coherent ofdm

[GURINDER SINGH]

what is the minimum BER we can get in coherent ofdm with wdm system for 4 channels????

[Abhishek Shrama]

Hello gurinder!
BER roughly always range from 10^-9 to 10^-12.As we consider communication to be error free after this.
check these discussions for more details of BER in OFDM:

OPTISYSTEM OFDM Example

http://www.dsplog.com/2012/01/01/symbol-error-rate-16qam-64qam-256qam/

BER cannot measured for 16-QAM-OFDM direct detection

co-ofdm

OptiSystem – Hot Topics

http://vixra.org/pdf/1208.0091v1.pdf
https://opus.lib.uts.edu.au/research/bitstream/handle/2100/67/27_FloresCenturion.pdf?sequence=1
https://optiwave.com/resources/latest-news/new-version-optisystem-12-2/

BER cannot measured for 16-QAM-OFDM direct detection

16 QAM/ 64 QAM for long distance

[alistu]

Hi Gurinder,

I think the bit-error-rate we get for any optical system depends on many factors and by changing these factors acquiring any amount for BER is possible. Among these factors are bit rate, channel length, and dedicated bandwidth to every WDM channel (in the case of the system you are talking about). So you have to design your system for the BER you wish to get.

Regards

[Ashu verma]

Hello Gurinder singh
I agree with the explanation given by Abhishek and Alisthu about the WDM OFDM.I just want to add that due to the long sequence length you will get the BER 0 i.e error free in optisystem samples (if you are using eye diagram analyzer) and falls to very high values of BER after that.If you are using the DUT signal analyzer i.e BER TEST SET then you will get the no. of error bite more clearly and you will get the noticeable results.Rest in general for the system BER is acceptable within 10^-9 to 10^-12.Visit the link given below as Ravil and Alessandro have given very good points in the discussion

BER / Eye Diagram

[GURINDER SINGH]

The various parameters i have taken are
sequence length-16384 bits
4 channels of frequencies 193.05,193.1,193.15 and 193.2
bit rate of each channel– 10 gbps
channel of 100Km having SMF and DCF
laser linewidth- 0.15 MHz
BER i am getting is– 0.000716 for one channel and 0.0001953 for the other
i am using sample of optisystem version 11
please guide me…

[Ashu verma]

Hello Gurinder

All the parameters you have taken are correct .Also one of examples from the optisystem samples having WDM-OFDm also worked well.For 10Gbps data rate you can enhance your system for more transmission distance by using EDFA amplifier.Can you attach your system file?You can increase laser power to some extent to get good results and do not forget to check constellation of the system.Generally the common problem in OFDM sample is to set the delay in OFDM demodulator.You have to putt the delay value correct and you will get BER 0 i.e error free for 100Km distance.It would be better to if you attach file and we check your system file to correct any parameters in file.
Thank you

[GURINDER SINGH]

file is attached..

[Ashu verma]

Thank you Gurinder singh for attaching file.I will work on it and get back to you soon .

[Alessandro Festa]

Hi Gurinder, have you tried to use a booster amplifier to launch more power into the spans? You are now launching -26dBm/channel which is very low!

[GURINDER SINGH]

i am using amplifier of gain 5db and noise figure 6 db

[GURINDER SINGH]

if anyone can suggest some good papers on this topic

[Alessandro Festa]

I would try to launch at least 0dBm/channels in the first span!

[GURINDER SINGH]

what do you mean by 0dbm/channel?? is it the power of CW laser at transmitter??

[Alessandro Festa]

No, I would use a booster edfa after the WDM (muxing the 4 channels) with 6dBm output power, which means 0dBm/channel.

[Ashu verma]

Hello Gurinder singh
I take a quick look on your system file and found it correct.I have just changed the value of cw laser on all the channels from -4 dB to – 2 dB and it starts working fine.You can change it and will get error free transmission.Good luck

[GURINDER SINGH]

ok thanks sam sung……what ber is coming now??

[Ashu verma]

Hello Gurvinder singh
By doing the modifications i have said above,you will get ber 0 i.e error free communication over the distance you have taken.you can enhance the total transmission distance by launching more power in to the fiber.But higher values of power rise the non linearities .You can use the NF noise figure 4dB instead 6dB in the optical amplifier.EDFa with less that 6dB NF are available now.

[Ashu verma]

Hello, there are some links that i have given below for the ofdm related work and concepts
http://ieeexplore.ieee.org/xpl/login.jsp?tp=&arnumber=7023476&url=http%3A%2F%2Fieeexplore.ieee.org%2Fiel7%2F4563700%2F7023462%2F07023476.pdf%3Farnumber%3D7023476
http://ieeexplore.ieee.org/xpl/login.jsp?tp=&arnumber=6826492&url=http%3A%2F%2Fieeexplore.ieee.org%2Fiel7%2F68%2F4581641%2F06826492.pdf%3Farnumber%3D6826492
http://ieeexplore.ieee.org/xpl/articleDetails.jsp?arnumber=7035548
http://ieeexplore.ieee.org/xpl/login.jsp?tp=&arnumber=6895371&url=http%3A%2F%2Fieeexplore.ieee.org%2Fiel7%2F4563700%2F6895367%2F06895371.pdf%3Farnumber%3D6895371
http://ieeexplore.ieee.org/xpl/login.jsp?tp=&arnumber=6895371&url=http%3A%2F%2Fieeexplore.ieee.org%2Fiel7%2F4563700%2F6895367%2F06895371.pdf%3Farnumber%3D6895371

[Ashu verma]

http://www.sciencedirect.com/science/article/pii/S0030402612008935
https://www.osapublishing.org/oe/abstract.cfm?uri=oe-15-20-13282
https://www.osapublishing.org/oe/abstract.cfm?uri=oe-18-20-20651
http://ieeexplore.ieee.org/xpl/login.jsp?tp=&arnumber=4785281&url=http%3A%2F%2Fieeexplore.ieee.org%2Fxpls%2Fabs_all.jsp%3Farnumber%3D4785281
http://www.ncbi.nlm.nih.gov/pubmed/2245347
Regards

[GURINDER SINGH]

i have to extend this for 8 channels. what changes i have to made in the parameters??

[Ashu verma]

Hello Gurinder singh,
Your system work well for 4 channels so To make 8 channel WDM system you can simply add 4 more channels and 8×1 multiplexer instead of 4×1.And 1×8 Demux at the receiver side simply by matching the frequency on the both sides.you can start frequency as you are taking in current system channels of frequencies 193.05,193.1,193.15,193.2,193.25,193.3,193.35 and 193.4.For coherent reception you have to place 4 more LO with same frequency as the channel frequency.Let me know if it helps
Regards

[GURINDER SINGH]

okk…i will do it and let you know.
I want to study the various equalizers we can use in this system…. kindly guide me for this.

[Ashu verma]

Hello Gurinder
Can you specify about the equalizers you want to use in OFDM system.Search the forum,as far as i know similar topic has been already discussed on volteera equilizer in OFDM on this forum.Damian also given a file for sample.Tell me if you will not able to fibd the topic then i will provide you the link.

[GURINDER SINGH]

i want to study the effect of both equalizers electronic and mlse equalizer…….please guide me how to connect equalizer in this system and what parameters i should take..

[Ashu verma]

Have you got the link i have mentioned above?

[GURINDER SINGH]

yes i have got it…..but i am not able to understand how to connect it in case of this system??/

[Ashu verma]

Optiwave optisystem file on MLSE guide you better,discussion on this topic has also done .check the optiwave example

Maximum-Likelihood Sequence Estimation (MLSE) Equalizer

Find the same topic on forum for detailed information

[GURINDER SINGH]

i have simulated the system for 8 channels taking the same parameters and first two channels are recovered. The BER is coming 0.4812 for one channel and 0.469 for the other……plz suggest how to reduce it

[alistu]

Hi Gurinder,

Can you please attach the osd file of your eight-channel simulated system? The former file you have attached gives acceptable constellation diagram. And the numbers you have achieved for your two channels shows that the system is probably completely corrupted. Please attach the file so that I can see if I can help you in any way.

[GURINDER SINGH]

i have attached the file….

[Ashu verma]

Hi Alisthu
In your above statement you mentioned that system file that Gurinder attached gives correct constellation but corrupted.Did you mean there is any problem in designed system or you concluded it by observing BER results? Because as far as i know,system was right.The only problem was in launched power.please clarify

[alistu]

Hi Sam Sung,

The constellation diagram I was talking about was the one related to the other file Gurinder had uploaded, which is named ofdm12212.osd. Now he is talking about a new WDM design with eight channels and the BER results he is giving shows that the constellation diagram for this new design is probably corrupted, as the BER has an amount about 0.5 .

[Ashu verma]

Ok Alisthu,got you 🙂

[GURINDER SINGH]

can we use optical filter at the receiver side???

[Ashu verma]

hello Gurinder
You can use optical filter at receiver side,however placement of optical filters in any optical system is very important.Take a look at a research article,reported work proved that with the use of optical filter at receiver side enhance the system performance.

http://ieeexplore.ieee.org/xpl/login.jsp?tp=&arnumber=1325266&url=http%3A%2F%2Fieeexplore.ieee.org%2Fiel5%2F68%2F29327%2F01325266.pdf%3Farnumber%3D1325266

[GURINDER SINGH]

okk….in co-ofdm wdm system, what kind of optical filters can we use??

[alistu]

Since in CO-OFDM the data is processed in the electrical domain and is only converted to optical domain afterwards (or in other words, the digital signal processing in done in electrical domain), I cannot think of any SPECIAL type of optical filter that can be used in here nor I have come into any in the papers I have read so far.

[Ashu verma]

Hi Alisthu
IAs i am not very much familier with OFDm ,but i have noticed a optical filter in ofdm samples givwn by the optiwave in the transmission part.have you noticed that ?

[alistu]

You are right about an optical filter being used in the OFDM samples. But as I stated in my last comment, I do not know any SPECIAL type of optical filters designated for CO-OFDM systems, as signal processing in done in electrical domain. The filter used in here is used to reduce noise from other frequencies and is not specific to the CO-OFDM systems. If you happen to know any, please let me know. Thank you.

[Ashu verma]

Thank you Alisthu,yeah sure if i will find any research article for special flitering in Co ofdm then definitly share with you.Also thanks for explaining about filter used in ofdm examole given by optiwave.

[GURINDER SINGH]

i want to analyse different type of filters like gaussian, rectangular, bessel etc. I have simulated the system with gaussian filter. kindly check it out if there is any mistake…..

[alistu]

Hi Gurinder,

It seems to be all right and when I run the simulation, it works. However, how come you have only used two of the outputs of the demultiplexer? Also, although it makes no change in the result as dispersion is accumulative, but is there any particular reason why you have used two SMF pieces?

[Alessandro Festa]

He is using EDFAs between the span, I think it is the reason why he divided into two pieces 25km eahc of SMF-28

[Ashu verma]

Hello Gurinder ,your system is fine as Alisthu said.You can compare the performance of different filters in ofdm system and find out the solution which filter gives you best results.Alisthu as i have noticed his system,on demux side the roughly use of taking two channel is that ,system also works for other channels too in the terms of Ber.Rest results may vary due to the wavelength dependence of Edfa.

[Ashu verma]

I am agree with Alessandro about the use of EDFa in the system.System on which gurinder woeking on is realized by incorporating symmetri arrangements by using two smf28.it is better to use a small distance fiber followed by the amplifier to get good results.however the dispersion compensation is used to get longer transmission distance.in one of my ofdm wdm system i got good results for even 6600 km using non linear compensation by opc.

[GURINDER SINGH]

yes i have used two smf’s and dcf for compensation…..as it gives better results….i will compare optical filters– gaussian, bessel, rectangular and raised cosine filter and find which is the best…..can i increase the distance further???? what is the acceptable value of ber in this system.

[alistu]

The acceptable bit-error-rate in a real communication system should be below 10e-9 or 10e-12 I guess. However, if you want to design a system which could measure BER so accurately, you need a very long sequence of data. In papers, it is not necessary that BER be that small. The important thing is that your work shows the performance of the system you are analyzing clearly.

[Ashu verma]

Hi Gurinder,i understand the concept of symmetric disperstion compensation that you have used in your system.It works very well in ofdm to increase the transmission distance.the value of ber would be 0 i.e error free if you are using ber analyzer however it would chande if you are using dut set.you can try different filters by inxreasing distance ,i think this would be the way to check the performance.

[Ashu verma]

I jst want to add that you can analyze SNR with the variation of distance .you can then plot the graphs distance vs SNR etc.

[GURINDER SINGH]

okkkk…..in this system i am getting BER as 10^-4 so is this the acceptable value???

[alistu]

For your purposes, maybe. But for a real system? Not absolutely. Please refer to my previous comment.

Cheers

[Ashu verma]

In general as stated by ITU ,ber should be 10^-9 to 10^-12.But if you are getting BER 10^-4 then it is not acceptable.you can decrease value of ber by increasing some power or decreasing distance.have you tried SNR? sNR should be more than 40 at tx side and vary at receiver side accordingly

[GURINDER SINGH]

okk… i will try to vary SNR value
thanks

[alistu]

Can you tell me about the sequence length you are using in your simulations? As a told you, you can measure BER with an accuracy related to your sequence length. The greater your sequence length, the more accurate your BER. for example, in order for you to measure 10e-9 BER, you need your sequence length to be 1000000000!

[GURINDER SINGH]

sequence length is 16384 bits

[alistu]

Okay. The least amount of bit-error-rate you can calculate according to your sequence length happens when only one bit is detected incorrectly. In your case, the BER will be: 1/16384=6.1e-5. You should not expect to get any less than this. Either you get this or you get zero BER if you improve your system performance.

[GURINDER SINGH]

okkk….as per your information if am getting BER 10^-4…so is it acceptable???? if i reduce the sequence length then can i achieve BER less than e-5???

[alistu]

By increasing sequence length you can measure smaller amounts of BER and that’s it! Increasing or decreasing sequence length has no effect of the performance of your system. it just helps view your system performance better. So if your system is bad, for example, there’s nothing you can do with sequence length to make its performance better. You have to change the design and other parameters to boost system performance.

[GURINDER SINGH]

Okkk…so what can i do to improve the BER???

[alistu]

There are plenty of parameters whose changing causes the BER to change as well, within your system design. You can improve launched power by increasing laser power, decrease the number of users in some multiple access systems, decrease BER or/and channel length (at the cost of getting a smaller distance.bitrate product), … . But before doing any of these, you must know why you are designing the system. Are you making the simulation as a part of the university project? What is your objective?

[GURINDER SINGH]

I m doing my thesis work

[alistu]

So you should do it in accordance with your thesis objectives. Unfortunately, I cannot help any further when I don’t know about the objectives.

[GURINDER SINGH]

i have simulated co-ofdm with wdm for 8 channels according to a paper i have read. But i am not able to get the correct results. please help…

[alistu]

Hi Gurinder, Please attach your optisystem simulation file so that I can take a look at it. You have mentioned that you haven’t been able to get the correct results from your WDM-OFDM system, but may I ask if your CO-OFDM system (before you use 8 of them to form a WDM system) has given the expected results?

[Ashu verma]

Hi Gurinder i think we had lot of discussion on this topic and the system files you have attached were corrected at times.As far as i know your 8 channel WDM system was correct for two channel on demultiplexed side.Now what results you need to know?please provide the paper you want to implement? Then we will able to help you

[GURINDER SINGH]

the paper i am implementing is attached…

[alistu]

Thank you Gurinder for sharing the paper. But in order for us to be able to troubleshoot your implemented osd system, you are going to have to upload it in here. I am familiar with the paper. I have made a suggestion in the forum below, which discusses a very similar problem to yours. Please take a look at my answer while you might want to attach your file:

DDO-OFDM-DWDM-64QAM

[GURINDER SINGH]

file is attached here…..

[GURINDER SINGH]

file attached

[alistu]

Your filters’ cut-off frequency is 0.62*Bit rate and your bit rate is 50Gbps; which makes your filters’ cut-offs 31GHz. On the other hand, your channel spacing is 50GHz, which is a small amount and considering your cut-off frequencies, you at least need 31×2=62 GHz channel spacing. I suggest that you put channel spacing equal to 70GHz to make it work.

[alistu]

I think even after you do so, you are going to have to reduce your bit rate and/or your optical fiber length, because the work you are trying to simulate has been done most probably in the absence of the nonlinear effects of the optical fiber. So you have to use shorter length unless you don’t enable the nonlinear effects.

Cheers!

[GURINDER SINGH]

if i want to keep channel spacing 50 GHz, then what should i take as filter’s cut-off frequency??

[alistu]

I think in the case of the simulation you are doing, you can reduce the filter’s cut-off frequency size to ” 0.3*bit rate ” (or even maybe a little less). In that case, the cut-off will be 15GHz and your channel spacing can even be 40GHz. You can use optical spectrum analyzers to observe how the channels are spaced and if that is enough or not. I hope I have made a clear point here.

[GURINDER SINGH]

okk….i will try to simulate it with cut off frequency 0.3*bit rate.

[alistu]

I suggest that you take a look at the following thesis done by the same person as the one giving the paper you are trying to reproduce the results of:

High Data Rate Coherent Optical OFDM System for Long-Haul Transmission – by Khaled Alatawi – University of Denver.

At the end of the thesis, the related published papers are there, including the one you have attached here. The thesis demonstrates the steps to simulate the system.

[GURINDER SINGH]

okkk….how to set OSNR in the system??

[alistu]

You mean to ask how to measure OSNR, right? OSNR is the ratio of signal to noise at the receiver side while the signal is still in optical domain and has not yet changed to electrical signal. I think one way to measure this parameter would be to use optical power meter located in the visualizers library.

[alistu]

I am sorry. You can use a component named “set OSNR” located in optical sources in the components library. This component makes your desirable optical SNR by adding the proper noise floor level according to the SNR you wish to have. This component has been added to the Optisystem library in version 13. However, if you merely want to measure SNR you can do as I told you before.

[GURINDER SINGH]

okk….where i should place set OSNR component in the system??

[alistu]

I haven’t actually worked with the component, but in the component help section it says that the component adds a defined noise floor to the signal. It somehow implies that the input should be the signal that you want to add the noise to. The component measures the amount of signal and noise in the input, and adds the noise according to them in the output.

[Ashu verma]

Hello Gurinder
I have read the discussion abut your system and use of sample file.I just want you to add WDM analyzer after total distance of your loop and run the system.Open WDM analyzer then scroll the under tab to extreme right ,here you go to see the OSNR at each wavelength.
Try it and let me know,i havent a access of optisystem 13 so i couldnt help you to do this.I will definately see your system file and will try to change if any possible modification required. Thank you

[GURINDER SINGH]

okkk…..i have set filter cut off frequency as u said…but i am not able to get the correct constellation diagram.

[alistu]

I also told you to reduce the number of loop iterations. Since you are using Optisystem v.12 OFDM components and they don’t have any type of equalizers like sending pilot signals and so on, you will not be able to get the same results as when you use optisystem 13 OFDM components. Set “loop iteration=1” for the start.

[GURINDER SINGH]

even by setting loop iterations to 1, i am not getting the correct results.

[alistu]

Maybe the bit rate should also be reduced for your system to give good results. I suggest that you put the bit rate equal to 10Gbps just to see if this works. you can also change the loop iteration to zero to see if the system works. If the system still didn’t give good results, then there is some other problem with it. Let me know of the result.

[GURINDER SINGH]

if i take bit rate 10gbps an loo iterations as 0 then there is a liitle improvement in the results.

[alistu]

Please attach the file with the adjustments I have told you to make. however, my guess is if you have correctly changed all filters cut-off frequencies , then you might have made a mistake calculating the power of optical amplifiers that are used in the fiber optic channel. i am waiting for you to attach the modified file.

[GURINDER SINGH]

file is attached…

[alistu]

I got your file and I am analyzing it now. can you tell me if you have implemented the whole system by yourself or perhaps you have used optisystem’s sample file and then you have made the necessary adjustments to that? the overall configuration is pretty much the same, but there are minor differences between some values.

[GURINDER SINGH]

yes i have used optisystem sample “coherent detection wdm optical ofdm”

[alistu]

So it appears you have made some changes in it, for example the constellation visualizer after the OFDM block which doesn’t have any use. Beside, some parameters are different from CO-OFDM sample. I recommend that you use optisystem CO-OFDM (not WDM) sample and then make 8 copies of it. Change the center frequency of each one and you will have a WDM CO-OFDM system that will work for sure.

[GURINDER SINGH]

okk. i will try to design it using co-ofdm sample.

[alistu]

I have honestly tried to find the problem but that I’m afraid that suggestion is all I can give to you Gurinder. there was a similar case yesterday which is hopefully solved by now doing the same thing. After implementing the system the way I told you to, please attach it in case you encountered any problems.

[GURINDER SINGH]

thanks alistu……damian i will add wdm analyser and let u know the OSNR.

[GURINDER SINGH]

damian….should i provide OSNR values of the modified system as guided by alistu or my original system??

[alistu]

Hi Gurinder, I think Sam Sung (who you have mistakenly confused with Damian) suggested the WDM analyzer just as a way to measure your OSNR. Otherwise the implementation problems you currently have will not be solved using this component. Therefore you’d better implement the system the way I told you so that it runs without any problems and then use WDM analyzer to measure the OSNR, or even “set OSNR” to set the OSNR.

[GURINDER SINGH]

I have modified the sample file and simulated with the sample parameters…..with 50 km SMF and 10 Km DCF. I am getting BER 1 for first channel and 0 for the second.

[alistu]

Hi Gurinder. Could you please upload the optisystem file so I can check it? There may be a problem concerning the central frequency of the channels. Maybe one of the lasers frequency is not set to the amount it has to be. please attach the file and I will let you know about the problem as soon as I analyze it.

[GURINDER SINGH]

okk…file is attached…

[alistu]

Thank you. I am running the simulation and it may take several minutes for the software to finish simulation. I think should have better asked you to just implement a WDM-OFDM system with two users. If that system would have worked correctly, then it was possible to expand it. That way simulations took less time than now. Let’s do this from next time.

[GURINDER SINGH]

okkk……let me know the mistakes when the simulation will be over..

[alistu]

Simulation is done and the good news is that it runs correctly and your problem is solved finally. The reason for the BER=1 in one of the channels is the symbol rotation problem. So if you put laser initial phase in the receiver of the user with bad BER equal to 110 in this case, you will have no error. Another thing you must pay attention to is the value of amplifiers in the line, which should correspond to the loss of the line.

[ali]

hi alitsu

you said:The reason for the BER=1 in one of the channels is the symbol rotation problem
But the constellation diagram does not show large spin.
The number 110 seems a little difficult to guess.
You can give me a little guide?

[alistu]

Hi Ali, As I mentioned I found the required phase shift to correct the constellation by error and trial, which means for example I changed the phase 45 degrees at first and observed the result, then changed it again and maybe again to get the right value for BER. By the way, if the error is around 0.5, it means each of the four symbol groups are in the adjacent quarter to the one they have to be in, and if the error is near 1, the case is as I have depicted in the two images we were discussing. Thought this might help you in error and trial.

Cheers!

[ali]

Very good’ve described.
Thank you

[alistu]

You’re welcome Ali!

[ali]

hi alitsu
some questions
1- bit rate is set in two places, one in the global parameters and other in the PRBS GEN
What’s the difference?
is have a relationship with toghter?
Because if change bit rate in global parameters result BER is different
2-One advantage of ofdm set cyclic prefix for avoid isi
Where should determine this parameter?i change in ofdm mod. but i did not see change in the BER
3-where i do download optisystem 13.0.3 ? because i have v. 13.0.2

[alistu]

Hi hadi,

1- Bit rate in PRBS is usually set as bit rate in the global parameters by using script mode. However, the value you choose for the one in PRBS is the one used in the simulation as Bit rate.

2- In order for the cyclic prefix effect to be seen, you need to use BER Test Set instead of BER Analyzer. I have explained that explicitly in this very page.

3- You can download Optisystem 13.0.3 64 bit from “Evaluation License Download” section in your profile.

Cheers!

[alistu]

I would strongly suggest that you use BER Test set instead of BER Analyzer for OFDM systems. The reason for this has been mentioned in the following forum by Optiwave team:

RoF with QPSK doesn’t work

In the file attached, I have done it for one channel. You can do it for the rest. Also I have corrected your amplifier values. I hope your problem is solved now.

[GURINDER SINGH]

okkk…i will look into it…..thanks for ur help

[alistu]

You’re welcome. Let me know if we could solve your problem. And if you use the new OFDM component offered in Optisystem 13, you will not have the rotation problem. Because in the new component, pilot symbols are employed to give an estimation of the channel and modify the symbols based on the estimation of the channel.

[GURINDER SINGH]

i am not able to open your file…

[alistu]

I downloaded and opened the file I had attached and there was nothing wrong with it.The only explanation for this problem is the fact that you are using an older version of Optisystem than I do. I am using v13.0.3. What is the version of your software? If that is the case, there is nothing you can do.

[GURINDER SINGH]

i am using optisystem 13.0….kindly tell the changes you have made or send the snapshot of it…

[alistu]

For one of the users that you have used BER analyzer, do the following:

1- Delete NRZ reference, NRZ pulse generator, the fork before the transmitter subsystem and BER analyzer.
2- Choose BER Test Set component from Test Set library of the Optisystem component library.
3- Place it near the QAM decoder. connect its input to the output of QAM decoder and its output to the input of the sybsystem related to your user’s transmitter.

[alistu]

I also made a change in your line. I adjusted the amplification of your optical amplifiers to the amount needed for the line. You have used a 50 km SMF with 0.2 dB/km attenuation. So the overall loss will be 50×0.2=10 dB. So change the amplification of the next amplifier to 10 dB. In the same way, you will have 4dB amplification for the amplifier after the DCF.

[GURINDER SINGH]

okk..i will make the necessary changes and let u know…
thanks

[alistu]

You’re welcome. And if you are interested to know more about the symbol rotation problem that I was referring to, you can have a look at the discussion below that I started about two months ago:

Symbol Rotation in OFDM

Different WDM channels face different rotations, simply because they have different wavelengths and the line has a different response to different wavelengths.

[GURINDER SINGH]

okk i will surely read it….what about the parameters of BER test set.?
i have tick the the ofdm block…subcarriers set to the value used in ofdm modulator. Is this right??

[alistu]

Now that you are using the OFDM component from optisystem 12 you don’t need to tick the OFDM box and you don’t need to set any value for the OFDM part of the BER Test Set. That part is used for Optisystem 13 OFDM component when you have pilot signals. Without that part, you are able to get the correct results.

[GURINDER SINGH]

okk….what value in BER test set gives the BER?? How i will get the graph of BER and Q factor??

[alistu]

In order to see the values measured by BER, you need to right click on the component in the layout and choose component results. Now in the component results window if you tick the display box for each of the parameters, it will be shown in the layout. I suggest that you tick BER, Bit Errors, and Sequence length per iteration.

[alistu]

As for what you asked about Q factor, I made a search and apparently this component doesn’t measure Q-factor. It empirically measures the BER and this is one important criteria in many papers. If you want to know why it is better to use BER Test Set in your system, have a look at the link I sent to you in my previous replies.

[GURINDER SINGH]

okk. can i now implement paper mentioned above on this system??

[GURINDER SINGH]

after simulation i am getting BER for first channel as 0.9991 and 0 for the other

[GURINDER SINGH]

file is attached…..

[alistu]

Again you are faced with the problem I told you about: Symbol rotation. Symbol rotation happens because of the delay induced by the optical fiber. Whenever this problem occurs while you have a clear constellation diagram (as you have in this example), try to change the initial phase of the laser corresponding to the user.

[alistu]

In these images I have explained the matter to you and others who might find this problem: The symbols in im1 are not in their own corresponding quarter. Why? because they have rotated and I told you the reason why. In the second image (im2), by applying a proper initial phase to the laser, they are rotated back, well, not exactly to their own place but at least to their corresponding quarter and that makes the detection correct in this example. Just so as you know, I applied initial phase=-145 in this case (and you can see this by comparing the images). I hope I made it clear to you now.

[ali]

hello alistu
How did Marking symbol?
Why did you symbol into these numbers?

[alistu]

Hi Ali, The reason that I used the symbols is to make it clear for Gurinder what happens to the symbols after changing the receiver laser initial phase, as a way to compensate for the symbol rotation imposed by optical fiber. I used windows “paint” to number the symbol groups.

[ali]

Hi alistu
No, I mean, you said it was not
I mean, the question was why the symbol (1) in the third quarter numbers did you?
Do you come from experimentation to -145 number? Or did the calculations?
thanks

[alistu]

Hi Ali, The numbers I used for symbol groups correspond to the quarter they have to be in. Before applying the phase shift I realized from BER, which was around 1, that I had to shift symbols. Then using error and trial I got to the number 145 for phase shift.

[GURINDER SINGH]

okkk…so i should apply initial phase to -145….

[alistu]

Yes, please do that and keep in mind that you should always do that for every channel of WDM-OFDM systems that are implemented by optisystem 12 OFDM component. And different channels see different dispersions, so this is going to be hard for the systems with several channels. And when you change the channel length, the phase rotation will change.

[GURINDER SINGH]

once i have solved this problem can i now implement the paper on this system??

[alistu]

I think I told you that there are some assumptions in this work that are not mentioned in the paper. Even this symbol rotation compensation is one of them. The paper doesn’t state anything about any type of equalization. And this paper has neglected nonlinear effects of the fiber as far as I know. So if you neglect them you might be able to reproduce the results.

[GURINDER SINGH]

okk … i will try to implement the paper

[ali]

hi
Why file attached with high O.S.N.R but is bad performance? What are the defect?

[alistu]

Hi ali, I see you have disabled self phase modulation and that’s why you have been able to simulate for such high power. But optical fiber has other destructive effects and I think in your case the problem is the length of the fiber (600km) without any equalizer and the symbol rotation. And as you know, each of the optical amplifiers have a noise figure as well.

[ali]

hi alitsu
To resolve the problem with the length of the fiber, and even more, what should I do?

[alistu]

Ali, I suggest that you use the OFDM components in the new Optisystem version, which make use of the OFDM pilot symbols as a way to estimate the optical fiber channel and therefore give much better results than you are getting now. In other words, the new component uses pilot signals as an “equalization” method.

[ali]

in section “BER TEST set properties\ofdm parameters\pilot symbols” i do change pilot symbols and change other parameters in the section but no better result.

[alistu]

BER Test Set just calculates the BER and does not affect the system. If you are not using pilot symbols then you should not change pilot symbols from zero or otherwise you will get the wrong results. If you have access to optisystem 13, then use the new OFDM components and then change pilot symbols accordingly in the BER Test Set.

[ali]

i use optisystem 13.
in section ofdm mod. no appear item pilot symbols .
if possible Please help clear
thanks

[alistu]

That’s because you are using optisystem 12 component in v13. As you can see in the image attached, the right block with five ports is the new OFDM component which has the aforementioned features, whereas you have been using the left one for your simulations. Please refer to optisystem samples for examples of the new component being used.

[ali]

i said,”in section ofdm mod. no appear item pilot symbols ” I mean, change the parameters IN ofdm modulator
but appear pilot symbols in section BER test set
if possible Changes you saying apply in sent my file?

[alistu]

Please attach a screenshot of the problem you are referring to if it is possible. I did not fully understand what you mean. Thank you.

[ali]

I sent the file

With these changes CONSTELLATION no better

[alistu]

What about the BER value? Can you attach the file also?

[ali]

i attach constellation+BER result

[alistu]

Thank you, but by “the file” I meant the osd file in which you have used new OFDM component. Also, you can start your work with an OFDM optisystem sample and make the necessary changes. have a look at the similar case in this topic which was solved by doing so.

[mukalla pavan]

hi
i am working on coherent detection 4-QAM WDM optical OFDM. i got BER of 0 with 4QAM. but when i am using 16QAM ,getting BER of 0.4809

so please suggest me what are the changes needed.

i am attaching the file

[Author]

Posts