Digital Baseband Signal Receivers - Digital Communication: Principles and System Modelling - page 58

Fig. 3.6 Detection of
sampling instant. ( a ) Signal
output of integrate and dump
filter with input 11010. ( b )
Noise output of integrate and
dump filter

s 0 (t)

VT

τ

(a)

T B

VT

t

−

τ

n 0 (t)

(b)

t

T

The filter output s o (t) is a ramp voltage, as integration of step is obviously a ramp
signal, as shown in Fig. 3.5 . At the end of bit interval (T), the magnitude of s o (t),

i.e. s o ( T )is VT τ VT τ

, wherever applicable, for logic 1 and 0 respectively,
whereas, n o (t) has random values at each bit interval.
Thus, at the end of each bit interval, the sample voltage due to signal is empha-
sized as compared to the sample voltage due to noise. Hence, the end of bit interval
is selected as sampling instant (Fig. 3.6 ).

or −

3.2.1 Noise Power and Variance

From the accuracy point of view, noise power is an important area of concern. If
we consider, the noise to be Additive White Gaussian Noise (AWGN), with power
spectral density (PSD) G n (f), the integrate and dump type filter processes the noise
PSD and result the output PSD G no (f) (Fig. 3.7 ). If we consider the transfer function
of the filter to be H(f), the relation between G n (f) and G no (f) would be

2

G no ( f ) = G n ( f ) | H n ( f ) | (3.5)

n o ( T )

o

Here, rms noise power N 0 = = σ = variance

+
Delay
{H(f)}

2

⎪ H n ( f ) ⎪

G no ( f ) = G n ( f )

G n ( f )

Fig. 3.7 Input and output noise PSD relation