Está en la página 1de 43

Digital Transmission Through Bandlimited Channels

Ph.D. Ismael Soto

Universidad de Santiago de Chile

June 5, 2023
Santiago, Chile

I. Soto c (U. de Santiago de Chile) Digital Transmission Through Bandlimited Channels 2017 1 / 42
Index

1 Filtros Analógicos y Digitales

2 Filtros FIR

3 Filtros IIR

4 Filtros ARMA (AutoRegresivo y Media en Movimiento)

5 El espectro de potencia de una señal PAM digital

6 Caracterización de canales de banda limitada y distorsión

7 Characterization of Intersymbol Interference

8 Communication System Design for Bandlimited Channels

I. Soto c (U. de Santiago de Chile) Digital Transmission Through Bandlimited Channels 2017 1 / 42
Filtros Analógicos y Digitales

Eligiendo adecuadamente los coeficientes, podemos diseñar filtros selectivos en


frecuencia que dejan pasar señales con componentes de frecuencia en ciertas bandas,
al tiempo que atenúan señales que contengan componentes frecuencia en otras
bandas.
H(w ) actúa como función de conformación espectral para las diferentes frecuencias
de la señal de entrada. En consecuencia, LTI es sinónimo de filtro.
El filtrado se utiliza en procesado digital de señales de diferentes formas: eliminación
de ruido indeseado, conformacion espectral para igualar canales de comunicación,
detección de señales radar, sonar, comunicaciones, etc.

I. Soto c (U. de Santiago de Chile) Digital Transmission Through Bandlimited Channels 2017 2 / 42
Filtros Analógicos y Digitales

Los filtros se clasifican en función de la banda pasante en: Pasa bajos, pasa altos,
pasa banda (pasa-banda) y banda eliminada. Están representados en las figuras
siguientes, en los que se puede apreciar que tienen ganancia unidad en la banda
pasante y cero fuera de la banda.

Figure 1: Pasa Bajo

I. Soto c (U. de Santiago de Chile) Digital Transmission Through Bandlimited Channels 2017 3 / 42
Figure 2: Pasa Alto

I. Soto c (U. de Santiago de Chile) Digital Transmission Through Bandlimited Channels 2017 4 / 42
Figure 3: Pasa Banda

I. Soto c (U. de Santiago de Chile) Digital Transmission Through Bandlimited Channels 2017 5 / 42
Figure 4: Banda Eliminada

I. Soto c (U. de Santiago de Chile) Digital Transmission Through Bandlimited Channels 2017 6 / 42
Figure 5: Pasa Todo

I. Soto c (U. de Santiago de Chile) Digital Transmission Through Bandlimited Channels 2017 7 / 42
Filtros FIR

Un filtro FIR de orden M se describe por la siguiente ecuación diferencia y [n] =


B0 x[n]+B1 x[n −1]+· · ·+BM x[n −M], lo que da lugar a la función de transferencia
H(z) = B0 + B1 z −1 + · · · + BM z −M .
La secuencia {BM } son los coeficientes del filtro.
No hay recursión, es decir, la salida depende sólo de la entrada y no de valores
pasados de la salida.
La respuesta es por tanto una suma ponderada de valores pasados y presentes de
la entrada. De ahi que se denomine Media en Movimiento (Moving Average)
La función de Transferencia tiene un denominador constante y sólo tiene ceros.
La respuesta es de duración finita ya que si la entrada se mantiene en cero durante
M periodos consecutivos, la salida será también cero.

I. Soto c (U. de Santiago de Chile) Digital Transmission Through Bandlimited Channels 2017 8 / 42
Filtros IIR

Los filtros IIR corresponden directamente al equivalente analógico. Una forma de


diseñar filtros IIR es creando la función de transferencia deseada en el dominio
analógico para transformarla al dominio z y después calcular los coeficientes del
filtro IIR mediante los cuales se obtiene la siguiente ecuación en diferencias:

y (n) = b0 x(n) + b1 x(n − 1) + b2 x(n − 2) + . . . + bM x(n − M)


(1)
−a1 y (n − 1) − a2 y (n − 2) − . . . . . . aN y (n − N)
donde las variables ak y bk son los coeficientes del filtro.

Dentro de las ventajas que ofrecen los filtros IIR sobre los tipo FIR encontramos.

I. Soto c (U. de Santiago de Chile) Digital Transmission Through Bandlimited Channels 2017 9 / 42
Prop. Los filtros IIR requieren menos memoria y menos instrucciones para
implementar su función de transferencia.
Un filtro IIR se diseña mediante el calculo de polos y ceros en el plano complejo.
El uso de polos confieren a un filtro IIR la capacidad de implementar funciones de
transferencia que es imposible realizar mediante filtros FIR.

Figure 6: Implementación de filtro IIR


I. Soto c (U. de Santiago de Chile) Digital Transmission Through Bandlimited Channels 2017 10 / 42
Filtros ARMA (Auto Regresivo y Media en Movimiento)

Es el filtro más general y es una combinación de los filtros MA y AR descritos


anteriormente. La ecuación diferencia que descibe un filtro ARMA de orden N es

y [n] + A1 y [n − 1] + A2 y [n − 2] + · · · + AN y [n − N]
(2)
= B0 x[n] + B1 x[n − 1] + · · · + BM x[n − M]
Y la funcion de transferencia
B0 + B1 z −1 + · · · + BM z −M
H(z) = (3)
1 + A1 z −1 + · · · + AN z −N
Un filtro de este tipo se denota por ARMA(N,M), es decir es Autoregresivo de
orden N y Media en Movimiento de orden M.
Su respuesta a impulso es también de duración infinita y por tanto es un filtro del
tipo IIR.

I. Soto c (U. de Santiago de Chile) Digital Transmission Through Bandlimited Channels 2017 11 / 42
Clasificación

EcuacionPDiferencia Tipo de Filtro


y [n] = Bm x[n−m] FIR (Finite Impulse Response),
No Recursivo,
Moving Average (MA) orden M,
Todo ceros
P IIR (Infinite Impulse Response),
Ak y [n − k] = x[n] Recursivo, Autoregresivo (AR) orden N,
P P Todo polos
Ak y [n − k] = k x[n − m] IIR Recursivo, ARMA (N, M),

I. Soto c (U. de Santiago de Chile) Digital Transmission Through Bandlimited Channels 2017 12 / 42
Example Finite Impulse Response (FIR)
M−1
X
yn = hk x(n − k) (4)
n=0

Figure 7: Program Code equation 4

I. Soto c (U. de Santiago de Chile) Digital Transmission Through Bandlimited Channels 2017 13 / 42
Input/output time-domain response.

Figure 8: This shows the input/output time-domain response of the two-tap moving
average filter. Note that the output is delayed by one sample because of the filter’s
transient.

I. Soto c (U. de Santiago de Chile) Digital Transmission Through Bandlimited Channels 2017 14 / 42
Example: Filter Python code with eight taps (part 1)

Figure 9: Example of a filter with eight taps (part 1)

I. Soto c (U. de Santiago de Chile) Digital Transmission Through Bandlimited Channels 2017 15 / 42
Example: Filter Python code with eight taps (part 2)

Figure 10: Example of a filter with eight taps (part 2)

I. Soto c (U. de Santiago de Chile) Digital Transmission Through Bandlimited Channels 2017 16 / 42
Outputs of the filter with eight taps

Figure 11: The top left shows the magnitude response of the eight-tap filter. The top
right shows the corresponding phase response. The bottom plot shows the input/output
time-domain response for the input at frequency ω = π3 .

I. Soto c (U. de Santiago de Chile) Digital Transmission Through Bandlimited Channels 2017 17 / 42
Filter Design Using the Window Method

Figure 12: Python code for filter design using the window method
I. Soto c (U. de Santiago de Chile) Digital Transmission Through Bandlimited Channels 2017 18 / 42
Filter Design Using the Window Method

Figure 13: The top plot shows the filter coefficients. The middle plot shows the
magnitude response in linear units. The bottom plot shows the magnitude response is
decibel units

I. Soto c (U. de Santiago de Chile) Digital Transmission Through Bandlimited Channels 2017 19 / 42
El espectro de potencia de una señal PAM digital
Def. Una señal digital PAM a la entrada de un canal de comunicación se representa
generalmente como:

X
v (t) = an g (t − nT ) (5)
n=−∞

Donde:
an : Secuencia de amplitudes de la fuente
g (t) : Forma de onda de los pulsos
T : Recı́proco de la tasa de sı́mbolos

Prop. Cada elemento de la secuencia an se selecciona de entre uno de los posibles


valores de amplitud, que son:

Am = (2m − M + 1)d, m = 0, 1, ..., M − 1 (6)


Donde:
d : is a scale factor that determines the Euclidean distance between any pair of
signal amplitudes.
I. Soto c (U. de Santiago de Chile) Digital Transmission Through Bandlimited Channels 2017 20 / 42
Def. El espectro de potencia de v(t) es simplemente la transformada de Fourier de
la auto-correlación media Rv (τ ); esto es:

Z ∞
Sv (f ) = Rv (τ )e −j2πf τ dt
−∞ (7)
1 2
= Sa (f ) |G (f )|
T
Donde Sa (f ) es el espectro de potencia de la secuencia de amplitud an y G (f ) es
la transformada de Fourier del pulso g (t). Sa (f ) se define como:

X
Sa (f ) = Ra (m)e −j2πfmT (8)
m=−∞

I. Soto c (U. de Santiago de Chile) Digital Transmission Through Bandlimited Channels 2017 21 / 42
Ejemplo

Determine the power spectrum of V (t) when {an } is an uncorrelated sequence and
gT (t) is the rectangular pulse shown in Figure 14.

Figure 14: Pulso Transmitido

I. Soto c (U. de Santiago de Chile) Digital Transmission Through Bandlimited Channels 2017 22 / 42
Example

Solución:
Z ∞
G (f ) = g (t)e −j2πft dt
−∞
√ sin πfT −jπfT
= T e
πfT
y
 2
sin πfT
Sv (f ) = σa2
πfT

I. Soto c (U. de Santiago de Chile) Digital Transmission Through Bandlimited Channels 2017 23 / 42
Example

Figure 15: Espectro de potencia de la señal transmitida (para σa2 = 1




I. Soto c (U. de Santiago de Chile) Digital Transmission Through Bandlimited Channels 2017 24 / 42
Characterization of Bandlimited Channels and Channel
Distortion

Def. Many communication channels, may be generally characterized as bandlimited


linear filters. Such channels are described by their frequency response C (f ), which
may be expressed as:

C (f ) = A(f )e jθ(f ) (9)


Where A(f ) is called the amplitude response and θ(f ) is called the phase response.
Def. Another characteristic that is sometimes used in place of the phase response

is the envelope delay or group delay, defined as:

1 dτ (f )
τ (f ) = − (10)
2π df

I. Soto c (U. de Santiago de Chile) Digital Transmission Through Bandlimited Channels 2017 25 / 42
Def. A channel is said to be nondistorting or ideal, if, within the bandwith W
occupied by the transmitted signal A(f ) = constant and θ(f ) is a linear function of
frequency.

Prop. On other hand,if A(f ) and τ (f ) are not constant within the bandwidth
occupied by the transmitted signal, the channel distorts the signal. This can be fixed
through an equalizer, which compensates for the linear distortion in the channel.

I. Soto c (U. de Santiago de Chile) Digital Transmission Through Bandlimited Channels 2017 26 / 42
Example: Now are presented examples of a distorted signal by the channel, and
the fixed signal through the equalizer:

Figure 16: Channel Input

I. Soto c (U. de Santiago de Chile) Digital Transmission Through Bandlimited Channels 2017 27 / 42
Figure 17: Channel Output

I. Soto c (U. de Santiago de Chile) Digital Transmission Through Bandlimited Channels 2017 28 / 42
Figure 18: Equalizer Output

I. Soto c (U. de Santiago de Chile) Digital Transmission Through Bandlimited Channels 2017 29 / 42
Characterization of Intersymbol Interference

In a digital communication system, channel distortion causes intersymbol


interference (ISI). We assume that the transmitted signal is a baseband PAM
signalw, expressed as:

X
s(t) = an g (t − nT ) (11)
n=0

The signal s(t) is transmitted over a baseband channel, which may be characterized
by a frquency response C (f ) represented as:

X
r (t) = an h(t − nT ) + w (t) (12)
n=0

Where:
h(t): g (t) ∗ c(t)
c(t): Impulse response of the channel
w (t): Additive noise in the channel

I. Soto c (U. de Santiago de Chile) Digital Transmission Through Bandlimited Channels 2017 30 / 42
To characterize ISI, the received signal is passed through a filter. In general the
optimum filter is matched to the received pulse h(t). Hence, the frequency response
of this filter is H ∗ (f ), output denoted as:

X
y (t) = an x(t − nT ) + v (t) (13)
n=0

if y (t) is sampled at times t=kT, we have:



X
y (kT ) = an x(kT − nT ) + v (kT ) (14)
n=0


X
yk = an xk − n + vk k = 0, 1, ... (15)
n=0

I. Soto c (U. de Santiago de Chile) Digital Transmission Through Bandlimited Channels 2017 31 / 42
The sample values yk can be expressed as:

1 X
yk = x0 (ak + an xk−n ) + vk (16)
x0
n=0n6=k

The term x0 is arbitrary scale factor, set to one for convenience. Then:

X
yk = ak + an xk−n + vk (17)
n=0n6=k

I. Soto c (U. de Santiago de Chile) Digital Transmission Through Bandlimited Channels 2017 32 / 42
Figure 19: Examples of eye patterns for binary and quaternary amplitude shift keying (or
PAM)

I. Soto c (U. de Santiago de Chile) Digital Transmission Through Bandlimited Channels 2017 33 / 42
Figure 20: Effect of ISI on eye opening

I. Soto c (U. de Santiago de Chile) Digital Transmission Through Bandlimited Channels 2017 34 / 42
Design for Bandlimited Channels

Class: In this section we consider the design of the transmitter and receiver filters
that are suitable for a baseband bandlimited channel. Two cases considered:
First Case: Design based on a transmitter and receiver filters that result in zero
ISI
Second Case:The design is based on transmitter and receiver filters that have a
specified amount of ISI.

I. Soto c (U. de Santiago de Chile) Digital Transmission Through Bandlimited Channels 2017 35 / 42
Signal Design for Zero ISI
The design of bandlimited signals with zero ISI was a problem considered by Nyquist
about 70 years ago. He demonstrated that a necessary and sufficient condition for
a signal x(t) to have zero ISI is:
(
1 if n = 0
x(nT ) = (18)
0 if x 6= 0

is that its Fourier transform X(f) satisfy:



X m
X (f + )=T (19)
m=−∞
T

One of the most commonly used signals in practice has a raised-cosine frequency
response characteristic, defined as:

T
 if 0 ≤ |f | ≤ 1−α2T
Xrc (f ) T2 [1 + cos πT
α (|f |) − 1−α
2T ] if 1−α
2T ≤ |f | ≤ 1+α
2T
(20)
1+α

0 if |f | > 2T

I. Soto c (U. de Santiago de Chile) Digital Transmission Through Bandlimited Channels 2017 36 / 42
The signal pulse xrc (t) having the raised-cosine spectrum is:

sin πt παt
T cos T
xrc (t) = πt 4α2 t 2
(21)
T 1− T2

I. Soto c (U. de Santiago de Chile) Digital Transmission Through Bandlimited Channels 2017 37 / 42
Figure 21: Raised-cosine frequency response and corresponding pulse shape (a)
Raised-cosine frequency responde (b) Pulse shapes for raised-cosine frequency response

I. Soto c (U. de Santiago de Chile) Digital Transmission Through Bandlimited Channels 2017 38 / 42
Signal Design for Controlled ISI

A transmit filter with excess bandwith may be employed to realize practical


transmitting and receiving filters for bandlimited channels. By allowing for a
controlled amount of ISI, we can achieve the rate of 2W symbols/second, as shown
in the next images:

I. Soto c (U. de Santiago de Chile) Digital Transmission Through Bandlimited Channels 2017 39 / 42
Figure 22: Impulse response and frequency response of tryncated discrete-time FIR filter
at transmitter

I. Soto c (U. de Santiago de Chile) Digital Transmission Through Bandlimited Channels 2017 40 / 42
Figure 23: Impulse response of the cascade of the transmitter filter with the matched
filter at the receiver

I. Soto c (U. de Santiago de Chile) Digital Transmission Through Bandlimited Channels 2017 41 / 42
END

Thanks for your attention

I. Soto c (U. de Santiago de Chile) Digital Transmission Through Bandlimited Channels 2017 42 / 42

También podría gustarte