I experimentally transmit an OFDM signal through 2 meters channel. When I estimate the channel at the receiving end, I find the channel coefficients very small. An example of the channel I find is:

h = [0.0823 + 0.0167i 0.0017 + 0.0033i 0.0010 + 0.0023i 0.0008 + 0.0018i 0.0004 + 0.0011i 0.0001 + 0.0001i -0.0004 + 0.0010i -0.0002 + 0.0006i];

In idea case, I supposed to get the first value 1 and all others zeros, but I do not know why that coefficients becomes very small. does that mean the signal is losing its power?

When I directly decode the signal without channel equalization, I get very small value of my signal. Is there a common reason or explanation for that signal deterioration?

NP The transmitted signal is optical in free space transmission. I am using LED as light source with 2.25 Voltage.

  • $\begingroup$ Please edit your question to describe what your channel is and post a picture or a sketch. Include not just the medium that's 2m long, but the receiving and transmitting apparatus. Is it two feet of coax, two feet of string, free space transmission with two antennas separated 2m apart, etc. If it is antennas, are they simple dipoles, 1/4-wave with ground plane, Yagis, etc. Also, if it's free space transmission, tell us the frequency. $\endgroup$
    – TimWescott
    Commented Oct 2, 2022 at 15:24
  • 1
    $\begingroup$ Floating point numbers are often… floating. I.e. the absolute magnitude of the number lacks a physical reference and it is up to the application to normalize if neded. $\endgroup$
    – Knut Inge
    Commented Oct 2, 2022 at 15:52
  • $\begingroup$ And without detail on the channel, there's no way of knowing what's reasonable. $\endgroup$
    – TimWescott
    Commented Oct 2, 2022 at 16:01
  • $\begingroup$ @TimWescott I have updated the question. $\endgroup$
    – Sajjad
    Commented Oct 2, 2022 at 17:31

1 Answer 1


First, I don't see those numbers as all that small. Second, given that there's computers, transmit electronics, and receive electronics in the mix, the raw channel gain could be anything.

It sounds like your path can be described as: computer #1 $\to$ some electronics $\to$ LED $\to$ free space $\to$ photodiode $\to$ some electronics $\to$ computer #2. (And it may be just one computer without loss of generality).

How much the signal changes numerically is going to be an arbitrary combination of the scaling in the two computers, the gains of the transmit and receive electronics, and the actual path gain between the LED and the photodiode. More importantly, that's not what matters.

In a communications system like this, what matters is the signal to noise ratio at the receiver, and any distortion -- as long as the SNR is high enough for your chosen communications protocol, and as long as the signal isn't so strong that nonlinearities are distorting the signal, then you're fine.

  • $\begingroup$ Thank you, the problem is that I get clipped time-domain signal. However when I was performing the same test two months ago, all was ok. I do not know what is the problem in that ! $\endgroup$
    – Sajjad
    Commented Oct 4, 2022 at 6:41
  • $\begingroup$ I posted that question here dsp.stackexchange.com/questions/84788/… $\endgroup$
    – Sajjad
    Commented Oct 4, 2022 at 7:59

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.