Show that the capacity of a Gaussian channel with NRZ/OOK modulation rapidly

converges with increasing signal power to the upper limit [C = 1 (bit/s) Hz] by plotting
the capacity of the channel in matlab.

Clues:
 Assume a binary input signal xi ∈ X = {x1, x2} taking the values x1 = A and x2 = −A,
and a continuous output y ∈ Y

 Define the mutual information as:

 Define the conditional probability p(y | xi) as:

 Show that the channel capacity is a function of the “bit SNR” parameter S2=a2 /σ 2 and
numerically determine the limit for increasing S2.

Assume that the binary-input AWGN channel with inputs ±A and noise variance σ 2 . The
transition probability density function for this channel is defined by:
1|Page
 − ( y−x i )2
p ( y∨x i )=
1
σ √2 π
exp [2 σ2 ]
where x = ± A. By symmetry, the capacity of this channel is achieved by symmetric input
probability mass function, i.e., by letting P [x = A] = P [x = —A] = 0.5. Using these input
probabilities and:
H ( X ; Y )=∑ ∫ p ( x i , y ) log ¿ ¿ ¿ ¿
i Y

¿ ∑ ∫ p ( x i) p ( y∨x i ) log ¿ ¿ ¿ ¿
i Y
But
2
− ( y−x i )
[2 e ]
[[ ]
2

p ( y∨x i )
∞
[ − ( y−x i )
] log 2σ 2

∫ p ( y∨x i ) log
Y
[ p( y) ] dy = ∫ σ √12 π e
−∞
2σ
2

e
− ( y− xi )
2σ
2
2

+e
− ( y+ xi )
2σ
2
2

]
dy

2
∞
1 [ − ( y−x i )
2σ
2 ] log 2
= ∫
−∞ σ √2 π
e
( 1+e
−2 y x i
σ
2
) dy

2
− ( u−z )
x y
∞
Define ¿ i ,u= , and g ( z ) = ∫ 1 e
[ ] log 2
σ σ −∞ √ 2 π
2
( 1+e−2 uz
du
)
A
Hence, H ( X ; Y )=0.5 g [ σ σ )]
( )+g(
−A

So what is left is to plot H(X;Y ) and generate a similar figure to that provided

 Your plot should look like this:

2|Page