- Email: [email protected]
Impact of tunable laser wavelength drift in a base-band and sub-carrier multiplexed system
Optics Communications 281 (2008) 4057–4060
Contents lists available at ScienceDirect
Optics Communications journal homepage: www.elsevier.com/locate/optcom
Impact of tunable laser wavelength drift in a base-band and sub-carrier multiplexed system E. Connolly, A. Kaszubowska-Anandarajah, P. Perry *, L.P. Barry Research Institute for Networks and Communication Engineering, School of Electronic Engineering, Dublin City University, Collins Avenue, Glasnevin, Dublin, Ireland
a r t i c l e
i n f o
Article history: Received 5 November 2007 Accepted 2 April 2008
Keywords: Tuneable lasers Ultra dense wavelength division multiplexing SGDBR Sub-carrier multiplexing
a b s t r a c t The potential use of very densely spaced wavelengths in FTTx systems to carry modest bit rate broadband connections can be implemented either with conventional base-band (BB) intensity modulation or using sub-carrier multiplexing (SCM) using Radio carriers. Such systems will typically use a long time frame time-sharing system to share a transmitting laser between a number of users. The impact of the adjacent channel interference due to wavelength drift of a tunable laser (TL) in such a system has been characterised for both the BB and SCM approaches. In the experiments described, a laser operating on a fixed wavelength represents the desired channel and an interferer is produced by using a TL that switches periodically between two other channels, one of which is adjacent to the desired channel. Although the TL output is blanked during the main switching transient, some wavelength drift occurs after the end of the blanking period which can cause interference to the adjacent channel. The BER measurements on the desired channel show that SCM is more resistant to this interference, allowing for closer channel spacing. For the TL tested, the BB data shows an error floor >1e 4 while the SCM data gave error free performance with a power penalty of 1.2 dB at 1e 9 in comparison to the back-to-back case. Ó 2008 Elsevier B.V. All rights reserved.
1. Introduction As the demand for broadband connectivity continues to increase, the maximum available capacity and cost effectiveness of FTTx networks have attracted much interest from researchers. In such schemes a single optical fibre from a local exchange feeds a splitter with an individual fibre link to each customer. Dynamic bandwidth allocation, using one or more TLs and passive wavelength routers, allows traffic to be routed through a passive optical network (PON) to different optical network units (ONUs) by simply switching the TL wavelength [1,2]. In such systems, the ONU cost can be minimised by operating at modest bit rates to take advantage of technology that is readily available for 100 Mbps Ethernet as shown in Fig. 1. In the residential market the traffic profile of each subscriber is typically very bursty and so the TL can be shared between a large numbers of clients without any appreciable reduction in their perceived quality of service. That is, these systems can use a flexible time division multiple access (TDMA) system with a very high contention ratio to improve overall system cost. The use of such systems for access networks has been motivated in the past [3] and more recently discussed in terms of Gigabit Ethernet (GbE) with multiple lasers [4]. The Korean experiences * Corresponding author. Tel.: +353 1 7005884; fax: +353 1 7005508. E-mail addresses: [email protected] (E. Connolly), philip@perryradio. com (P. Perry). 0030-4018/$ - see front matter Ó 2008 Elsevier B.V. All rights reserved. doi:10.1016/j.optcom.2008.04.006
discussed in [5] show the real deployment of such systems and, of particular interest here, discussed the use of 125 Mbps WDM-SCMPON systems for residential access. Clearly such a system can use multiple TL’s connected to the same PON to improve capacity but since the switching events of one TL will generally not be synchronised with switching events on another TL, there arises the potential for a switching event on one TL to interfere with a static channel that is adjacent to its target channel. As TL’s are not used on the uplink channel, the work presented here will only consider the downlink data transmission. To investigate such a system, we used a fixed wavelength laser as the static channel and introduced adjacent channel interference by periodically switching a TL between the adjacent channel and another channel on the DWDM grid. The TL output is suppressed during the main switching transient to prevent spurious emissions interfering with other channels, however, the TL usually still exhibits some wavelength drift after this blanking period. The data signal was not gated during these tests, so that the BER measured on the static channel was dominated by the periodic errors caused by the drift of the switching channel and these errors were aggregated over the full switching period. In this paper we present results for a scenario where the combination of bit rate and burst duration yields a burst size of 20 Bytes (the size of an Internet Protocol (IP) packet with zero payload) compared to our work presented in [6], which used packet sizes of 62 kBytes (equivalent to 40 typical IP packets). Here we investigate
4058
E. Connolly et al. / Optics Communications 281 (2008) 4057–4060
Fig. 1. Access PON with multiple tunable lasers and high contention ratio. Fig. 3. SCM transmitter used for simulation.
Fig. 4. SCM receiver used for simulation.
Fig. 2. (a) Base-band scheme and (b) sub-carrier multiplexed scheme.
the use of sub-carrier multiplexing (SCM) and relatively simple heterodyne detection [7] to provide a four fold reduction in channel spacing to 12.5 GHz compared to the standard 50 GHz currently available. We examine the impact of the wavelength drift of a TL directly after wavelength switching operations for ONUs employing both traditional base-band (BB) amplitude shift keying and for the proposed SCM system. In our implementation, the SCM signal is demodulated using a mixer based analogue circuit using the same local oscillator (LO) for transmitter and receiver to emulate performance from a phase locked LO at the receiver. Both systems used the same optical filter with bandwidth of 10 GHz as shown in Fig. 2; this determines the detection bandwidth of the BB case.
ceiver simply consisted of external modulation and direct detection. For both the SCM and BB transmitters, the interfering channel was positioned 6 GHz offset from the monitored channel to simulate the worst case TL drift. Both data channels are generated by external modulation with 155 Mbit/s of either NRZ data (for BB case) or BPSK data on a 500 MHz carrier (for SCM case). The interfering channel’s signal strength was adjusted by use of a variable optical attenuator to emulate the effect of the OBPF on the amplitude of the interference. For the results shown here, the interferer was attenuated by 5 dB. The simulations showed a 3 dB reduction in sensitivity for the SCM case compared to the BB case which is due the optical power present in the unmodulated residual carrier of the SCM signal. The results shown in Fig. 5 are for an interference level of 5 dBc and show that the SCM system gives good performance when the RF carriers of adjacent channels are in quadrature and poor performance when they are in phase or in anti-phase. The BER curves for the BB case have been shifted by 3 dB to account for the 3 dB penalty between the back-to-back BB and SCM case (due to the residual carrier). This ensures that the back-to-back curve shown
2. Simulations Since the dynamic drift of the TL (shown by a dashed line) is sufficiently large to make the signal fall inside the optical filter, the BB detected data (Fig. 2a) will experience interference due to the intensity demodulation used. In the SCM case (Fig. 2b), however, in-band interference does not impact the detected data signal as the demodulation process is phase coherent in the electrical domain so it will select the desired channel that is at the centre of the OBPF. To verify that the adjacent channel interference due to TL drift could be reduced in the SCM case, simulations were carried out using the VPI optical network simulator to compare the degradation of performance due to the presence of interference. The diagrams in Figs. 3 and 4 show the simulation systems for the SCM transmitter and receiver respectively. The BB transmitter and re-
Fig. 5. Simulation results for SCM and base-band systems.
4059
E. Connolly et al. / Optics Communications 281 (2008) 4057–4060
3. Experimental setup and results To simulate an AWGR based dynamically bandwidth allocated network, we modulated two lasers: a TL at 1544.924 nm (channel 52) and a static laser at 1544.824 nm (12.5 GHz away), representing an interfering channel and the desired channel respectively. Both lasers were externally modulated with 155 Mb/s data using 2.5 GHz electro-optic modulators biased at quadrature. The interfering channel was de-correlated using 40 m of fibre, and the power levels were matched using an optical attenuator. One port of a 12.5 GHz spaced wavelength routing node was simulated by filtering the desired (static) channel using a fibre Bragg grating (FBG) with a bandwidth of 10.8 GHz. The demultiplexed channel was then passed through a variable optical attenuator before entering the direct detection receiver which consists of an EDFA, an optical filter, a photo diode and an electrical amplifier. After detection the BB signal was filtered using a LPF and fed directly to an error detector. To represent degraded optical signal to noise ratios which may be present in optically amplified PON’s [8] the signal was attenuated prior to an optical amplifier and spontaneous–spontaneous beat noise minimised using a relatively broad 2 nm filter prior to direct detection and electrical signal processing. As a reference, base-band data was transmitted using a 117 MHz electrical filter in the transmitter, and processed in the receiver with an identical low pass filter (Fig. 6a). For the SCM system, at the transmitter the 155 Mbit/s data was BPSK modulated onto a 500 MHz local oscillator before double side band electro-optic modulation, whilst the received signal was demodulated to base-band using a second mixer before low pass filtering (Fig. 6b). The specific TL used here had an integrated semiconductor optical amplifier (SOA) at the output which was used to blank or attenuate any spurious components generated during the wavelength tuning process (approximately 60 ns) [9]. After this time the wavelength is specified to be within 15 GHz of the destination wavelength and a wavelength locker circuit performs the fine tuning. The wavelength drift during this locking time has been character-
Fig. 6. (a) Base-band experimental setup and (b) sub-carrier multiplexed experimental setup.
Freq [GHz].
in Fig. 5 is for both the SCM and BB case. The BB performance shown is comparable to the SCM case with adjacent channel carriers being offset by 45°. For the network topology shown in Fig. 1 this means that the central station must select the phase of the modulating RF carrier to achieve quadrature between adjacent channels when BPSK modulation is used. Intuitively this corresponds to setting the constellation points of the interfering channel to be mid way between the constellation points for the desired channel. By inference, a system using QPSK would require an offset between adjacent channels of 45° to minimise the adjacent channel interference due to TL drift. The maximum group delay variation at the edge of the OBPF was 20 ps, which translates to a phase shift of less than 4° at the RF frequency used which introduces a small uncertainty into the experiment. In the experimental setup in this work, then the phase of the RF carriers in adjacent channels was tuned to minimise the impact of the interference. For narrow spacing, the residual drift of a tunable laser becomes significant when compared to the channel spacing. For a DD system, this leads to large error floors associated with the tuning transient when the interfering laser is passed by the filter. However, for the SCM system, the effective filter bandwidth is restricted to the data rate, and so more drift may be tolerated. In the tested scenario with short data bursts that periodically use the channel adjacent to the static channel, it is expected that SCM will outperform the BB system. The goal of this paper, then, is to quantitatively compare SCM and BB systems with this aggressive switching regime.
8 6 4 2 0 -2 -4 0
100
200
300
400
500
600
700
Time [ns] Fig. 7. TL wavelength drift after channel 42–52 transition.
ised for a large number of channel transitions [10], and Fig. 7 presents the wavelength drift of the TL (as a function of time from the beginning of the switch) for the transition between channels 42 (1548.915 nm) and 52. The TL wavelength locker can be seen to activate 30 ns after the TL comes out of blanking causing a rapid fluctuation in output wavelength for a small period of time (15 ns), after which the wavelength drift is characterised by a damped oscillation. From the plot it can be seen that the maximum amplitude of the damped oscillation for this transition is around 8 GHz, which is well within the 15 GHz specified for the TL. As the optical filter used to select the static channel has a bandwidth of 10 GHz, the signal will be within the filter pass band for several tens of nanoseconds, but will not fall within the modulation bandwidth of the BPSK modulation in the SCM case. In order to characterise the adjacent channel interference due to the wavelength drift of the TL we measured the bit error rate (BER) of the desired channel (static laser) for three different modes of TL operation: (1) when TL was static at channel 42 (512.5 GHz away from the fixed laser), (2) when TL was static at channel 52 (12.5 GHz away from the fixed laser) and (3) when TL was switched repetitively between channel 42 and 52 every 1 ls. The BER vs. received optical power plot for the BB data is shown in Fig. 8a. It can be seen that there is a negligible power penalty on the desired laser channel when the TL is in static mode at channel 52 (12.5 GHz away) in comparison to when it is at channel 42 (512.5 GHz away). This demonstrates that the optical filter used can select out a channel from a 12.5 GHz spaced DWDM system with no interference from an adjacent channel. It can also be seen that, when the TL is switched repetitively as described above,
4060
E. Connolly et al. / Optics Communications 281 (2008) 4057–4060
a 1.E+00
b 1.E+00 42-52
52
42
1.E-04 1.E-06 1.E-08
42-52
1.E-02 BER
BER
1.E-02
52
42
1.E-04 1.E-06 1.E-08
1.E-10 -35
-33
-31 -29 Received Power [dBm]
-27
1.E-10 -35
-33
-31 -29 Received Power [dBm]
-27
Fig. 8. BER vs. received optical power for (a) BB data and (b) SCM data for different modes of operation of the TL.
wavelength drift of the TL causes an adjacent channel interference, which places an error floor on the performance characteristic of the signal above 1e 4. The same BER measurements were performed for the SCM data and are presented in Fig. 8b. In this case we achieved error free performance in all three cases, with a residual power penalty of 1.2 dB at a reference BER of 10 9 replacing the error floor when the interfering channel is switched repetitively. 4. Discussion and conclusion The cross-channel interference due to the wavelength drift of a TL has been investigated in BB and SCM systems with closely spaced data channels in a PON in the access network. Simulation results show that this interference can be reduced in the SCM system by setting the RF carriers in adjacent WDM channels to be in quadrature when BPSK is used. Our experimental results show that the impact of the wavelength drift from the TL after switching is much more severe for the BB data (BER > 1e 4) than for the SCM signal (Power Penalty of 1.2 dB at 1e 9) even though the separation between the SCM signals was smaller than for the BB data. This difference in performance is due to the fact that in the case of the BB DWDM system, any portion of the light from the TL that leaks through the OBPF interferes with the data carried by the static laser. However, in the case of SCM with heterodyne detection at the receiver the side bands of the two lasers have to spectrally overlap, as a result of TL drift, before significant degradation of the quality of the signal transmitted by the fixed laser is observed. This indicates that using SCM in a low bit rate DWDM access PON
with TLs enables the wavelengths to be more closely spaced without a significant increase in ONU complexity. Acknowledgements This material is based upon work supported by the Science Foundation Ireland under Grants 03/IN.1/1340 and 03/IN.3/I427. The TL used is an AltoNet 1200 FTL Tx Module from Intune Networks Ltd., Ireland. The authors would also like to thank A.K. Mishra, A.D. Ellis and D. Cotter of the Tyndall National Institute, Cork, Ireland for the assistance with preparing this paper. References [1] A. Kaszubowska-Anandarajah, Fast Tunable Lasers in Radio-Over-Fibre Access Networks, Paper We. D1.2, ICTON 2006, Nottingham, UK. [2] C. Bock, IEEE J. Light Technol. 23 (12) (2005) 3981. [3] C.R. Giles et al., IEEE PTL 8 (1I) (1996). [4] F-T. An et al., IEEE J. Light Technol. 22 (11) (2004). [5] C-H. Lee et al., OSA J. Opt. Network. 6 (5) (2007). [6] E. Connolly, Adjacent Channel Interference due to Wavelength Drift of a TL in Base-Band and Subcarrier Multiplexed System, Paper ThEE3, LEOS 2006, Montreal, Canada. [7] C. Bock, Wavelength Independent RSOA-based ONU for FTTH PON Implementation of Switched Ethernet Services, Paper Mo. 4.3.3, ECOC2005, Glasgow, UK. [8] C.W. Chow, Reduction of Rayleigh noise in 10 Gb/s DWDM-PONs by Wavelength Detuning and Phase Modulation Induced Spectral Broadening, Paper We 4.5.5, ECOC 2006, Cannes, France. [9] A.J. Ward, IEEE J. Sel. Topics Quant. Electron. 11 (1) (2005) 149. [10] E. Connolly, Cross Channel Interference due to Wavelength Drift of Tunable Lasers in DWDM Networks, Paper Mo. P.13, ICTON 2006, Nottingham, UK.
Contents lists available at ScienceDirect
Optics Communications journal homepage: www.elsevier.com/locate/optcom
Impact of tunable laser wavelength drift in a base-band and sub-carrier multiplexed system E. Connolly, A. Kaszubowska-Anandarajah, P. Perry *, L.P. Barry Research Institute for Networks and Communication Engineering, School of Electronic Engineering, Dublin City University, Collins Avenue, Glasnevin, Dublin, Ireland
a r t i c l e
i n f o
Article history: Received 5 November 2007 Accepted 2 April 2008
Keywords: Tuneable lasers Ultra dense wavelength division multiplexing SGDBR Sub-carrier multiplexing
a b s t r a c t The potential use of very densely spaced wavelengths in FTTx systems to carry modest bit rate broadband connections can be implemented either with conventional base-band (BB) intensity modulation or using sub-carrier multiplexing (SCM) using Radio carriers. Such systems will typically use a long time frame time-sharing system to share a transmitting laser between a number of users. The impact of the adjacent channel interference due to wavelength drift of a tunable laser (TL) in such a system has been characterised for both the BB and SCM approaches. In the experiments described, a laser operating on a fixed wavelength represents the desired channel and an interferer is produced by using a TL that switches periodically between two other channels, one of which is adjacent to the desired channel. Although the TL output is blanked during the main switching transient, some wavelength drift occurs after the end of the blanking period which can cause interference to the adjacent channel. The BER measurements on the desired channel show that SCM is more resistant to this interference, allowing for closer channel spacing. For the TL tested, the BB data shows an error floor >1e 4 while the SCM data gave error free performance with a power penalty of 1.2 dB at 1e 9 in comparison to the back-to-back case. Ó 2008 Elsevier B.V. All rights reserved.
1. Introduction As the demand for broadband connectivity continues to increase, the maximum available capacity and cost effectiveness of FTTx networks have attracted much interest from researchers. In such schemes a single optical fibre from a local exchange feeds a splitter with an individual fibre link to each customer. Dynamic bandwidth allocation, using one or more TLs and passive wavelength routers, allows traffic to be routed through a passive optical network (PON) to different optical network units (ONUs) by simply switching the TL wavelength [1,2]. In such systems, the ONU cost can be minimised by operating at modest bit rates to take advantage of technology that is readily available for 100 Mbps Ethernet as shown in Fig. 1. In the residential market the traffic profile of each subscriber is typically very bursty and so the TL can be shared between a large numbers of clients without any appreciable reduction in their perceived quality of service. That is, these systems can use a flexible time division multiple access (TDMA) system with a very high contention ratio to improve overall system cost. The use of such systems for access networks has been motivated in the past [3] and more recently discussed in terms of Gigabit Ethernet (GbE) with multiple lasers [4]. The Korean experiences * Corresponding author. Tel.: +353 1 7005884; fax: +353 1 7005508. E-mail addresses: [email protected] (E. Connolly), philip@perryradio. com (P. Perry). 0030-4018/$ - see front matter Ó 2008 Elsevier B.V. All rights reserved. doi:10.1016/j.optcom.2008.04.006
discussed in [5] show the real deployment of such systems and, of particular interest here, discussed the use of 125 Mbps WDM-SCMPON systems for residential access. Clearly such a system can use multiple TL’s connected to the same PON to improve capacity but since the switching events of one TL will generally not be synchronised with switching events on another TL, there arises the potential for a switching event on one TL to interfere with a static channel that is adjacent to its target channel. As TL’s are not used on the uplink channel, the work presented here will only consider the downlink data transmission. To investigate such a system, we used a fixed wavelength laser as the static channel and introduced adjacent channel interference by periodically switching a TL between the adjacent channel and another channel on the DWDM grid. The TL output is suppressed during the main switching transient to prevent spurious emissions interfering with other channels, however, the TL usually still exhibits some wavelength drift after this blanking period. The data signal was not gated during these tests, so that the BER measured on the static channel was dominated by the periodic errors caused by the drift of the switching channel and these errors were aggregated over the full switching period. In this paper we present results for a scenario where the combination of bit rate and burst duration yields a burst size of 20 Bytes (the size of an Internet Protocol (IP) packet with zero payload) compared to our work presented in [6], which used packet sizes of 62 kBytes (equivalent to 40 typical IP packets). Here we investigate
4058
E. Connolly et al. / Optics Communications 281 (2008) 4057–4060
Fig. 1. Access PON with multiple tunable lasers and high contention ratio. Fig. 3. SCM transmitter used for simulation.
Fig. 4. SCM receiver used for simulation.
Fig. 2. (a) Base-band scheme and (b) sub-carrier multiplexed scheme.
the use of sub-carrier multiplexing (SCM) and relatively simple heterodyne detection [7] to provide a four fold reduction in channel spacing to 12.5 GHz compared to the standard 50 GHz currently available. We examine the impact of the wavelength drift of a TL directly after wavelength switching operations for ONUs employing both traditional base-band (BB) amplitude shift keying and for the proposed SCM system. In our implementation, the SCM signal is demodulated using a mixer based analogue circuit using the same local oscillator (LO) for transmitter and receiver to emulate performance from a phase locked LO at the receiver. Both systems used the same optical filter with bandwidth of 10 GHz as shown in Fig. 2; this determines the detection bandwidth of the BB case.
ceiver simply consisted of external modulation and direct detection. For both the SCM and BB transmitters, the interfering channel was positioned 6 GHz offset from the monitored channel to simulate the worst case TL drift. Both data channels are generated by external modulation with 155 Mbit/s of either NRZ data (for BB case) or BPSK data on a 500 MHz carrier (for SCM case). The interfering channel’s signal strength was adjusted by use of a variable optical attenuator to emulate the effect of the OBPF on the amplitude of the interference. For the results shown here, the interferer was attenuated by 5 dB. The simulations showed a 3 dB reduction in sensitivity for the SCM case compared to the BB case which is due the optical power present in the unmodulated residual carrier of the SCM signal. The results shown in Fig. 5 are for an interference level of 5 dBc and show that the SCM system gives good performance when the RF carriers of adjacent channels are in quadrature and poor performance when they are in phase or in anti-phase. The BER curves for the BB case have been shifted by 3 dB to account for the 3 dB penalty between the back-to-back BB and SCM case (due to the residual carrier). This ensures that the back-to-back curve shown
2. Simulations Since the dynamic drift of the TL (shown by a dashed line) is sufficiently large to make the signal fall inside the optical filter, the BB detected data (Fig. 2a) will experience interference due to the intensity demodulation used. In the SCM case (Fig. 2b), however, in-band interference does not impact the detected data signal as the demodulation process is phase coherent in the electrical domain so it will select the desired channel that is at the centre of the OBPF. To verify that the adjacent channel interference due to TL drift could be reduced in the SCM case, simulations were carried out using the VPI optical network simulator to compare the degradation of performance due to the presence of interference. The diagrams in Figs. 3 and 4 show the simulation systems for the SCM transmitter and receiver respectively. The BB transmitter and re-
Fig. 5. Simulation results for SCM and base-band systems.
4059
E. Connolly et al. / Optics Communications 281 (2008) 4057–4060
3. Experimental setup and results To simulate an AWGR based dynamically bandwidth allocated network, we modulated two lasers: a TL at 1544.924 nm (channel 52) and a static laser at 1544.824 nm (12.5 GHz away), representing an interfering channel and the desired channel respectively. Both lasers were externally modulated with 155 Mb/s data using 2.5 GHz electro-optic modulators biased at quadrature. The interfering channel was de-correlated using 40 m of fibre, and the power levels were matched using an optical attenuator. One port of a 12.5 GHz spaced wavelength routing node was simulated by filtering the desired (static) channel using a fibre Bragg grating (FBG) with a bandwidth of 10.8 GHz. The demultiplexed channel was then passed through a variable optical attenuator before entering the direct detection receiver which consists of an EDFA, an optical filter, a photo diode and an electrical amplifier. After detection the BB signal was filtered using a LPF and fed directly to an error detector. To represent degraded optical signal to noise ratios which may be present in optically amplified PON’s [8] the signal was attenuated prior to an optical amplifier and spontaneous–spontaneous beat noise minimised using a relatively broad 2 nm filter prior to direct detection and electrical signal processing. As a reference, base-band data was transmitted using a 117 MHz electrical filter in the transmitter, and processed in the receiver with an identical low pass filter (Fig. 6a). For the SCM system, at the transmitter the 155 Mbit/s data was BPSK modulated onto a 500 MHz local oscillator before double side band electro-optic modulation, whilst the received signal was demodulated to base-band using a second mixer before low pass filtering (Fig. 6b). The specific TL used here had an integrated semiconductor optical amplifier (SOA) at the output which was used to blank or attenuate any spurious components generated during the wavelength tuning process (approximately 60 ns) [9]. After this time the wavelength is specified to be within 15 GHz of the destination wavelength and a wavelength locker circuit performs the fine tuning. The wavelength drift during this locking time has been character-
Fig. 6. (a) Base-band experimental setup and (b) sub-carrier multiplexed experimental setup.
Freq [GHz].
in Fig. 5 is for both the SCM and BB case. The BB performance shown is comparable to the SCM case with adjacent channel carriers being offset by 45°. For the network topology shown in Fig. 1 this means that the central station must select the phase of the modulating RF carrier to achieve quadrature between adjacent channels when BPSK modulation is used. Intuitively this corresponds to setting the constellation points of the interfering channel to be mid way between the constellation points for the desired channel. By inference, a system using QPSK would require an offset between adjacent channels of 45° to minimise the adjacent channel interference due to TL drift. The maximum group delay variation at the edge of the OBPF was 20 ps, which translates to a phase shift of less than 4° at the RF frequency used which introduces a small uncertainty into the experiment. In the experimental setup in this work, then the phase of the RF carriers in adjacent channels was tuned to minimise the impact of the interference. For narrow spacing, the residual drift of a tunable laser becomes significant when compared to the channel spacing. For a DD system, this leads to large error floors associated with the tuning transient when the interfering laser is passed by the filter. However, for the SCM system, the effective filter bandwidth is restricted to the data rate, and so more drift may be tolerated. In the tested scenario with short data bursts that periodically use the channel adjacent to the static channel, it is expected that SCM will outperform the BB system. The goal of this paper, then, is to quantitatively compare SCM and BB systems with this aggressive switching regime.
8 6 4 2 0 -2 -4 0
100
200
300
400
500
600
700
Time [ns] Fig. 7. TL wavelength drift after channel 42–52 transition.
ised for a large number of channel transitions [10], and Fig. 7 presents the wavelength drift of the TL (as a function of time from the beginning of the switch) for the transition between channels 42 (1548.915 nm) and 52. The TL wavelength locker can be seen to activate 30 ns after the TL comes out of blanking causing a rapid fluctuation in output wavelength for a small period of time (15 ns), after which the wavelength drift is characterised by a damped oscillation. From the plot it can be seen that the maximum amplitude of the damped oscillation for this transition is around 8 GHz, which is well within the 15 GHz specified for the TL. As the optical filter used to select the static channel has a bandwidth of 10 GHz, the signal will be within the filter pass band for several tens of nanoseconds, but will not fall within the modulation bandwidth of the BPSK modulation in the SCM case. In order to characterise the adjacent channel interference due to the wavelength drift of the TL we measured the bit error rate (BER) of the desired channel (static laser) for three different modes of TL operation: (1) when TL was static at channel 42 (512.5 GHz away from the fixed laser), (2) when TL was static at channel 52 (12.5 GHz away from the fixed laser) and (3) when TL was switched repetitively between channel 42 and 52 every 1 ls. The BER vs. received optical power plot for the BB data is shown in Fig. 8a. It can be seen that there is a negligible power penalty on the desired laser channel when the TL is in static mode at channel 52 (12.5 GHz away) in comparison to when it is at channel 42 (512.5 GHz away). This demonstrates that the optical filter used can select out a channel from a 12.5 GHz spaced DWDM system with no interference from an adjacent channel. It can also be seen that, when the TL is switched repetitively as described above,
4060
E. Connolly et al. / Optics Communications 281 (2008) 4057–4060
a 1.E+00
b 1.E+00 42-52
52
42
1.E-04 1.E-06 1.E-08
42-52
1.E-02 BER
BER
1.E-02
52
42
1.E-04 1.E-06 1.E-08
1.E-10 -35
-33
-31 -29 Received Power [dBm]
-27
1.E-10 -35
-33
-31 -29 Received Power [dBm]
-27
Fig. 8. BER vs. received optical power for (a) BB data and (b) SCM data for different modes of operation of the TL.
wavelength drift of the TL causes an adjacent channel interference, which places an error floor on the performance characteristic of the signal above 1e 4. The same BER measurements were performed for the SCM data and are presented in Fig. 8b. In this case we achieved error free performance in all three cases, with a residual power penalty of 1.2 dB at a reference BER of 10 9 replacing the error floor when the interfering channel is switched repetitively. 4. Discussion and conclusion The cross-channel interference due to the wavelength drift of a TL has been investigated in BB and SCM systems with closely spaced data channels in a PON in the access network. Simulation results show that this interference can be reduced in the SCM system by setting the RF carriers in adjacent WDM channels to be in quadrature when BPSK is used. Our experimental results show that the impact of the wavelength drift from the TL after switching is much more severe for the BB data (BER > 1e 4) than for the SCM signal (Power Penalty of 1.2 dB at 1e 9) even though the separation between the SCM signals was smaller than for the BB data. This difference in performance is due to the fact that in the case of the BB DWDM system, any portion of the light from the TL that leaks through the OBPF interferes with the data carried by the static laser. However, in the case of SCM with heterodyne detection at the receiver the side bands of the two lasers have to spectrally overlap, as a result of TL drift, before significant degradation of the quality of the signal transmitted by the fixed laser is observed. This indicates that using SCM in a low bit rate DWDM access PON
with TLs enables the wavelengths to be more closely spaced without a significant increase in ONU complexity. Acknowledgements This material is based upon work supported by the Science Foundation Ireland under Grants 03/IN.1/1340 and 03/IN.3/I427. The TL used is an AltoNet 1200 FTL Tx Module from Intune Networks Ltd., Ireland. The authors would also like to thank A.K. Mishra, A.D. Ellis and D. Cotter of the Tyndall National Institute, Cork, Ireland for the assistance with preparing this paper. References [1] A. Kaszubowska-Anandarajah, Fast Tunable Lasers in Radio-Over-Fibre Access Networks, Paper We. D1.2, ICTON 2006, Nottingham, UK. [2] C. Bock, IEEE J. Light Technol. 23 (12) (2005) 3981. [3] C.R. Giles et al., IEEE PTL 8 (1I) (1996). [4] F-T. An et al., IEEE J. Light Technol. 22 (11) (2004). [5] C-H. Lee et al., OSA J. Opt. Network. 6 (5) (2007). [6] E. Connolly, Adjacent Channel Interference due to Wavelength Drift of a TL in Base-Band and Subcarrier Multiplexed System, Paper ThEE3, LEOS 2006, Montreal, Canada. [7] C. Bock, Wavelength Independent RSOA-based ONU for FTTH PON Implementation of Switched Ethernet Services, Paper Mo. 4.3.3, ECOC2005, Glasgow, UK. [8] C.W. Chow, Reduction of Rayleigh noise in 10 Gb/s DWDM-PONs by Wavelength Detuning and Phase Modulation Induced Spectral Broadening, Paper We 4.5.5, ECOC 2006, Cannes, France. [9] A.J. Ward, IEEE J. Sel. Topics Quant. Electron. 11 (1) (2005) 149. [10] E. Connolly, Cross Channel Interference due to Wavelength Drift of Tunable Lasers in DWDM Networks, Paper Mo. P.13, ICTON 2006, Nottingham, UK.