Home Forums GENERAL BER Analyzer Accuracy of Measurement

• Author
Posts
• #23300
alistu
Participant

Hi all,

In order to measure BER of the order -20, for example, one should have a sequence of 10^23 bits at least, so that when one bit is not received correctly, a BER of order 10^-20 is obtained. But this is done in BER analyzer without such a sequence length being used. I would like to know how it is possible and on what basis the BER analyzer works in this regard.

Thank you all in advance.

• #23395
alistu
Participant

Sorry, I have apparently made an obvious mistake in the above post. When I said the sequence should be 10^23 bit at least, I meant 10^20. So that when we have error in one bit, the bit error rate becomes: 1/(10^20) = 10^(-20). I am curious to know the answer and appreciate any comments. Thank you.

• #23442
Damian Marek
Participant

The BER Analyzer calculates the BER in the same manner as in the attachment Heitor gave in this post:

VALUE OF Q IN ENCRYPTED SIGNAL

The Q factor is calculated from the characteristics of the eye diagram, so you do not need extremely large sequence lengths.

• #23460
Ashu verma
Participant

Damian according to this ,there is no need of long sequence length to calculate BER.However in case of OFDm, in almost all the forum topics we say that due to long sequence length value of BER is either o error free and then abruptly decrease fimally to 1.By using BER DUT set, error in bits increase very quickly with even small distance.Is there any relation of ling sewuence length in OFDM ber? Ihave tried to write what i want to know.Hope you can also understand

• #23504
alistu
Participant

Thank you very much Damian for the useful link. You have suggested to me and the others to use BER Test Set in OFDM systems, even though my OFDM system is defined by my own Matlab program. I know that BER Test Set has the advantage of taking pilot signals and such factors into consideration. Other than that Don’t you think the use of BER Analyzer is better since it can give small BER without the need to use long sequences? Thank you.

• #23459
Ashu verma
Participant

Thank you Damian for attaching the link.This link cleared most of doubts.I think calculation of Q-factor also can be done with the difference of digital logic level of 1 and 0 divided by variance in 1 plus variance of 0.After calculating this
BER=1/2 erf(Q/root 2). As given in this tutorial
http://www.srcf.ucam.org/~fw246/docs/thesis/mphil/M__AppendixB_73_.pdf
Attache link by damian is similar to i have attached here but in simple and experiments

• #23462
alistu
Participant

Thank you Sam Sung for the files. The Fiber Optic Communications Educator Kit you have attached contains useful information especially for working in the labs I suppose. I think in my case Heitor’s PFD file is somehow a more concise yet precise answer. It also provides interesting basic info on eye pattern.

• #23482
Ashu verma
Participant

Welcome Alistu,
I have suggested above link because i have worked on this manual in a lab and i found this very helpful for even a beginner. Heitor’s PFD file is very good and give very important points to understand about BER and Q-Factor.

• #23501
alistu
Participant

Yes, thank you. I studied both and learned from both. I was merely suggesting to anyone visiting this topic that if they are looking for a straight answer to the question asked, they can take a look at the paper uploaded by Heitor. But I think the white paper also had a lucid explanation and it was really useful for me.