An Expression for BER

with Gaussian Noise

 The Q-function
fv(v)
Suppose that v is Gaussian with m = 0
and = 1. P[vT]=1-Q(T) Q(T) = P[v>T]

The probability that v is greater than a

particular value T is given by the Q-
function
Q(T) P[v T] m = 0, = 1 v
T
1
There is no closed-form expression for 0.9
Q(T). Its value must be found 0.8
numerically, e.g. 0.7
from tables, or 0.6
0.5
the MATLAB function qfunc(T)
0.4

Properties
0.3 1-Q(T) P[v>T]
0.2 P[vT]
Q(T)
0.1
P[v T] 1 Q(T)
0
-3 -2 -1 0 1 2 3
Q(0) 12 T
 Probabilities for Other Gaussians
If y is Gaussian with If y is Gaussian with In general,
my 0 and y = 1, my = 0 and y 1, if y is Gaussian with
my 0 and y 1,
P[y > T] = Q T - my T
P[y > T] = Q

T - my
mv 0 mv 0 P[y > T] = Q
0.4 0.4
fv fy my 2 fv my 0 y
0.2
v 1 0.2 v 1
y 1 fy y 2
0 y 0 y
-5 0 5 -5 0 5

1 1
Q(T-2) Q(T)
Q(T)
0.5 0.5
Q(T/2)

0 T 0 T
-5 0 5 -5 0 5
 Expressions for Pe0 and Pe1
If IN = 0, there is an error if f(y if IN=0)
OUT = 1
The noise pushes y above T Pe0

Pe0 P[y T if IN 0]

T rmin
Q rmin T rmax y

If IN = 1, there is an error if
OUT = 0 f(y if IN=1)
The noise pushes y below T
Pe1 P[y T if IN 1] Pe1

T rmax
1 Q
rmin rmax y
T
 Predicting BER
If 0 and 1 input bits are equally likely

BER Pe0 Pe1 Q

1
2
1
2
1
2 T rmin
1 Q
1
2 T rmax

f(y if IN=0) f(y if IN=1)

Pe0
Pe1

rmin T rmax y rmin T rmax y

f(y if IN=0) f(y if IN=1)

BER

rmin T rmax y