that the effects of RB noise and crosstalk due to residual downstream signal-induced waveform fluctuations are dominant factors limiting the achievable downstream and upstream performances. Under optimum SOA and RSOA operating conditions as well as practical downstream and upstream optical launch powers, 10Gb/s downstream and 6Gb/s upstream over 40km SMF transmissions of conventional double sideband AMOOFDM signals are feasible without in-line optical amplification and chromatic dispersion compensation. In particular, the aforementioned transmission performance can be improved to 23Gb/s downstream and 8Gb/s upstream over 40 km SMFs when SSB-SCM is utilized in the downstream systems.
It is worth addressing that, very recently we have demonstrated experimentally real-time end-to-end OOFDM transceivers operating at a record-breaking signal bit rates of 11.25Gb/s. Making use of the demonstrated real-time OOFDM transceivers, experimental investigations are currently being undertaken in our research group to verify the theoretical predictions presented in the present paper, and results will be reported in due course.
Acknowledgements
This work was partly supported by the European Community's Seventh Framework Programme (FP7/2007-2013) within the project ICT ALPHA under grant agreement n° 212352, and in part by The Royal Society Brian Mercer Feasibility Award. The works of J. L. Wei, X. Q. Jin and X. Zheng were also supported by the School of Electronic Engineering and the Bangor University.
References and links
1. K. Grobe and J.-P. Elbers, “PON in adolescence: from TDMA to WDM-PON,” IEEE Commun. Mag. 46(1), 26–34 (2008). [CrossRef]
2. P. Healey, P. Townsend, C. Ford, L. Johnston, P. Townley, I. Lealman, L. Rivers, S. Perrin, and R. Moore, “Spectral slicing WDM-PON using wavelength-seeded reflective SOAs,” Electron. Lett. 37(19), 1181–1182 (2001). [CrossRef]
3. S. L. Woodward, P. P. Iannone, K. C. Reichmann, and N. J. Frigo, “A spectrally sliced PON employing Fabry-Perot lasers,” IEEE Photon. Technol. Lett. 10(9), 1337–1339 (1998). [CrossRef]
4. D. K. Jung, H. Kim, K. H. Han, and Y. C. Chung, “Spectrum-sliced bidirectional passive optical network for simultaneous transmission of WDM and digital broadcast video signals,” Electron. Lett. 37(5), 308–309 (2001). [CrossRef]
5. H. Takesue and T. Sugie, “Wavelength channel data rewrite using saturated SOA modulator for WDM networks with centralized light sources,” J. Lightwave Technol. 21(11), 2546–2556 (2003). [CrossRef]
6. W. Lee, S. H. Cho, M. Y. Park, J. H. Lee, C. Kim, G. Jeong, and B. W. Kim, “Optical transceiver employing an RSOA with feed-forward current injection,” OFC/NFOEC, (Anaheim, CA, USA, 2007), Paper OTuH1.
7. W. Hung, C. K. Chan, L. K. Chen, and F. Tong, “An optical networking unit for WDM access networks with downstream DPSK and upstream re-modulated OOK data using injection-locked FP laser,” OFC/NFOEC, (Anaheim, CA, USA, 2003), Paper TuR2.
8. N. Deng, C.-K. Chan, and L.-K. Chen, “A centralized-light-source WDM access network utilizing inverse-RZ downstream signal with upstream data remodulation,” Opt. Fiber Technol. 13(1), 18–21 (2007). [CrossRef]
9. J. M. Tang and K. A. Shore, “30 Gb/s signal transmission over 40-km directly modulated DFB-laser-based single-mode-fibre links without optical amplification and dispersion compensation,” J. Lightwave Technol. 24(6), 2318–2327 (2006). [CrossRef]
10. J. L. Wei, A. Hamié, R. P. Giddings, and J. M. Tang, “Semiconductor optical amplifier-enabled intensity modulation of adaptively modulated optical OFDM signals in SMF-based IMDD systems,” J. Lightwave Technol. 27(16), 3678–3689 (2009). [CrossRef]
11. J. L. Wei, X. L. Yang, R. P. Giddings, and J. M. Tang, “Colourless adaptively modulated optical OFDM transmitters using SOAs as intensity modulators,” Opt. Express 17(11), 9012–9027 (2009), http://www.opticsinfobase.org/abstract.cfm?uri=oe-17-11-9012. [CrossRef] [PubMed]
12. R. P. Giddings, X. Q. Jin, and J. M. Tang, “First experimental demonstration of 6Gb/s real-time optical OFDM transceivers incorporating channel estimation and variable power loading,” Opt. Express 17(22), 19727–19738 (2009). [CrossRef] [PubMed]
13. J. L. Wei, A. Hamié and J. M. Tang, “Optimization and comparison of the transmission performance of RSOA/SOA intensity-modulated optical OFDM signals for WDM-PONs,” OFC/NFOEC, (San Diego, California, USA, March 21–25, 2010). Paper JThA53.
14. X. Zheng, J. L. Wei, and J. M. Tang, “Transmission performance of adaptively modulated optical OFDM modems using subcarrier modulation over SMF IMDD links for access and metropolitan area networks,” Opt. Express 16(25), 20427–20440 (2008), http://www.opticsinfobase.org/abstract.cfm?uri=oe-16-25-20427. [CrossRef] [PubMed]
15. R. P. Giddings, E. Hugues-Salas, X. Q. Jin, J. L. Wei and J. M. Tang, “ Colourless real-time optical OFDM end-to-end transmission at 7.5Gb/s over 25km SSMF using 1GHz RSOAs for WDM-PONs,” OFC/NFOEC, (San Diego, California, USA, March 21–25, 2010). Paper OMS4.
16. J. Yu, M.-F. Huang, D. Qian, L. Chen, and G.-K. Chang, “Centralized lightwave WDM-PON employing 16-QAM intensity modulated OFDM downstream and OOK modulated upstream signals,” IEEE Photon. Technol. Lett. 20(18), 1545–1547 (2008). [CrossRef]
17. T. Duong, N. Genay, P. Chanclou, B. Charbonnier, A. Pizzinat, and R. Brenot, “Experimental demonstration of 10 Gbit/s for upstream transmission by remote modulation of 1 GHz RSOA using Adaptively Modulated Optical OFDM for WDM-PON single fiber architecture,” in European Conference on Optical Communication (ECOC), (Brussels, Belgium, 2008), PD paper Th.3.F.1.
18. C.-W. Chow, C.-H. Yeh, C.-H. Wang, F.-Y. Shih, C.-L. Pan, and S. Chi, “WDM extended reach passive optical networks using OFDM-QAM,” Opt. Express 16(16), 12096–12101 (2008), http://www.opticsinfobase.org/abstract.cfm?URI=oe-16-16-12096. [CrossRef] [PubMed]
19. C. W. Chow, C. H. Yeh, C. H. Wang, F. Y. Shih, and S. Chi, “Demonstration of signal remodulation long reach carrier distributed passive optical network using OFDM-QAM signal,” in ECOC (Vienna, Austria, 2009), paper 8.5.2.
20. P. Gysel and R. K. Staubli, “Statistical properties of Rayleigh backscattering in single-mode fibers,” J. Lightwave Technol. 8(4), 561–567 (1990). [CrossRef]
21. C. Arellano, K.-D. Langer, and J. Prat, “Reflection and multiple Rayleigh backscattering in WDM single-fiber loopback access networks,” J. Lightwave Technol. 27(1), 12–18 (2009). [CrossRef]
22. J. Ko, S. Kim, J. Lee, S. Won, Y. S. Kim, and J. Jeong, “Estimation of performance degradation of bidirectional WDM transmission systems due to Rayleigh backscattering and ASE noises using numerical and analytical models,” J. Lightwave Technol. 21(4), 938–946 (2003). [CrossRef]
23. G. P. Agrawal, Fibre-Optic Communication Systems, (Wiley, 1997).
24. J. M. Tang and K. A. Shore, “Maximizing the transmission performance of adaptively modulated optical OFDM signals in multimode-fiber links by optimizing analog-to-digital converters,” J. Lightwave Technol. 25(3), 787–798 (2007). [CrossRef]
25. X. Tian, A. P. Freundorfer, and L. Roy, “Noise analysis of a photoreceiver using a P-I-N and GaAs HBT distributed amplifier combination,” IEEE Microw. Wirel. Compon. Lett. 13(6), 208–210 (2003). [CrossRef]
