- This topic has 16 replies, 3 voices, and was last updated 8 years, 7 months ago by Damian Marek.
-
AuthorPosts
-
-
April 27, 2016 at 8:22 am #38562Irfan MorshedParticipant
Hi everybody,
My project is basically about a 3 user OCDMA system for 16 QAM, where I have done spectral amplitude coding (SAC) using FBGs. I have used BER test set and the BER I am getting is 0.5 which is bad, and the eye diagram is also horrible. I have used optical amplifier of gain 10, FBGs encoded and decoded using Walsh code (you can check the wavelength and bandwidth values, I have done it by using complementary/balanced detection), laser linewidth of 300 GHZ ( since wavelength = 2.4 nm), and sequence length = 2048 bits. I also changed sequence length to 65,536 bits to support BER test set, still no change in results. All other design parameters have ideal values. However, the M-ary threshold detector values are kept as the default values for 64 QAM. The constellation and eye diagrams are both horrible.
Could anyone please be kind enough to help me locate and solve the problem? Your one tiny help can make things much easier for me here. My thesis work is getting badly hampered due to this. I have been brainstorming since weeks, still no outcome.
Please anyone help me out! Its a sincere request.
Please.
-
April 27, 2016 at 8:45 am #38564Irfan MorshedParticipant
Here’s a brief on how Spectral Amplitude Coding (SAC) was done using Walsh Code :
We have 4 wavelength values –
lamda 1 = 1550 nm
lamda 2 = 1550.8 nm
lamda 3 = 1551.6 nm
lamda 4 = 1552.4 nmWalsh code for Encoding :
User 1 (1010) : 2 FBGs at encoder coded with lamba 1 and lamda 3 (lamda 2,4 are cut off)
User 2 (1100) : 2 FBGs at encoder coded with lamba 1 and lamda 2
User 3 (1001) : 2 FBGs at encoder coded with lamba 1 and lamda 4Walsh code for decoding using complementary/balanced detection :
User 1 (1010) : 2 top FBGs at decoder coded with lamba 1 and lamda 3, 2 bottom FBGs at decoder coded with lamba 2 and lamda 4
User 2 (1100) : 2 top FBGs at decoder coded with lamba 1 and lamda 2, 2 bottom FBGs at decoder coded with lamba 3 and lamda 4
User 3 (1001) : 2 top FBGs at decoder coded with lamba 1 and lamda 4, 2 bottom FBGs at decoder coded with lamba 2 and lamda 3 -
April 27, 2016 at 8:57 am #38565Irfan MorshedParticipant
Also at the transmitter side, upconversion from baseband to passband is done, from 2.5 Gigs to 60 Gigs. And at receiver side, 60 gigs RF passband signal is converted back to 2.5 Gigs baseband signal.
-
April 28, 2016 at 1:46 pm #38600Damian MarekParticipant
Irfan,
I had written a long response, but apparently it was not posted properly! I’ll post the main points here, but in that project you had been using the white light source, which is I think the correct way to do it.
The main issue is that the white light source was only generating noise bins and not a sampled signal. The proper way is to convert the noise bins, as done in our OptiSystem example in the \Metro and access systems\PON and CDMA systems\ folder.
The other problem is that the signal rate is pretty low when compared to the bandwidth of these SAC-OCDMA systems. Following the Nyquist sampling theorem you need at the sample rate to be at the very least twice the bandwidth of your signal. Your electrical signal is approximately 60 Ghz in bandwidth (in fact a bit more), so the sample rate should be at the very least 120 GHz. I would suggest a bit more. It seems like you have changed the project from the previous topic you made because here the Bit rate is 2.5 GHz.Previously the Bit rate was only about 30 MHz, so I calculated that you would need 8192 Samples per bit to correctly describe this large bandwidth signal. Attached is the previous project.
Some of the filters were also to low to pass the actual signal, so I modified them. I really suggest you try designing systems starting with a back to back configuration, that is no optical fiber and remove the noise generated from the photodetectors. The communication systems are complex and one should always try the simplest transmission scheme before moving on to an actual physical implementation.
Although there are still some details to iron out, the attached project is close to working. Hint: You might want to play with the power levels of the transmitter.
Damian
-
April 28, 2016 at 2:42 pm #38605Irfan MorshedParticipant
Thanks a lot for your feedback brother. I will check your osd file from my workplace tomorrow, since no optiwave at home.
But I need to somehow show a decent enough eye diagram and BER value for a conventional 16 QAM system, for performance analysis purpose.
Does your sent osd file work with regards to that?
Please kindly let me know now. I am mega-stressed because of not achieving desirable results since weeks. I hope your updated file managed that brother.
I just hope.
-
April 28, 2016 at 3:25 pm #38606Irfan MorshedParticipant
Hi,
In the osd file that you have sent me, apparently the minimum BER = 1.
And the project is all about achieving a BER less than 10^(-9).
Also Q factor and eye height is zero.
But you said your file is close to working? How is that?
I am highly confused.
-
April 28, 2016 at 4:58 pm #38608Damian MarekParticipant
Yes the project is close to working but not quite yet. Try increasing the signal power, when I was using your project I found the receiver power quite low and I do not think the signal was being reproduced faithfully at the output.
-
-
April 28, 2016 at 3:26 pm #38607Irfan MorshedParticipant
I dont think your osd file works.
-
April 28, 2016 at 5:08 pm #38609Irfan MorshedParticipant
I used an optical amplifier with gain 10 and noise margin 4, as you can see in the attached osd file with the post (at the top). The sending power becomes 66 dbm and received power becomes 26 dbm roughly. What should I do now?
Also did u make any changes in the osd file that you sent me? It seems to be same as my old project with no change.
-
April 30, 2016 at 2:56 am #38663Irfan MorshedParticipant
Why cannot it be calculated?
-
April 30, 2016 at 3:04 am #38665Irfan MorshedParticipant
Since my project is designed on version 14, and nobody apparently has that version, plus my college submission is due in next few days and I have no time on hand, so I am taking help from this project.
It was in the forum, and some of you already solved it once. But when I ran it, it says unable to be calculated. I thought it was a successful project, but it is not even running.
Why is that so?
-
May 4, 2016 at 2:42 pm #38844Damian MarekParticipant
It does seem that the older version of that sample project does not run in OptiSystem 14.0. You could probably fix it by removing the power combiner/splitter and replacing them with components from the default library. The reason being these components were changed in the latest release.
You can find these sample files on your own PC in the samples folder. Usually C:\Users\%YOURUSERNAME%\Documents\OptiSystem 14.0 Samples\Metro and access systems\PON and CDMA systems.
-
May 4, 2016 at 2:52 pm #38845Damian MarekParticipant
By the way, it is proving a bit more difficult to get your original project working. There seems to be an important relationship between the FBGs’ bandwidth and the actual bit rate of the signal. These problems have more to do with understanding the theory then the software however. The main points you were missing from the simulation perspective were those highlighted above. Now the work is to correctly design the system to work for a QAM type modulation.
-
May 5, 2016 at 3:20 am #38868Rajguru M. MohanParticipant
thank you Damian Marek sir for your response and guidance on OCDMA.
I also find difficulty in OCDMA modulations.
Rally you clear all doubts related to it.
Thanks alot -
May 5, 2016 at 3:24 pm #38908Irfan MorshedParticipant
Mr. Damian,
As you have previously mentioned, since I am upconverting 2.5 GHz signal to 60 gigs, you said I would need 8192 samples per bit. I made the changes and improved signal filtering, and added all your suggested corrections.
This time, the project is taking hours to calculate. I cancelled calculation after it exceeded over 2 hours!
Osd file attached.
May I kindly know what is the issue with this one?
-
May 9, 2016 at 8:40 am #39022Damian MarekParticipant
This is why I downconverted the signal to 2GHz other wise the required bandwidth is very large. There is not much you can do here, if you want to simulate this large of a bandwidth signal, you will be forced to use that many samples and require that long of a simulation.
-
-
May 5, 2016 at 3:27 pm #38910Irfan MorshedParticipant
Suggestions by others are also most welcome and eagerly anticipated! 🙂
-
-
AuthorPosts
- You must be logged in to reply to this topic.