find the dependence ratio

hi, i have a graph above. Now i want to comment in this graph. For example, if the channels increase 4 times, the OSNR decrease 'x' times. What function should i use to find 'x'?

Answer 1

[Salaheddin Hosseinzadeh]

Dear Son,

You can make a function if you have a series of measurements.

You can compare results with a base signal, and then find a equation for it using cftool

Lets say channel(ch) = 2 then OSNR reduced 1

ch = 3, ONSR = 2

ch = 4, OSNR = 3

and so on

What you need to do is then go to the cftool

cftool

then assign the x and y data

x is the ch

y is the OSNR

Find a fit for it, whichever method you prefer.

Then you have an equation or relation between ch and OSNR which allows you to predict and tell the dependency of CH and OSNR.

Hope this helps

Good Luck!

[son]

many thanks. that is what i want.

[son]

hello could u please explain the result:

Linear model Poly2:

     f(x) = p1*x^2 + p2*x + p3
  Coefficients (with 95% confidence bounds):
         p1 =     -0.1474  (-0.3859, 0.09112)
         p2 =        5.23  (0.5256, 9.934)
         p3 =       5.431  (-11.6, 22.47)

Goodness of fit:
  SSE: 0.3463
  R-square: 0.9994
  Adjusted R-square: 0.9983
  RMSE: 0.5884

[Salaheddin Hosseinzadeh]

That's good Son,

So you managed to find a polynomial fit with max degree of 2

That's great, congradulations.

Now you have the equation which is

f(X) = (-14.74)X^2 + 5.23X + 5.431

Where x is I guess number of channels

Answer 2

[Salaheddin Hosseinzadeh]

Hi Son,

I guess you want to compare between your signals to see how close the are! At least that what I understood from your unclear question.

To see this you need to use Xcorr I believe to see the correlation between your signals.

To find the corss correlation in MATLAB you can either use crosscorr function or xcorr

doc crosscorr
doc xcorr

type those above in MATLAB to see the matlab documentation about them, they are straightforward to use.

If this is not what you're looking for just let us know.

Good Luck.

[son]

as u can see, the graph for 3 subplot is the the same curve but it decrease by somewhat value.

There are three case:number of channels is 10, 40,80 for each case => i have three graphs

now i want to comment something like: if the number of channels increase 4 times, the OSNR decrease ...

[Salaheddin Hosseinzadeh]

Then what is the problem?

Find the max or mean of one graph and compare it to the min or max of the other graph!

By comparing I mean either division or subtraction, depending on what's your scale? linear or logarithmic. Which yours is in dB.

[son]

yeah, but i want to find a function to show this dependence. is it possible?

if i know that function, i can find the OSNR decrease for the number of channels increase 2.5 WITHOUT plot a graph.