Root Raised Cosine (RRC) Filters and Pulse Shaping in Communication Systems

Erkin Cubukcu
Abstract
This presentation briefly discusses application of the Root Raised Cosine (RRC) pulse shaping in the
space telecommunication. Use of the RRC filtering (i.e., pulse shaping) is adopted in commercial
communications, such as cellular technology, and used extensively. However, its use in space
communication is still relatively new. This will possibly change as the crowding of the frequency
spectrum used in the space communication becomes a problem. The two conflicting requirements in
telecommunication are the demand for high data rates per channel (or user) and need for more
channels, i.e., more users. Theoretically as the channel bandwidth is increased to provide higher data
rates the number of channels allocated in a fixed spectrum must be reduced. Tackling these two
conflicting requirements at the same time led to the development of the RRC filters. More channels with
wider bandwidth might be tightly packed in the frequency spectrum achieving the desired goals. A link
model with the RRC filters has been developed and simulated. Using 90% power Bandwidth (BW)
measurement definition showed that the RRC filtering might improve spectrum efficiency by more than
75%. Furthermore using the matching RRC filters both in the transmitter and receiver provides the
improved Bit Error Rate (BER) performance.
In this presentation the theory of three related concepts, namely pulse shaping, Inter Symbol
Interference (ISI), and Bandwidth (BW) will be touched upon. Additionally the concept of the RRC
filtering and some facts about the RRC filters will be presented.

Avionic Systems Analysis

Root Raised Cosine Filters

&
Pulse Shaping in Communication Systems

Erkin Cubukcu
Friday, May 18, 2012

Outline

Avionic Systems Analysis

Pulse Shaping, Intersymbol Interference (ISI),

and Bandwidth
Ideal Low Pass Filter (LPF)
Raised Cosine (RC)
Root Raised Cosine (RRC)
Facts about RRC
Link Modeling with the RRC Filters
Signal Spectra
BER Plots
Conclusions

5/7/2012

Pulse Shaping, Intersymbol Interference

(ISI), and Bandwidth

Avionic Systems Analysis

Digital Modulation

5/7/2012

Pulse Shaping, Intersymbol Interference

(ISI), and Bandwidth (cont)

Avionic Systems Analysis

Theory

Two conflicting requirements!:

Demand for High data rates (more information).
Need for Narrow Bandwidth (more users, more channels, less noise).
If channels are too narrow the symbols will be too wide, hence. At sampling
points (in time) there will be signal (tails) of the previous and next symbols.
This is called Inter-symbol Interference (ISI)
One possible solution is to use an Ideal Low Pass Filter (ILPF) (rectangular
in frequency).

5/7/2012

Ideal Low Pass Filter (ILPF)

Avionic Systems Analysis

No ISI
Narrow bandwidth channel

sin t / T
t / T

Issues
Physically unrealizable and difficult to approximate.
If attainable, require extreme precise synchronization, synchronization
is a problem (jitter in the system might be detrimental).

5/7/2012

Solution : Raised Cosine (RC) Filters

It is shown by Nyquist that

Avionic Systems Analysis

If the frequency characteristic has odd symmetry at the cutoff frequency,

the impulse response will have zeros at uniformly spaced intervals.

Much simpler to attain

Effects of jitter may be minimized

T
T
1
H ( f ) 1 cos
f
2T

2

h (t )

5/7/2012

sin t / T cos t / T
t / T 1 ( 4 2 t 2 / T 2 )

0 f

T
2

1
1
f

T
T
2
2

(Eq. 1)

2T

(Eq. 2)

Plot of Raised Cosine (RC) filter

Avionic Systems Analysis

Frequency Response

Impulse Response

5/7/2012

Raised Cosine Filter (Cont)

Avionic Systems Analysis

Impulse response now has a sinc term that ensures that it has zero
crossings as like ideal low pass filter.
sin t / T
t / T

In addition, it has another term

cos t / T
1 ( 4 2t 2 / T 2 )

That decays in time hence reduces the tails reducing the impact of
jitter.

5/7/2012

Bandwidth of Raised Cosine (RC) Filter

sin t / T
t / T

Avionic Systems Analysis

c t t / T

Where the sampling time is T

c T T / T

cT

T / c /( 2f c ) 1 / 2 f c

If ideal LPF were used the baseband bandwidth would be

B f c 1 / 2T

Since

Nyquist Bandwidth

H( f ) 0 f

2T

Bandwidth for a realizable RC filter

B (1 ) / 2T

Nyquist Bandwidth times

So the baseband transmission bandwidth

B (1 ) f c
5/7/2012

Avionic Systems Analysis

Root Raised Cosine (RRC) Filter

The overall channel transfer function must be Raised Cosine (RC) as
discussed above.
One way of achieving it is to take square root of the raised cosine filter in
frequency domain and use this new filter in the Tx and Rx. This is the so
called Root Raised Cosine filter.
H rrc ( )

H ( ) rc

1
(1 cos / 2 c )
2

2 c

When the transmitter and receiver filters are cascaded one gets raised
cosine filter transfer characteristic
H rc ( ) H rrc ,tx ( ) H rrc , rx ( )

Or
H rc ( )
5/7/2012

H rc ( ) H rc ( )
10

Root Raised Cosine Filter (RRC) (cont)

Avionic Systems Analysis

By taking square root of RC filter frequency response, one gets.

H( f )
2

2T

T
1
1
cos
f

2T

1
1
f

2
2
T
T

2
T

Finding its impulse Response is a little bit tricky.

h (t )

2
T

sin[(1 )t / T ]
4t / T
1 (4t / T ) 2

cos[(1 )t / T ]

Impulse response can also be generated numerically using IFFT.

5/7/2012

11

Avionic Systems Analysis

Facts about RRC

RRC theoretically has infinite number of taps so it has infinite attenuation in the stop
band. However, in implementation its length should be reduced to a finite value.

Decreasing the number of samples (filter delay) reduces the stop band attenuation.

The rolloff factor is a measure of the excess bandwidth of the filter, i.e., the bandwidth
occupied beyond the Nyquist bandwidth of 1/2T, where 1/T is symbol rate.

As rolloff increases eye in the eye diagram opens up. This means that if there were
no bandwidth restrictions it would be easier on the receiver if one used a large rolloff.
(However, for bandwidth efficiency rolloff should be smaller.)

Smaller rolloff gives narrower bandwidth. However, its side lobes increases so
attenuation in stop band is reduced.

RRC filters are implemented in the base band as a digital filter. Since implementing
narrow (high Q) filters in the RF bands is difficult.

5/7/2012

12

Avionic Systems Analysis

Facts about RRC (cont.)

Another issue to consider in pulse shaping is the Peak-to-Average Power Ratio

(PAPR).

PAPR is determined by combination of

Modulation
Constellation of the signal
Pulse shaping

High PAPR reduces power amplifier efficiency since it must operate with large back
off (higher PAPR requires higher back off and/or more linearity)

PAPR of an RRC will increase with

reduced excess bandwidth5
increased filter length5

5/7/2012

13

Link Modeling with the RRC Filters

Avionic Systems Analysis

A SIMULINK Model was developed and simulated to study the spectral

efficiency and Bit Error Rate (BER) performance of the RRC filters. Below
is a high level block diagram of this model.

5/7/2012

14

Avionic Systems Analysis

Link Modeling with the RRC Filters (Cont.)

Tx Model with RRC Pulse Shaping

The modulator generates one symbol for each pair of data bits. The
symbols generated by the modulator is up sampled and pulse shaped
(filtered) to comply with the channel bandwidth restrictions. Typically, the
pulse shaping is the last stage of transmitter before (DAC and) PA.

5/7/2012

15

Avionic Systems Analysis

Link Modeling with the RRC Filters (Cont.)

Rx Model with RRC Filter

RRC filter filters out the signal (i.e., equalizes to give a nearly zero
Intersymbol Interference, ISI, if Tx is also using RRC filter pulse shaping).
The output is sampled at the optimum points (i.e., down-sampled) to give
one sample per constellation symbol. Demodulator finds out which
quadrant the received sample falls and based on that decision generates
a pair of bits.

5/7/2012

16

RC Filter Impulse Response, rolloff 0.4, 0.7, and 1

Symbol rate 1 Msps

-3

14

Raised Cosine Filter Response

x 10

RC, rolloff 0.4

RC, rolloff 0.7
RC, rolloff 1

Avionic Systems Analysis

12
10

Magnitude

8
6

4
2

0
-2
-4
-4

5/7/2012

-3

-2

-1

0
time (s)

17

Ideal RRC Frequency Response

with rolloff 0.4, 0.7, and 1.0
Just the plot of the equation

Symbol rate 1 Mbps

Root Raised Filter Response
Ideal RRC, rolloff 0.4
Ideal RRC, rolloff 0.7
Ideal RRC, rolloff 1

-10
-20
-30
-40
Mag. (dB)

Avionic Systems Analysis

-50
-60
-70
-80
-90
-100

5/7/2012

0.5

1.5

2
Freq. (Hz)

2.5

3.5

4
6

x 10

18

Frequency Response of Realizable Root Raised

Cosine Filter with Rolloff 0.4, 0.7, and 1
Symbol rate 1 Mbps

Root Raised Filter Response

0
rolloff 0.4
rolloff 0.7
rolloff 1

Avionic Systems Analysis

-10
-20
-30

Mag. (dB)

-40
-50
-60
-70
-80
-90
-100

5/7/2012

0.5

1.5

2
Freq. (Hz)

2.5

3.5

4
6

x 10

19

Effect of the length of the Filter (Realizable)

delay 2, 4, and 8 sym (rolloff 0.4)
Delay is half of the filter length in time. Here is given as the symbol length
Root Raised Filter Response
0
del 2 sym
del 4 sym
del 8 sym

Avionic Systems Analysis

-10
-20
-30

Mag. (dB)

-40
-50
-60
-70
-80
-90
-100

5/7/2012

0.5

1.5

2
Freq. (Hz)

2.5

3.5

4
6

x 10

20

OQPSK Tx RRC Output Spectrum for a Symbol Rate of 6 Msps

w/ Category A Masks
OQPSK Spectrum at the RRC filter o/p, Sym Rate 6 Msps, OSR 16
0

-10

Tx o/p
Data Rate > 2 Msps
Data Rate < 2 Msps

rolloff is 0.35
Group Delay is

4 sym

power (dB)

Avionic Systems Analysis

-20

-30

-40

-50

-60

-70

5/7/2012

-2

-1.5

-1

-0.5

0
0.5
freq (Hz)

1.5

2
7

x 10

Rolloff 0.35, Group delay 4, 2-sided BW = 3*1.35 = 4 MHz.

Measured 99% 2-sided BW = 3.48 MHz
Measured 90% 2-sided BW = 2.78 MHz

21

BPSK Tx RRC Output Spectrum for a Sym Rate of 2 Msps

w/ Category A Masks
BPSK Spectrum at the RRC filter o/p, Sym Rate 2 Msps, OSR 16
0
Tx o/p
Data Rate > 2 Msps
Data Rate < 2 Msps

rolloff is 0.35

-10

Group Delay is

4 sym

power (dB)

Avionic Systems Analysis

-20

-30

-40

-50

-60

-70

-1.5

-1

-0.5

0
freq (Hz)

0.5

1.5
7

x 10

Group delay 4 sym, rolloff 0.35

2-sided BW = 2*1.35 = 2.7 MHz.
99% measured 2-sided BW = 2.32 MHz.
90% measured 2-sided BW = 1.87 MHz.

5/7/2012

22

Power Amplifier (PA) Output Spectrum

for 6 Msps OQPSK
Mod and Tx o/p Spectra, OSR 16
0
Rolloff is 0.3.
Group del is 4 symb.
-20

The spectrum of
the Modulator
output is also
shown in the plot.

-30
-40
power (dB)

Avionic Systems Analysis

-10

Mod o/p
Tx o/p
Data Rate > 2 Msps
Data Rate < 2 Msps
NTIA Spurious Emissions

-50
-60
-70
-80
-90
-100

5/7/2012

-2

-1.5

-1

-0.5

0
freq (Hz)

0.5

1.5

2
7

x 10

23

OQPSK Tx Output Spectrum for a Rect Pulse at a

Sym Rate of 6 Msps w/ Category A Masks
OQPSK Spectrum at the Tx o/p (Rect Pulse), Sym Rate 6 Msps, OSR 16
0
Tx o/p
Data Rate > 2 Msps
Data Rate < 2 Msps

Avionic Systems Analysis

-10

power (dB)

-20

-30

-40

-50

-60

-70

-2

-1.5

-1

-0.5

0
0.5
freq (Hz)

1.5

2
7

x 10

Rectangular pulse shaping does not comply with Category A spectral mask requirements.
90% measured 2-sided BW = 4.96 MHz.

5/7/2012

24

BPSK Tx Output Spectrum for a Rectangular Pulse,

for a Symbol Rate of 2 Msps w/ Category A Masks
BPSK Spectrum at the Tx o/p (Rect Pulse), Sym Rate 2 Msps, OSR 16
0
Tx o/p
Data Rate > 2 Msps
Data Rate < 2 Msps

Avionic Systems Analysis

-10

power (dB)

-20

-30

-40

-50

-60

-70

-1.5

-1

-0.5

0
freq (Hz)

0.5

1.5
7

x 10

Rectangular pulse shaping does not comply with Category A spectral mask requirements 6.
99% measured 2-sided BW = 2.22 MHz.
90% measured 2-sided BW = 3.34 MHz.

5/7/2012

25

OQPSK, 6 Msps
OQPSK, 6 Msps, Extrapolated (Cubic Spline)

10

Theory
Tx RRC & Rx RRC, Extr
Tx RRC & Rx No filt, Extr
Tx RRC & Rx I&D, Extr
Tx Rect & Rx No Filt, Extr
Tx Rect & Rx I&D, Extr
Tx Gauss (BT=0.35) & Rx I&D, Extr
Tx Gauss (BT=0.85) & Rx I&D, Extr

-1

-2

10

-3

10

BER

Avionic Systems Analysis

10

-4

10

Please Note that

the curves are
(Cubic Spline)
extrapolated after
10-11 dB EbNo
and
Simulations are
ran for +1000
error bits unless
stated otherwise.

-5

10

-6

10

-7

10

-8

10

-2

5/7/2012

-1

6
7
Eb/No

10

11

12

13

14

15

26

BPSK, 2 Msps
BPSK, 2 Msps, Extrapolated (Cubic Spline)

10

Theory
w/Tx RRC & Rx RRC,Extr
w/Tx RRC & Rx No Filt,Extr
w/Tx RRC & Rx I&D,Extr
w/Tx Rect & Rx No Filt,Extr
w/Tx Rect & Rx I&D,Extr
w/Tx Gauss & Rx I&D,Extr

-1

-2

10

-3

10

BER

Avionic Systems Analysis

10

-4

10

*Please Note that

the curves are
(Cubic Spline)
extrapolated after
10-11 dB EbNo
and
Simulations are
ran for +1000
error bits unless
stated otherwise.

-5

10

-6

10

-7

10

-8

10

-2

5/7/2012

-1

6
7
Eb/No

10

11

12

13

14

15

27

Conclusions

Avionic Systems Analysis

RRC is more bandwidth efficient than NRZ (i.e., no pulse shaping,

rectangular waveform).
Matching TX and RX filters achieves optimum Bit Error Rate (BER)
performance.
This study provides insight and guidance in the system design for
spectral efficiency and Bit Error Rate (BER) performance.

5/7/2012

28

References
1.

Avionic Systems Analysis

2.
3.
4.

5.
6.

S. Daumont, R. Basel, Y. Lout, "Root-Raised Cosine filter influences on PAPR

distribution of single carrier signals", ISCCSP 2008, Malta, 12-14 March 2008.
Proakis, J. (1995). Digital Communications (3rd ed.). McGraw-Hill Inc. ISBN 0-07113814-5.
Glover, I.; Grant, P. (2004). Digital Communications (2nd ed.). Pearson Education
Ltd. ISBN 0-13-089399-4.
Tavares, L.M.; Tavares G.N. (1998) Comments on "Performance of Asynchronous
Band-Limited DS/SSMA Systems" . IEICE Trans. Communications., Vol. E81-B, No.
9
Chtelain, Benot, and Gagnon, Franois, Peak-to-Average Power Ratio and
Intersymbol Interference Reduction by Nyquist Pulse Optimization
Constellation Program Command, Control, Communication, and Information (C3I)
Interoperability Standards Book Volume 2: Spectrum and Channel Plan
3.4..2.1 Spectral Emissions Mask for Spurious Emissions (NTIA), p 42
3.4.2.2 Spectral Emissions Mask for Category A Missions, p 43

5/7/2012

29