error when I simulated a bidirectional system

[Joseph Jiang]

hello everyone，I want to know why it has an error when i set the bit rate above 512bps using the bidirectional device.It will be ok
when the bit rate below 512bps.can anybody know this problem?thank you!

[Damian Marek]

Can you please attach your project file? It might be a sampling rate problem, or the fact of having a very slow bit rate.

[Joseph Jiang]

Dear damian,I am so sorry to reply you late.I have upload my attachment（optisystem 7 in win7）.I want to simulate a system to demnonstrate the influence of EDFA in the bidirectional optical link.In my application ,i want to set the time window to almost 1 second.however,It occurs to a error I refered to last time.is it because that my computer is out of memory.(my computer has 4g memory ).I also have some questions to ask you!

1、it seems that The device optical power meter does not measure all the optical power but some wavelength power near the center wavelength.
2、I want to know whether the OSNR the device wdm analyzer measured include the information I modulated to the laser.
thank you for you reply in advance.

[Damian Marek]

Hi Joseph,

There are a couple problems I saw with your project file. You need to either add signal delay to every bidirectional component manually or check the Initial Delay box in Layout Parameters->Signals. For more information on how we model bidirectional devices please review our tutorial at:

Lesson 5: Bidirectional Simulation — Working with Multiple Iterations

1. The power meter should calculate the average power over all wavelengths.

2. Yes, it should include the modulated part in the signal.

Finally, for a bit rate of 10 Gbps simulating a time window of 1 second will require a lot of memory. For a samples per bit value of 64, the total number of samples required is ~640e9, and so using a 8 byte double as an estimate the total memory required is roughly 5120 GB’s!

[Joseph Jiang]

HI,Damian,i appreciate your reply.I have modified the errors you told me.I set the bit rate 100Mbps.But the error was the same.I think it is also the memory”s problem.I want to know if i want to simulate the bidirectonal device,how can i set the global parameters ,bit rate ,samples per bit etc.

[Damian Marek]

Set the bit rate to the experiment or design you want to test, usually anywhere from 1-2 Mbps to 100 Gbps. The samples per bit determines the accuracy of the simulation, generally 32-64 is perfectly fine. Finally set the sequence length to determine how many bits you will simulate. A larger number of bits will make BER calculations more accurate, but generally 1024, is a good number to start with.

[Joseph Jiang]

Dear Damian,I Know you said to me.In my application,I want to set the bit rate 100mbps,the time window 1 second.however,as you know the sequence length should be set very large and because of the useage of bidirectional device like bidirectional optical fiber,It will spend so many time to simulate sucessfuly(may be one day). what’s more,the momery of computer should be enough large as you said(my computer‘s memory is 4G).Damian，do you have any solution to my problem.thank you for your help in advance.I Will wait in hope for your reply.

[Damian Marek]

There might be another way to simulate what you want. Is there a specific reason you want to simulate a time window for 1 s?

[Joseph Jiang]

Dear Damian,I want to simulate a TDM system as the project I uploaded.I have thought over my application maybe 1s time window is not necessary.my application is a time transfer via optical link system.The remote signal will be sent after 1 second.IF I want to set the time window 1 second ,i am so interested to the solution you said another way.By the way ,what is the difference between optical delay and time delay.
thankyou so much for spare your valuable time.

[Damian Marek]

Hi,

From my point of view you should not need to simulate the system for a full second, and in fact it will just not be possible. The time scale difference between gigahertz (10^-9 seconds) and even megahertz (10^-6 seconds) is just too gigantic. If you are looking at a transient part of the system you could just set the remote signal to be activated at say around 0.1 microsecond.

A time delay simulates the physical effect of time lag. In effect it will shift the optical signal right in the optical time domain visualizer. You could use it to model the estimated time it takes light to travel through a fiber or say through free space.

An optical delay on the other hand is a simulation tool used in bidirectional components. It aids in initializing bidirectional components. There is a better explanation of how bidirectional systems are simulated at:

Lesson 5: Bidirectional Simulation — Working with Multiple Iterations

Cheers

[Joseph Jiang]

Dear Damian,I have still a question about the device time delay.As far as i know,the signal is translated period once the sequence length setted.I wanna if it is cyclic time delay when the TIME DELAY is selected.how ever,i have made a project as uploaded file to vertify my suppose.
the bit sequence was set 101101011 and it was transmitted via a time delay device.The delay time was set 1/bit rate,that is to say ,the bit sequence will be delayed one bit.however,i found it was not the same as i thought.I need your help!thank you!may you have a very nice day !

[Damian Marek]

I just did a quick compare of the optical time domain and it seems to be working correctly to me. What is the problem you have with the results?

[Joseph Jiang]

dear damian，There are no problem when you use optical time domain visualizer.last time ,i selected the Dual Port Oscilloscope Visualizer and i can not find the difference.I want to know why it can be seen difference in optical domain rather than electrical domain. thank you！have a great time!

[Damian Marek]

Hi Joseph,

The issue here is that the output power of the electrical signal is very low compared to the input so you can’t even see the output signal. Try zooming into the microWatt range to view the second signal or incorporate a electrical gain component so that they are of comparative size.

[Joseph Jiang]

yes,you are right.than you for your answer.have a nice day!

[Joseph Jiang]

HI,Damian，thank you for your reply.In my project,I want the remote optical signal to be sent after receiving the local optical signal.That is to say,I expect the remote optical signal to be delayed equal to the time that local optical signal transfered to the remote receiver via optical link.however ,i think the time delay device is what i need,but i not sure .I want to get your help.additonaly,I have other questions:
1、 when should I select the parameter “convert to noise bin”in the global parameters
2、can I set color of the plot line in the report.
3、I do not konw how to use the device”electric carrier analyzer”.why are there two filter and how to set the frequency and bandwidth?\
thank you in advance!

[Damian Marek]

Hi Joseph,

I’m having trouble running your simulation. I’m not sure exactly why at this moment, does it work for you? As for your questions:

1) For your application I would just convert noise bins. The link below might help you better with understanding how they work. Basically they function as a computationally inexpensive way to include noise in a simulation.

EDFA Noise center frequency

2) The color of your line can be changed by clicking the top left tag on the figure. There should be a drop down menu that appears and then you can navigate to Properties->Curve.

3) If you right click the component and go to Component Properties you can set the filter frequency and bandwidth from there.

Cheers! (I’ll keep working on your file to see where the problem is!)

[Joseph Jiang]

Hi,Damian,
I can run my silulation ,but it may take some time.I have not know whether it is right to use the device “time delay”in my application .
As for my questions:
1、The example you gave me does not tell me the function of ‘convert noise bin’.It only told the center frequency and bandwidth of noise.I just want to know when it should be selected.In the doc’optisystem_getting_started ‘says that “Selects whether noise within a sampled band’s frequency range is added to the sampled signal or represented separately as noise bins. The default value is disabled, which means the noise propagate is separated from the signals. It can affect the Erbium doped fiber amplifiers and the photo detectors”.I REALLY WANT TO KNOW HOW IT CAN AFFECT THE Erbium doped fiber amplifiers and the photo detectors.
2、I have found what you said.
3、I just want to know why are there two filter in the device”electric carrier analyzer”.I want to measure the electric SNR.is it right to select the componet.
I am waitting for your reply!thank you for your help!

[Damian Marek]

Hi Joseph,

You are using the time delay correctly!

1) Noise bins are a convenient signal representation for noise. Each “bin” is given a bandwidth and amplitude, thus a wide range of frequencies can be modeled and makes calculating SNR much easier. A sampled signal will need a much higher sample rate to model a wider range of frequencies and it is not quite as easy to determine the power of the noise.

What the convert noise bins option does is when a sampled signal overlaps the noise bin signal the noise in the noise bins is converted to a sampled signal and added. Normally I would suggest using the convert noise bins (in OptiSystem 13.0 this option is now set by default), because you are not working with a WDM type network, which would benefit from the large frequency representation. That being said since you are interested in the electrical SNR value, you will HAVE to use noise bins. Our WDM Analyzer can calculate the SNR of a sampled optical signal but the electrical carrier analyzer is not able to yet.

Another reason why you would choose to convert noise bins is because some types of models don’t modify noise bins, because they might be a time domain effect, like nonlinearity, so converting the noise bins will increase the accuracy of the simulation, although since noise power is usually quite low it has a minimal impact. Let me know if this was a sufficient explanation!

3) You can use this component fine and it will compare noise bins to the signal power to calculate the SNR. There are two frequencies simply to allow you to compare values. You need a filter to be able to define the power of a frequency signal. Other wise the “power” is given by W/Hz.

Hope this clarifies the issues!

[Joseph Jiang]

Excellent Damian,thank you for your reply.It really help me a lot.however,when i was using the optisystem,I have found that the default set of noise bins spacing and noise bandwidth in many devices is not the same.for example,the noise bins spacing and noise bandwidth of Directly Modulated Laser Measured are 1THZ and 100GHZ,but 13THZ and 125GHZ in Bidirectional Optical Fiber.I wander if it is necessary to set them the same to begin my simulation.thank you !may you have a good day !

[Damian Marek]

No you should not have to set them manually. OptiSystem will automatically re-size them to preserve the most accurate version of the noise bins after adding them.

[Joseph Jiang]

OK，thank you for your reply.how about the noise center frequency?does the optisystem set it automatically?thanyou so much!

[Damian Marek]

There is a default center frequency for most components. Components like Lasers set the noise center at the lasing frequency, whereas the Erbium doped fiber has an explicit parameter in the Noise tab to set the center frequency.

[Joseph Jiang]

DAMIAN,I have found that if i select convert noise bin(the noise that overlaps the signal will be converted to sampled signal,the sampled signal power will be added ),the osnr and snr measured by wdm analyzer will be higher than not be selected.I think that is not the exactly right osnr and snr.how about electrical carrier analyzer?is it measured the right SNR if i selected convert noise bin,since i am interested in the snr in electrical domain.thank you !

[Joseph Jiang]

dear damain,could you give me a help.I have confused for this question a long time and i have also asked the qusetion here https://optiwave.com/forums/topic/convert-noise-bin/#post-15316

[Author]

Posts