Bit Error Rate

[SAHIL SINGH]

Hi Everyone on the Optiwave Forum,

I have a very important query which i feel is worth sharing and discussing on the forum… Can anyone please tell me whether or not Bit Error Rate (BER) is dependent on the type of sequence generators (PSEUDO RANDOM OR USER DEFINED )? In one of my designs on the optisystem software i noticed that if i vary the user defined SEQ Generator with Pseudo Random SEQ Generator.the bit error rate (BER) also changes?Can anyone please explain to me the reason behind this ? Hope to hear from all of you soon…

Thanking in anticipation

Regards
Sahil Singh

[Dr. Dhiman Kakati]

Hi Sahil How much deviation of the result you have noticed please let me know. In most general cases the PRBS generator is used as the bit sequence generated is unique every time with “generate random seed enabled.” In user defined one what are the parameter you have modified for simulation that also a deciding factor for BER output.

Regards,
Dhiman

[jyoti raina]

Hi Sahil Sir.. Hello Dhiman Sir..
I also want to ask if there is some difference or impact of the pseudo random binary sequence generator.. I have never thought on it..
Thanks Sahil sir for bringing up such a good question.

With regards
Jyoti raina

[Dr. Dhiman Kakati]

Hi Jyoti,
I am not getting what kind of difference you need for pseudo random binary sequence generator with which component. I am confused.

Regards,
Dhiman

[jyoti raina]

Hello Dhiman Sir..
I meant what sahil sir has mentioned about the change of BER with respect to changing of PRBS generator.. I hope you got my point. Thanking you..

With Regards
Jyoti Raina

[Dr. Dhiman Kakati]

Hi Jyoti,
If your design is ok then there should not be any significant change at the output BER Because we do not design a system that is specifically for a specific sequence. PRBS means input data and it can be any (Bit here) and system performance (i.e.) should not change.
Thanking you for the more explanation.

Regards,
Dhiman

[SAHIL SINGH]

Hi all.
I think i got it now. I think every time when i take a different bit sequence or the pattern of bits the bit error rate varies. It may be because of different bit pattern that causes the bit error rate to vary.
Anyway thank you jyoti and Dhiman for your efforts.

Regards
Sahil Singh

[Dr. Dhiman Kakati]

Hi sahil, The BER should not change significantly (may be there negligible change) as input data can be of any type and the system behavior should be the same.

Regards,
Dhiman

[jyoti raina]

Hello Sahil Sir..
I agree with dhiman sir.. The BER should not be affected that much.
May be it is because of random seed problem that your BER is varying..There has been lot of talk about random seed option.. Sahil Sir check if you have unchecked it.Thanking you.

With Regards
jyoti raina

[Ubaid Bhat]

hi sahil singh….
as already mentioned by dhiman kakati the BER of the system should not change very much by enabling the random seed in prbs generator.if your system design is good then there should not be any change in BER with changing the bit sequence. we simulate the design for different bit sequences because in practicality there will always be different bit sequences that will go down the system.so i think you should look into your design for this.hope you got my point.
with regards
ubaid bhat

[SAHIL SINGH]

Hi all,

As far as the point of view of Dhiman is concerned that the BER should not change significantly (may be there negligible change) as input data can be of any type and the system behavior should be the same… I completely agree with your point.. I had a made a mistake in my design and just corrected it… The design seems to work okay now… I thank you so much for your concern and valuable suggestions… Really glad that you took time in order to resolve my query… All the best

Regards
Sahil Singh

[jyoti raina]

Hi Sahil Sir..
Glad to know that your problem is solved now.. I also thought the same but i was just waiting for your response.. I think Dhiman sir was right the BER should not change significantly (may be there negligible change) as input data can be of any type and the system behavior should be the same.. Anywayz a good discussion. Thanking you..

With Regards
Jyoti Raina

[SAHIL SINGH]

Hi ubaid bhat,

I really thank you for your concern and i would like to let you know that i have made a note of what Dhiman has said… Really appreciate your efforts..
Regards
Sahil Singh

[Ranjeet Kumar]

Hi Sahil singh,
Yeah, I agree with thepoints that BER results will chane with changing types of signal.
Here i am uploading the image which contains the formula of calculations of BER and Q-factor.
The two parameter Algorithm defines the numerical method to use for the BER estimation namely Gaussian and average gaussian.
In this we could see the factor on which BER depends are where P0 and P1 are the probabilities of the symbols, M is the number of samples for the logical 0, and N is the number of samples for the logical 1, average values and standard deviations of the sampled values respectively, and S is the threshold value.If the signal is mixed with the noise, the Average Gaussian method is modified to calculate the average error patterns. The the Average Gaussian method can estimate the BER per bit or per pattern, the Worst-case Gaussian searches for the min BER for each bit or pattern instead of calculating the average values
So, BER depends upon types of signals.

Hope i could explain it correctly.
Seeking your response.

[SAHIL SINGH]

Hi Ranjeet Kumar,

Thanks for proviving the useful information…. You have mentioned it correctly that I BER results will do change when we change the type of the signal..thamks for providing the image containing the formulas of calculations of BER and Q-factor… I appreciate your efforts for providing Very nice and detailed explanation..

Regards
Sahil Singh

[Manoj Kumar]

Hi Sahil singh,

Yeah, I agree with thepoints that BER results will chane with changing types of signal.
Here i am uploading the image which contains the formula of calculations of BER and Q-factor.
The two parameter Algorithm defines the numerical method to use for the BER estimation namely Gaussian and average gaussian.
In this we could see the factor on which BER depends are where P0 and P1 are the probabilities of the symbols, M is the number of samples for the logical 0, and N is the number of samples for the logical 1, average values and standard deviations of the sampled values respectively, and S is the threshold value.If the signal is mixed with the noise, the Average Gaussian method is modified to calculate the average error patterns. The the Average Gaussian method can estimate the BER per bit or per pattern, the Worst-case Gaussian searches for the min BER for each bit or pattern instead of calculating the average values

[Author]

Posts