This document summarizes a study of the effects of various parameters on the performance of dispersion-managed dense wavelength-division multiplexing (DWDM) transmission systems for ultra-high speed communication. It was found that:
1) Power penalty increases with the number of channels due to four-wave mixing but saturates after a critical number of channels. Power penalty also increases with smaller channel spacing and higher bit rates.
2) Bit error rate decreases with larger channel spacing and lower bit rates due to reduced nonlinear interference. Fibers with residual dispersion and shorter channel lengths have better error performance.
3) The error performance of a multichannel system degrades at high input powers due to four-wave mixing noise
This document summarizes a study of the effects of various parameters on the performance of dispersion-managed dense wavelength-division multiplexing (DWDM) transmission systems for ultra-high speed communication. It was found that:
1) Power penalty increases with the number of channels due to four-wave mixing but saturates after a critical number of channels. Power penalty also increases with smaller channel spacing and higher bit rates.
2) Bit error rate decreases with larger channel spacing and lower bit rates due to reduced nonlinear interference. Fibers with residual dispersion and shorter channel lengths have better error performance.
3) The error performance of a multichannel system degrades at high input powers due to four-wave mixing noise
This document summarizes a study of the effects of various parameters on the performance of dispersion-managed dense wavelength-division multiplexing (DWDM) transmission systems for ultra-high speed communication. It was found that:
1) Power penalty increases with the number of channels due to four-wave mixing but saturates after a critical number of channels. Power penalty also increases with smaller channel spacing and higher bit rates.
2) Bit error rate decreases with larger channel spacing and lower bit rates due to reduced nonlinear interference. Fibers with residual dispersion and shorter channel lengths have better error performance.
3) The error performance of a multichannel system degrades at high input powers due to four-wave mixing noise
Iqbal Hasan #1 , Mashad Uddin Saleh #2 , Nabeela Kauser #3 , Dr. Mohammad Faisal *4 Department of Electrical and Electronic Engineering, Bangladesh University of Engineering and Technology. 1 tonmoysky@yahoo.com 2 mashad2810@yahoo.com 3 smileyjeny@gmail.com 3 mdfaisal@eee.buet.ac.bd
Abstract In this paper, we numerically study the characteristics of dispersion managed (DM) fiber. We show the effect of nonlinearities in the dense wavelength-division-multiplexing (DWDM) transmission performances. We also show the effect of varying the frequency spacing and channel lengths for various numbers of channels on the power penalty in DWDM transmission system. Finally we show these effects for different bitrates e.g. 40-Gb/s, 80-Gb/s, 120-Gb/s and 160-Gb/s. Here we take the design of a dispersion compensating fiber and residual dispersion into account.
Keywords Dense wavelength division multiplexing, dispersion, residual dispersion, frequency spacing, power penalty, optical fiber.
I. INTRODUCTION Now-a-days a remarkable increase has been seen in data traffic and the need for an ultrahigh-capacity transmission system has no other alternative. To meet this demand, a lot of experiments and reports have been observed. There have been researches on the effects on increasing the no of channels in the DWDM transmission systems. However, there have been few reports on the effects of residual dispersion, different frequency spacing, bit error rate (BER) on the performance of DWDM system as ultrahigh speed communication. In this paper, we figure out numerically the relationship between dispersion managed optical fiber and DWDM transmission performances. We show these effects in SSMF (Standard Single Mode Fiber), DSF (Dispersion Shifted Fiber) and NZDSF (Nonzero Dispersion Shifted Fiber). We show the power penalties for different channels, frequency spacing and bitrates. We describe the BER of these DWDM transmission systems for variation of different parameters. Moreover we take into consideration the effect and characteristics of DCF (Dispersion Compensated Fiber) and thereby clarify the feasibility of these different transmission lines.
II. MODELLING OF DWDM TRANSMISSION SYSTEM Fig 1 demonstrates a model for our DWDM transmission system. The signals were multiplexed with an equal channel spacing of 25GHz from frequency 193.55 THz.
Fig. 1 DWDM transmission system model In our model, we considered 60Km channel length of SSMF with dispersion parameter D c as 17ps/nm- km compensated by DCF. The attenuation coefficient , nonlinear coefficient , responsivity R, Dispersion slope d dD c were set as 0.24 dB/km, 5.8410 -6 m -2 W -2 , 0.85 A/W, 0.055 ps/nm 2 -km respectively. III. EFFECT OF NO OF CHANNELS OVER DWDM SYSTEM As our first observation, we inspected the effect of increasing the no of channels over the DWDM system. We found that the power penalty increased as the no of channels increased. But after a critical no of channels, the power penalty almost saturates. In fig 2, the whole procedure was repeated for different input powers of 1mW, 1.5mW and 2mW respectively and similar results were found. However, it is observed that the power penalty is minimized for the lower input power.
Fig. 2: Calculated relationship between power penalty and no of channels In the fig 2 above, the increase of power penalty with the increase in channels is due to the FWM (Four wave mixing) effect in the fiber. Because of the FWM process a large number of new sideband signals are generated. The total number of FWM products because of the interaction of N number of channels is, Total no. of FWM Products = 2 3 2 1 N N
Each of the newly generated sideband frequency takes power from of the main channel frequency spectrum, thus the available power at the receiver get reduced. So power penalty increases as the total no. of channels increase. And this increase in power penalty due to FWM effect is proportional to the cube of transmitting power. IV. EFFECT OF CHANNEL SPACING IN DWDM SYSTEM Due to the increase in demand we need to move from the earlier WDM system to DWDM system. In DWDM system channel spacing is 50GHz or less than that whereas WDM system channel spacing was 100GHz or above. This reduction in frequency spacing has some obvious effect over the DWDM system. As we can see from the fig 3, power penalty increases as we tends to smaller channel spacing.
Fig 3: Effect of channel spacing in DWDM system
V. ERROR PERFORMANCE OF DWDM SYSTEM We observed the effect of channel spacing and bitrate on BER for different received power in DWDM system. In fig 4, we can see that the greater the channel spacing, the lesser the input power is needed for expected BER of 10 -9 . Smaller channel spacing leads to greater power penalty. So, for BER of 10 -9 , greater input power is needed for smaller channel spacing like 25GHz,50GHz while relatively low power is needed for channel spacing of 100GHz or greater.
Fig. 4 Relationship between BER and received power for different channel spacing System error performance at higher bitrates degrades. From our model it is found that BER of 10 -9 is achievable for 40Gbps/channel and 160Gbps/channel system at received power of -22dBm and -18dBm. So the higher the bitrate is, the higher received power per channel, i.e. higher input power is necessary. However the best BER of 10 -8 is possible for ultra-high speed of 320Gbps/channel system. We consider a 100km DCF fiber for our model in fig 5 below.
Fig. 5 Relationship between BER and received power for different bitrates Fiber with residual dispersion performes better than DCF fibers. For same input power, fibers with residual dispersion has lower BER than DCF fiber. From fig 6, it is obvious that fibers with negative residual dispersion performes better than other fibers.
FWM generation efficiency decreases with the increase of fiber chromatic dispersion, it is clear that dispersive fibers are better than low dispersion fibers, since they reduce FWM effect and thereby enhance the system performance. But, on the other hand, chromatic dispersion produces pulse broadening, i.e., ISI which ultimately badly impairs the performance. So, we must apply some compensation mechanism to keep overall dispersion as minimum as possible.
Fig. 6 Relationship between BER and received power for different residual dispersion
Fig 7 shows the performance of SSMF fiber over DCF fiber.And due to high dispersion of SSMF error performance of it is better than DCF fiber.
Fig. 7 Relationship between BER and received power for different fibers
Fiber length also plays an important role on the sysytem performance. As the length increases, error performance degrades. Its obvious that for greater length the power penalty increases, as the relation between power penalty due to FWM effect and channel effective length L eff is, power penalty L eff
2
From fig 8, its evident that shorter channel length has better perfomance. Error rate increases as the channel length increases.
Fig. 8 Relationship between BER and received power for different channel lengths
Multichannel systems face the severe effect of FWM while single channel systems face shot noise, thermal noise and other non-linearity like SBS, SRS only. At low signal power, the FWM noise is small compared to the shot noise and thermal noise and the performance is limited by these noises. As the signal power increases, the BER decreases. As signal power increases further, the FWM noise becomes dominant and the BER starts to increase. That is, the system is impaired by FWM interference and the performance is degraded for large power levels. As single channel system does not face any FWM effect, its error performance improves as input power increases. While for multichannel system error performance improves as the input power increases to a certain threshold value. For our case, it is -17dBm which can be observed from fig 9. After this threshold value FWM effect becomes very dominant and error performance degrades as input power increases.
Fig. 9 Relationship between BER and received power for single channel and multichannel VI. ALLOWABLE INPUT POWERS FOR VARIOUS FIBER As we discussed earlier, fibers with high dispersion parameter has a better control over FWM effect. So NZDSF and SSMF fibers have a lower power penalty profile than DSF and DCF fibers. From fig 10 below, we observe that SSMF and NZDSF fibers have very low power penalty than DSF fiber for same input power.
Fig. 10 Relationship between power penalty and input power for different types of fiber
VII. CONCLUSION Error performances in DWDM system for high-speed communication have been observed and analyses have been made in our work. It has been found that very high speed e.g. 160Gbps, 320Gbps can be used for DWDM communication system. It may require some extra considerations for error reduction, but it will improve the data transmission efficiency by a huge extent. By consulting our work, a decent idea for taking any decision regarding DWDM system can be taken. This work may eventually help explain other phenomena observed during DWDM system performances.