Performance Enhancement and Capacity Enlargement for a DWDM-PON System Utilizing an Optimized Cross Seeding Rayleigh Backscattering Design

: In this work, a record of 16 channels, with future channel spacing in the telecommunication standardization sector of the International Telecommunications Union G.694.1 (ITU-T G.694.1) for Dense Wavelength Division Multiplexing (DWDM) (i.e., 12.5 GHz), is simulated and tested. This work is done to realize a proposed high capacity DWDM-Passive Optical Network (DWDM-PON) system. These speciﬁcations are associated with enhancing the upstream (US) capacity to 2.5 Gb / s over a 25 km Single-Mode Fiber (SMF) transmission and producing a noteworthy average Bit Error Rate (BER) of 10 − 12 during the system’s evaluation process. These performance indicators are achieved through design optimization of the cross-seeding Rayleigh Backscattering (RB) elimination technique. This optimization has successfully reduced (compared to the cross-seeding related literature) the simulated DWDM-PON components and maintained an e ﬀ ective Rayleigh Backscattering elimination with the aforementioned system’s performance enhancement and capacity enlargement.


Introduction
Today, Time Division Multiplexing-Passive Optical Network (TDM-PON) standard is used. TDM-PON splits the optical power over a limited number of users based on a synchronized time slot in which all users share the available bandwidth. Due to the increasing demand for advanced multimedia applications, TDM-PON becomes insufficient enough for the high data rate that is required [1]. Wavelength Division Multiplexing-Passive Optical Network (WDM-PON) is the future of the next-generation network, which achieves the full usage of bandwidth by assigning for each user his own wavelength that creates a virtual connection between the Central Office (CO) and the user terminal [2].
To meet the rapid growth in current bandwidth requirements, DWDM-Passive Optical Networks (DWDM-PONs) have become the optimum choice. This type of network is considered to be the future of ultra-capacity networks that require an increasing number of channels by decreasing the channel spacing [3]. DWDM-PON also uses a remarkable low-cost Optical Network Unit (ONU) colorless operation [3]. Finally, the efficient utilization of the wavelength is another important requirement [4].
To meet low-cost ONU colorless operation requirements, several techniques, devices, and architectures are proposed, tested, and evaluated. A famous technique to realize ONU colorless operation is the use of reflective transmitters. These transmitters can be obtained by using

The System's Design, Architecture, and Operating Principles
The basic system structure, operating description, and design enhancements for this work are presented in this section. The proposed high capacity DWDM-PON system, based on modified cross-seeding architecture, is shown in Figure 1. The Optical Line Terminal (OLT) can be divided into four stages, as illustrated in Figure 1. (A) Stage 1 represents the input light source, which consists of eight CW light sources that have a wavelength spacing of 25 GHz and combine together to be forced into Stage 2. (B) At Stage 2, the optical light sources are utilized to generate subcarriers by using the Double Sideband Optical Carrier Suppression (DSB-OCS) technique. In this work, the DSB-OCS technique is used to increase the system capacity to a record of 8 light sources with 16 subcarriers (i.e., channels) by decreasing the channel spacing to 12.5 GHz. Generally, DSB-OCS has high receiver sensitivity with a small power penalty [24,25]. At this stage, an Erbium-Doped Fiber Amplifier (EDFA) is used. Each sideband (i.e., subcarrier) acts as a Centralized Light Source (CLS) to be re-modulated for US subcarriers, as shown in Figure 1. (C) The subcarriers are demultiplexed as odd and even subcarriers to ready them for transmission through the two-bidirectional fibers, feed fiber 1 (FF1) and feed fiber 2 (FF2). This part of the operation takes place in Stage 3. In this work, this function is performed with much lower complexity compared to the related studies that utilized the cross-seeding technique [22,26,27], but without affecting the performance, as performance will be described later in Section 4. (D) Then, each odd and even subcarrier is modulated by on-off key (OOK) modulation at Stage 4. OOK is the simplest modulation format and relatively inexpensive. Thus, OOK is used to meet the ONU requirements, as mentioned before. The transmitter (Tx) in Figure 2 consists of a Continuous Wave (CW) laser source with a wavelength of (193.025 THz) and an output power of 0 dBm. It is externally modulated at 10 Gb/s with a Non-Return-to-Zero (NRZ) pseudorandom binary sequence in a Mach-Zehnder Modulator (MZM) with a 30 dB extinction ratio. The NRZ format has been widely implemented, mainly due to its signal bandwidth and its relatively easy generation [28]. In addition, stage 4 contains a receiver (Rx) for the US signal, as shown in Figure 2. A p-i-n photodiode/ photodetector (PIN) is used with Low-Pass Filter (LPF) to demodulate the US signal. The PIN specifications are 10 nA dark current and 1 A/W responsivity. The same Tx and Rx will be used again in Stage 7. The odd DS signals are multiplexed to be transmitted through FF1. Similarly, even DS signals are multiplexed to be transmitted through FF2. The remaining part of this work can be classified as follows. The basic cross-seeding system design, architecture, and operating principles are demonstrated in Section 2. The detailed system architecture and parameters are presented in Section 3. In Section 4, tracing, testing, and an evaluation of the system are carried out. Section 5 includes a comparison between this work and the related literatures. The conclusion is introduced in Section 6.

The System's Design, Architecture, and Operating Principles
The basic system structure, operating description, and design enhancements for this work are presented in this section. The proposed high capacity DWDM-PON system, based on modified crossseeding architecture, is shown in Figure 1. The Optical Line Terminal (OLT) can be divided into four stages, as illustrated in Figure 1. (A) Stage 1 represents the input light source, which consists of eight CW light sources that have a wavelength spacing of 25 GHz and combine together to be forced into Stage 2. (B) At Stage 2, the optical light sources are utilized to generate subcarriers by using the Double Sideband Optical Carrier Suppression (DSB-OCS) technique. In this work, the DSB-OCS technique is used to increase the system capacity to a record of 8 light sources with 16 subcarriers (i.e., channels) by decreasing the channel spacing to 12.5 GHz. Generally, DSB-OCS has high receiver sensitivity with a small power penalty [24,25]. At this stage, an Erbium-Doped Fiber Amplifier (EDFA) is used. Each sideband (i.e., subcarrier) acts as a Centralized Light Source (CLS) to be remodulated for US subcarriers, as shown in Figure 1. (C) The subcarriers are demultiplexed as odd and even subcarriers to ready them for transmission through the two-bidirectional fibers, feed fiber 1 (FF1) and feed fiber 2 (FF2). This part of the operation takes place in Stage 3. In this work, this function is performed with much lower complexity compared to the related studies that utilized the cross-seeding technique [22,26,27], but without affecting the performance, as performance will be described later in Section 4. (D) Then, each odd and even subcarrier is modulated by on-off key (OOK) modulation at Stage 4. OOK is the simplest modulation format and relatively inexpensive. Thus, OOK is used to meet the ONU requirements, as mentioned before. The transmitter (Tx) in Figure 2 consists of a Continuous Wave (CW) laser source with a wavelength of (193.025 THz) and an output power of 0 dBm. It is externally modulated at 10 Gb/s with a Non-Return-to-Zero (NRZ) pseudorandom binary sequence in a Mach-Zehnder Modulator (MZM) with a 30 dB extinction ratio. The NRZ format has been widely implemented, mainly due to its signal bandwidth and its relatively easy generation [28]. In addition, stage 4 contains a receiver (Rx) for the US signal, as shown in Figure  2. A p-i-n photodiode/ photodetector (PIN) is used with Low-Pass Filter (LPF) to demodulate the US signal. The PIN specifications are 10 nA dark current and 1 A/W responsivity. The same Tx and Rx will be used again in Stage 7. The odd DS signals are multiplexed to be transmitted through FF1. Similarly, even DS signals are multiplexed to be transmitted through FF2.  Stage 5 represents the heart of the cross-seeding technique that eliminates RB. This is done by transmitting the odd DS signals after re-modulation (i.e., Stages 6, 7) as the US through FF2, as indicated in Figure 1. The same procedure will be applied to even DS signals but by passing them through FF1 after re-modulation. Recall that transmission via conventional full-duplex bidirectional techniques suffers from a huge level of RB [28,29], which forces researchers to find solutions, as indicated in Section 1.
Finally, in Stage 6, the RN consists of two multiplexers (MUX) and two demultiplexers (DeMUX). MUX and DeMUX are simpler and lower in cost than Array Waveguide Grating (AWG), which is a complex component consisting of a coupler and fiber grating [30]. Txs and Rxs are used for the demodulation/re-modulation in Stage 7 after the received signal is split by a 20:80 coupler. Only 20% of the signal power is used to extract the DS information by using the same receiver structure mentioned in Stage 4. The remaining 80% of the signal power is used for the re-modulation of the US by OOK, and its structure is mentioned in Stage 4. Using the Saving Array Waveguide Grating (AWG), an internal feed fiber, and RSOA is a merit of this work's design compared to related studies that utilized cross-seeding techniques [22,26,27].

Extended System Architecture and Parameters
The proposed cross-seeding based DWDM-PON system is presented in details in Figure 2. The general stages, which are presented in Figure 1, are re-introduced with details in Figure 2.
It is noteworthy to mention that the main system's stage specifications and their corresponding values are mentioned in the body of Section 2 and Table 1. In the following, the remaining commercial components' specifications will be addressed. The first launched wavelength is located at 193.025 THz with 0 dB launched power using a CW light source. By utilizing other seven CW light sources Stage 5 represents the heart of the cross-seeding technique that eliminates RB. This is done by transmitting the odd DS signals after re-modulation (i.e., Stages 6, 7) as the US through FF2, as indicated in Figure 1. The same procedure will be applied to even DS signals but by passing them through FF1 after re-modulation. Recall that transmission via conventional full-duplex bidirectional techniques suffers from a huge level of RB [28,29], which forces researchers to find solutions, as indicated in Section 1.
Finally, in Stage 6, the RN consists of two multiplexers (MUX) and two demultiplexers (DeMUX). MUX and DeMUX are simpler and lower in cost than Array Waveguide Grating (AWG), which is a complex component consisting of a coupler and fiber grating [30]. Txs and Rxs are used for the demodulation/re-modulation in Stage 7 after the received signal is split by a 20:80 coupler. Only 20% of the signal power is used to extract the DS information by using the same receiver structure mentioned in Stage 4. The remaining 80% of the signal power is used for the re-modulation of the US by OOK, and its structure is mentioned in Stage 4. Using the Saving Array Waveguide Grating (AWG), an internal feed fiber, and RSOA is a merit of this work's design compared to related studies that utilized cross-seeding techniques [22,26,27].

Extended System Architecture and Parameters
The proposed cross-seeding based DWDM-PON system is presented in details in Figure 2. The general stages, which are presented in Figure 1, are re-introduced with details in Figure 2.
It is noteworthy to mention that the main system's stage specifications and their corresponding values are mentioned in the body of Section 2 and Table 1. In the following, the remaining commercial Appl. Sci. 2019, 9, 4520 5 of 12 components' specifications will be addressed. The first launched wavelength is located at 193.025 THz with 0 dB launched power using a CW light source. By utilizing other seven CW light sources and the DSB-OCS technique, the system capacity is enlarged to 16 subcarriers (i.e., channels) by decreasing the channel spacing to 12.5 GHz. A 6.25 GHz sine generator that feeds a standard MZM with a 30 dB extinction ratio and an EDFA that has a 20 dB gain and a 4 dB noise figure are used in stage 2. In stage 3, a DeMUX with 10 GHz 3dB bandwidth and 0 dB insertion loss is utilized. Here, channels that reached the Tx in stage 4 achieve DS modulation via an externally modulated 10 Gb/s Non-Return-to-Zero (NRZ) pseudorandom binary sequence using the previous MZM specifications. As said before, the Rx in stage 4 will be used for demodulating the returned US signal with PIN and an LPF. The PIN specifications are 10 nA dark current and 1 A/W responsivity. A 4th order LPF with 0 dB insertion loss, 100 dB depth, and a cut off frequency of 1.875 GHz is used. As said before, the same Tx and Rx will be used again in Stage 7, which will provide simplicity and harmony to the proposed design. The MUX/DeMUX used in stage 4 and 6 for the US and DS channel selections have 10 GHz 3 dB bandwidth, 0 dB insertion loss, and 100 dB depth. The specifications of the bidirectional fibers used in Stage 5 are listed in Table 1.
Points (a-g), indicated in Figure 2, are used to trace, test, and evaluate the proposed design, as will be provided in Section 4.

Results and Discussion
To verify this work's goals, system tracing and evaluation are presented in this section. Figure 3a-g represents the optical spectra for signals processed through the proposed cross-seeding system, as indicated in Figure 2.
Appl. Sci. 2019, 9, x FOR PEER REVIEW 5 of 12 and the DSB-OCS technique, the system capacity is enlarged to 16 subcarriers (i.e., channels) by decreasing the channel spacing to 12.5 GHz. A 6.25 GHz sine generator that feeds a standard MZM with a 30 dB extinction ratio and an EDFA that has a 20 dB gain and a 4 dB noise figure are used in stage 2. In stage 3, a DeMUX with 10 GHz 3dB bandwidth and 0 dB insertion loss is utilized. Here, channels that reached the Tx in stage 4 achieve DS modulation via an externally modulated 10 Gb/s Non-Return-to-Zero (NRZ) pseudorandom binary sequence using the previous MZM specifications. As said before, the Rx in stage 4 will be used for demodulating the returned US signal with PIN and an LPF. The PIN specifications are 10 nA dark current and 1 A/W responsivity. A 4th order LPF with 0 dB insertion loss, 100 dB depth, and a cut off frequency of 1.875 GHz is used. As said before, the same Tx and Rx will be used again in Stage 7, which will provide simplicity and harmony to the proposed design. The MUX/DeMUX used in stage 4 and 6 for the US and DS channel selections have 10 GHz 3 dB bandwidth, 0 dB insertion loss, and 100 dB depth. The specifications of the bidirectional fibers used in Stage 5 are listed in Table 1.
Points (a-g), indicated in Figure 2, are used to trace, test, and evaluate the proposed design, as will be provided in Section 4.

Results and Discussion
To verify this work's goals, system tracing and evaluation are presented in this section. Figure  3a-g represents the optical spectra for signals processed through the proposed cross-seeding system, as indicated in Figure 2.    Table 1.
After being demultiplexed, the even and odd DS channels are modulated through IM with the NRZ format, each with 10 Gb/s. Finally, these channels are applied to DS MUX to provide the crossseeding odd and even spectra ready to be transmitted through FF1 and FF2, respectively. These spectra are presented in Figure 3b,c and are considered to correspond to the system signals' trace at points (b) and (c) in Figure 2. Figure 3d represents the spectrum of a chosen odd DS channel spectrum (i.e., λ1-1 = 193.019 THz as an example for the odd channels) after transmitting 25 km through FF1 and after being demultiplexed at point (d) in Figure 2 before an ONU receiver. Figure 3f represents this channel after transmitting back through the retransmitting path (US path) with a record of 2.5 Gb/s (i.e., point (e) in Figure 2). One needs to remember (as in Figure 2) that the retransmitting path (US path) includes a CLS module, MUX, transmitting through FF2 (not in FF1, which is used in the DS path and is the core of the cross-seeding technique that eliminates RB), and, finally, DeMUX to reach a point (e) Figure 2.   Table 1.
After being demultiplexed, the even and odd DS channels are modulated through IM with the NRZ format, each with 10 Gb/s. Finally, these channels are applied to DS MUX to provide the cross-seeding odd and even spectra ready to be transmitted through FF1 and FF2, respectively. These spectra are presented in Figure 3b,c and are considered to correspond to the system signals' trace at points (b) and (c) in Figure 2. Figure 3d represents the spectrum of a chosen odd DS channel spectrum (i.e., λ 1-1 = 193.019 THz as an example for the odd channels) after transmitting 25 km through FF1 and after being demultiplexed at point (d) in Figure 2 before an ONU receiver. Figure 3f represents this channel after transmitting back through the retransmitting path (US path) with a record of 2.5 Gb/s (i.e., point (e) in Figure 2). One needs to remember (as in Figure 2) that the retransmitting path (US path) includes a CLS module, MUX, transmitting through FF2 (not in FF1, which is used in the DS path and is the core of the cross-seeding technique that eliminates RB), and, finally, DeMUX to reach a point (e) Figure 2.
A similar analysis is done by choosing the even DS channel λ 1-2 = 193.031 THz as an example for the even channels. Figure 3e,g shows this channel spectrum after transmitting the DS path through FF2 and then returning using US path through FF1. These spectra correspond to the system trace at points (f) and (g) in Figure 2.
Finally, Figure 4 represents the eye diagrams for the selected odd and even channels after the retransmitting path (i.e., points (e) and (g) in Figure 2). Figure 4a,b represents the eye diagram evaluation for the previously chosen odd and even channels that are presented in Figure 3f,g (i.e., λ 1-1 and λ 1-2 ) while Figure 4c,d represents the eye diagram representation for the extra chosen channels (i.e., λ 8-1 and λ 8-2 ).
Appl. Sci. 2019, 9, x FOR PEER REVIEW 7 of 12 A similar analysis is done by choosing the even DS channel λ1-2 = 193.031 THz as an example for the even channels. Figure 3e,g shows this channel spectrum after transmitting the DS path through FF2 and then returning using US path through FF1. These spectra correspond to the system trace at points (f) and (g) in Figure 2.
Finally, Figure 4 represents the eye diagrams for the selected odd and even channels after the retransmitting path (i.e., points (e) and (g) in Figure 2). Figure 4a,b represents the eye diagram evaluation for the previously chosen odd and even channels that are presented in Figure 3f,g (i.e., λ1-1 and λ1-2) while Figure 4c,d represents the eye diagram representation for the extra chosen channels (i.e., λ8-1 and λ8-2). The results indicate that the worst simulated BER in the chosen samples can be observed for channel λ1-1 of 10 −12 , which corresponds to a quality factor (Q) of 6.88, while a remarkable simulated BER of 10 −13, which corresponds to a quality factor (Q) of 7.03, is noticed for channel λ1-2. An average BER better than 10 −12 can be observed for the remaining channels.
To provide a deeper understanding of Figure 4, it is useful to review some of the eye diagram basics and characteristics. The red curve in Figure 4 represents the ideal received bit pattern only, while the blue pattern represents a normalized pattern for the 2 bits actually received, with the 1 bit "eye-opening" in the center of the display and 1/2 bit to the left and right of the center eye to capture the rise/fall-time transitions [31][32][33]. The results indicate that the worst simulated BER in the chosen samples can be observed for channel λ 1-1 of 10 −12 , which corresponds to a quality factor (Q) of 6.88, while a remarkable simulated BER of 10 −13, which corresponds to a quality factor (Q) of 7.03, is noticed for channel λ 1-2 . An average BER better than 10 −12 can be observed for the remaining channels.
To provide a deeper understanding of Figure 4, it is useful to review some of the eye diagram basics and characteristics. The red curve in Figure 4 represents the ideal received bit pattern only, while the blue pattern represents a normalized pattern for the 2 bits actually received, with the 1 bit "eye-opening" in the center of the display and 1/2 bit to the left and right of the center eye to capture the rise/fall-time transitions [31][32][33].
The eye diagram is a useful tool for the qualitative analysis of the signals used in digital transmission. This diagram provides an at-a-glance evaluation of the system performance and can offer insight into the nature of the channel's imperfections. Careful analysis of this visual display can give the user a first-order approximation of the signal-to-noise ratio, clock timing jitter, and skew [31][32][33].
Several key values can be extracted directly from Figure 4, such as the One Level, Zero Level, Eye Amplitude, eye height, crossing point, rise time, and fall time. Then, system performance indicators, such as the Eye Crossing Percentage, signal to noise ratio, eye width, Duty Cycle Distortion, BER, and quality factor, can be directly calculated using simple and direct relations based on the key values [31]. Finally, it is worth mentioning that today's manufacturers provide devices that directly display all previous values and system indicators once the system is tested [31,32].

Comparison with Related Studies
This section provides a detailed specification comparison between this work and related literature. It is worth mentioning that the key differences between this work's technique (cross-seeding) and other techniques related to the critical requirements for the DWDM-PON system are already discussed in Section 1. In addition, the reasons, which have motivated us to use the cross-seeding technique, are also discussed there. Table 2 represents a detailed specification and value comparison between this work and related studies. Again, as mentioned in Section 1, the comparison includes literature not only for DWDM-PON systems and WDM-PON, but also for Hybrid WDM/TDM-PON and other related systems. For example, in [5], the number of channels, US, and DS rates are remarkable, but the problems associated with the use of (FP-L) (mentioned in Section 1) limit the transmission distance, the ability to use more channels (i.e., channel spacing), and the BER. Besides the FP-L influence, this work utilizes a higher Dispersion Compensation Fiber (DCF) length of −340 ps/nm to provide the aforementioned performance characteristics in Table 2 for 15 km. The effect of, and price of, the high length DCF can be found in many dispersion compensation-related studies, such as [35,36]. Finally, an in-band cross talk, which cannot be avoided, is observed through the system's testing [5].
Another nominated example can be reviewed in [21]. The effect of using high gain ROSA limits the US/DS bitrates, the ability to use more channels (i.e., channel spacing), and the BER. To enhance the merged TDM capacity, a high data rate optical source may be used [37]. Finally, the BER enhances to 10 −9 if the number of channels is reduced to only one (as in this work) [21].
A Long-Reach Bidirectional Centralized WDM-PON system, with attractive operating characteristics, is achieved in [19]. The use of centralized light sources is a desirable solution to reduce system monitoring and maintenance cost [19,38]. However, the utilization of OCS-SCM with its advantages and disadvantages (as in Section 1) limits the number of channels due to the large channel spacing required. A review of this system's architecture leads to a moderate-to-high level of complexity. This complexity is due to the existence of two EDFAs-AWG and long DCF [19].
Dual-arm modulator configurations are proposed in [20] to perform wavelength shifting for the ONUs to mitigate the RB impairment in bidirectional centralized light PONs. Although the proposed configurations reduce the dependence of the system on devices like EDFAs, AWG, and long DCF, the use of a combination of amplitude and phase modulation dramatically limits the US and DS rates [20]. On the other hand, this work, with its low operating rate, can provide perfect integration between a WDM-PON system (as a step toward 5G systems) and other low bit rate commercial 5G systems, such as the single carrier modulation (SCM) visible light communication (VLC) systems, as well as their applications [39,40]. The limited number of channels and the non-availability of such a modulator at that time are additional limitations [20].
The highest observed US and DS rates are accomplished with a record of 40 Gb/s in [10]. This work utilizes an integrated semiconductor optical amplifier with a reflective electro-absorption modulator (SOA-REAM). The idea of integrating everything into one compact device is currently pioneering and seems promising. Unfortunately, severe inter-symbol interference and strong beat noise are major limitations [10]. A material like a linear/non-linear photonic crystal (Phc) could mitigate these limitations. With its remarkable integration ability, low power consumption, and remarkable performance in variety of applications [41][42][43], the functionality of SOA-REAM can be achieved with attractive operational characteristics. In [10], several techniques are proposed to overcome these limitations, but the effects of these limitations dominate, since the optimum allowable BER = 2 × 10 −4 , with 20 Km and one channel.

Conclusions
This work simulates an enhanced cross-seeding-based DWDM-PON system with future channel spacing in the ITU-T G.694.1 standard for DWDM (i.e., 12.5 GHz). Attractive performance characteristics have been targeted through this work. These characteristics include enhancing the downstream (DS) channel capacity, the US bit rate, minimizing the active components, and maintaining the remarkable BER performance. Utilizing this work's design enhancement, the proposed DWDM-PON has successfully achieved a record of 16 channels, a US capacity of 2.5 Gb/s over 25 km of transmission, and a noteworthy average BER of 10 −12 . A comparison between this work and related studies, which are not only based on DWDM-PON systems but also cover other types of high-speed optical networks, was performed to ensure the validity and effectiveness of the proposed system.

Conflicts of Interest:
The authors declare no conflict of interest.