Alice send the message to Bob. Bob receive the signal. Eve is passive eavesdropping attack. He is also received the signal whenever Alice send to Bob or Bob send to Alice.
Alice received signal: ra = h.v + nb
Bob received signal: rb = h.s +na
h = circularly symmetric gaussian distributed channel response with mean zero
na & nb are circularly symmetric gaussian distributed additive noise with mean zero Alice and Bob respectively
v = coded symbols transmitted by bob
s = coded symbols transmitted by alice
dot '.' = hadmard product
Now, Hadmard product between what we send and what we get that is:
wa = h.v.s + nb.s
wb = h.v.s + na.v
Now, quantization is applied at above equation and BER (Bit Error Rate) is calculated:
BER of Alice-Bob = [ | Summation of (Qa(i) - Qb(i)) | ] / N where i= 1,2,...N and N = total length of Qa
Qa and Qb is quantized sequence of Alice and Bob
Eve Side:
we = he.v.s + ne
he = circularly symmetric gaussian distributed channel response with mean zero
ne = circularly symmetric gaussian distributed additive noise with mean zero
Qe is quantized sequence of Eve
BER of Eve = [ | Summation of (Qa(i) - Qe(i)) | ] / N where i= 1,2,...N and N = total length of Qa
As SNR (signal to noise ratio) increases BER of Alice-Bob decreases. But BER of Eve should be slighlty change. For Example:
SNR (dB) BER of Alice-Bob BER of Eve
5 dB 0.3 0.45
10 dB 0.21 0.44
15 0.15 0.43
20 0.09 0.42
25 0.04 0.41
I get the approx same result of BER of Alice-Eve VS SNR (dB). But I didn't get the approx same result for BER of Eve vs SNR (dB).
My question is How BER of Eve is slightly change? What's the formula? Relation between BER and SNR?