Vous êtes sur la page 1sur 51

1-1

Part II : Wireless Communication

Chapter 1

Modulation of Digital Data


This chapter contains some slides from chapter 5 of Data Communications and networking, 3rd ed. by B. Forouzan. The borrowed slides contain the copyright notice of McGraw-Hill at the bottom of the slide. Most of these slides are either modified or supplemented by additional information. McGraw-Hill The McGraw-Hill Companies, Inc., 2004

1-2

In order to get a digital signal into the air (through an antenna) it has to be modulated first. The modulation assumes a high frequency (radio frequency) sinusoidal oscillation, called carrier, into which the data signal is somehow impressed before being sent to antenna. Antenna transforms the electric oscillations into electromagnetic waves (radio waves). Radio waves propagate from transmitting antenna to all receiving antennas, where they induce electrical signals equal (although attenuated and delayed) to the signal in transmitting antenna. Each receiver then needs to take the impress off the carrier and thus recover the original data signal. This process reversed to modulation is called demodulation.

Data signal d(t)

Signal in antenna s(t)

Carrier c(t) = sin(2pfct )


McGraw-Hill The McGraw-Hill Companies, Inc., 2004

1-3

Frequencies used in telecommunications

Taken from: Wireless Communications and Networks, by W. Stallings, Prentica Hall 2002

McGraw-Hill

The McGraw-Hill Companies, Inc., 2004

1-4

A carrier signal has three parameters which can be used for impressing:

c(t) = A sin(2 fct + )


Amplitude Frequency Phase

ASK = Amplitude Shift Keying FSK = Frequency Shift Keying PSK = Phase Shift Keying QAM = Quadrature Amplitude Modulation (combination of ASK and PSK)
McGraw-Hill The McGraw-Hill Companies, Inc., 2004

1-5

Bit rate and baud rate


Bit rate: number of bits per second Baud rate: number of signal units per second If a signal unit is composed of n bits, then the bit rate is n times higher than baud rate

Question: An analog signal carries 4 bits in each signal unit. If 1000 signal units are sent per second, find the baud rate and the bit rate, Answer: baud rate = 1000 bauds/sec bit rate: 1000*4 = 4000 bps

McGraw-Hill

The McGraw-Hill Companies, Inc., 2004

1-6

ASK
s(t) = (Ao+ A d(t)) sin(2 fc t)

McGraw-Hill

The McGraw-Hill Companies, Inc., 2004

1-7

Spectrum of ASK

ASK is not noise immune


NOTICE: The bandwidth shown above is the minimal required bandwidth for
the ASK. In practice the required bandwidth is much larger. These issues will be discussed in slide Spectral Analysis of Digital Signals. (Similar remark holds for FSK and PSK considered in the following slides.)
McGraw-Hill The McGraw-Hill Companies, Inc., 2004

1-8

FSK
s(t) = Asin(2 (fc + f d(t))t)

McGraw-Hill

The McGraw-Hill Companies, Inc., 2004

1-9

Spectrum of FSK

FSK is noise immune but requires large bandwidth

McGraw-Hill

The McGraw-Hill Companies, Inc., 2004

1-10

PSK
s(t) = Asin(2 fct+ d(t)) = = Asin(2 fct+ d(t)) = = AdNRZ(t)sin(2 fct)

McGraw-Hill

The McGraw-Hill Companies, Inc., 2004

1-11

Spectrum of PSK
This is the minimal bandwidth (the Nyquist bandwidth). In practical applications is used BW = 2 Nbaud (see later)

+ PSK is noise immune and requires small bandwidth PSK with two phases is called 2-level PSK (2-PSK), or binary PSK (B-PSK)
McGraw-Hill The McGraw-Hill Companies, Inc., 2004

1-12

Constellation of two-level PSK

McGraw-Hill

The McGraw-Hill Companies, Inc., 2004

1-13

Four-level PSK
Excellent performance of 2-PSK encourages us to go with 4-PSK, also called quadrature PSK (Q-PSK)

90o
McGraw-Hill

180o

180o

270o

0o

The McGraw-Hill Companies, Inc., 2004

1-14

Constellation of 4-PSK

4-PSK has more efficient usage of bandwidth than 2-PSK, because each signal unit has two bits. For the same bandwidth, the data bit rate doubles.
McGraw-Hill The McGraw-Hill Companies, Inc., 2004

1-15

Example
Find the bandwidth for a 4-PSK signal transmitting at 2000 bps. Transmission is in half-duplex mode.

For 4-PSK the baud rate is one half of the bit rate, 1000 bauds per second. The bandwidth for any-level PSK is equal to the baud rate, therefore, the bandwidth for 2000 bps 4-PSK is 1000 Hz.
To be revised (this slide is using Nyquist BW)
McGraw-Hill The McGraw-Hill Companies, Inc., 2004

1-16

8-PSK

The idea can be extended to 8-PSK, 16-PSK, 32-PSK,. The limitation is the ability of equipment to distinguish small differences in signals phase.
McGraw-Hill The McGraw-Hill Companies, Inc., 2004

1-17

Example
Given a bandwidth of 5000 Hz for an 8-PSK signal, what are the baud rate and bit rate?

For PSK the baud rate is the same as the bandwidth, which means the baud rate is 5000. But in 8-PSK the bit rate is 3 times the baud rate, so the bit rate is 15,000 bps.
To be revised (this slide is using Nyquist BW)
McGraw-Hill The McGraw-Hill Companies, Inc., 2004

1-18

Quadrature Amplitude Modulation (QAM)


Combination of ASK and PSK which helps making a contrast between signal units. The number of amplitude shifts should be lower than the number of phase shifts due to noise susceptibility of ASK.

McGraw-Hill

The McGraw-Hill Companies, Inc., 2004

1-19

Example of 8-QAM signal

McGraw-Hill

The McGraw-Hill Companies, Inc., 2004

1-20

Bit and baud rate comparison


Modulation ASK, FSK, 2-PSK 4-PSK, 4-QAM 8-PSK, 8-QAM 16-QAM 32-QAM 64-QAM 128-QAM 256-QAM
McGraw-Hill

Units Bit Dibit Tribit Quadbit Pentabit Hexabit Septabit Octabit

Bits/Baud 1 2 3 4 5 6 7 8

Baud rate N N N N N N N N

Bit Rate N 2N 3N 4N 5N 6N 7N 8N

The McGraw-Hill Companies, Inc., 2004

1-21

Example
A constellation diagram consists of eight equally spaced points on a circle. If the bit rate is 4800 bps, what is the baud rate?

The constellation indicates 8-PSK with the points 45 degrees apart. Since 23 = 8, 3 bits are transmitted with each signal unit. Therefore, the baud rate is 4800 / 3 = 1600 baud

McGraw-Hill

The McGraw-Hill Companies, Inc., 2004

1-22

Example
Compute the bit rate for a 1000-baud 16-QAM signal. A 16-QAM signal has 4 bits per signal unit since log216 = 4. Thus, 1000*4 = 4000 bps

Example
Compute the baud rate for a 72,000-bps 64-QAM signal. A 64-QAM signal has 6 bits per signal unit since log2 64 = 6. Thus, 72000 / 6 = 12,000 baud
McGraw-Hill The McGraw-Hill Companies, Inc., 2004

Hartley-Shannon Theorem

1-23

The question is how many levels in amplitude, frequency or phase we can go. If the difference between two adjacent levels is too small and comparable to the level of noise in the communication channel, then we cannot discriminate the levels. The noise puts the limit on the number of levels. s (t )
4 2 0 0 5 10 15 20 25 30 35 40 45 50

Data signal

n (t )

4 2 0 0 5 10 15 20 25 30 35 40 45 50

Noise

s (t ) + n (t )

4 2 0 0 5 10 15 20 25 30 35 40 45 50

Signal with noise

Cant discriminate
SDSU McGraw-Hill

Discrimination possible here


Marko Vuskovic, 2004 The McGraw-Hill Companies, Inc., 2004

Hartley-Shannon Theorem (cont.)


The theoretical limit of number of levels is: Root-mean-square of signal with noise __________________________________ M= Root-mean-square of noise Root-mean-square of signal and noise:

1-24

sRMS

1 2 = S = s t dt ( ) T! 0

nRMS

1 2 = N = n t dt ( ) T! 0

(S and N are average power of signal and noise respectively): Root-mean-square of signal with noise:

(s (t ) + n (t ))RMS

1 1 1 1 2 2 2 ( ( ) ( )) ( ) 2 ( ) ( ) ( ) = s t + n t dt = s t dt + s t n t dt + n t dt = S + N ! ! ! T! T T T 0 0 0 0

(s (t ) + n (t ))RMS S +N S +N M= = = n (t )RMS N N
SDSU McGraw-Hill

(1)

= 0 if signal and noise are uncorrelated and noise has zero mean
Marko Vuskovic, 2004 The McGraw-Hill Companies, Inc., 2004

Hartley-Shannon Theorem (cont.)


On the other hand we know that the data bit rate C is:

1-25

C = 2W log 2 M

(2)

where W is signal bandwidth. After we combine (1) and (2):

C = 2W log 2

S +N N

" S # C =W log 2 $ 1 + % & N'

This is the famous Hartley_Shannon equation which defines the upper theoretical limit of a communication channel capacity.

SDSU McGraw-Hill

Marko Vuskovic, 2004 The McGraw-Hill Companies, Inc., 2004

Spectral Analysis of Digital Signals


In wired media the required bandwidth was determined by the rule:

1-26

B=

Rb 1 = 2 2Tb

B - bandwidth in [Hz] = [sec-1] Rb - bit rate in [bps] Tb - bit interval in [sec]

This rule is based on the assumption that the worse case signal is a periodic signal, i.e. a train of pulses (0101010101). The first harmonic of such signal contains 81% of energy, which satisfies the 50%-energy rule.
.

In case of wireless communication the media is more hostile (noise, interferences, longer propagation distances, Doppler effect, fading,). In addition, the signal through the media is more complex due to modulation and spectrum spreading. Therefore, we need to use larger bandwidth, which is determined based on a more exact spectral analysis of the signal.
.

Since the energy of the signal in a given frequency range (band) [f1,f2] can be determined by: f2

E = ! P ( f )df
f1

where P(f) is the power spectral density (PSD) of the signal, we will characterize signals by this important function.
SDSU McGraw-Hill Marko Vuskovic, 2004 The McGraw-Hill Companies, Inc., 2004

1-27

Random Binary Sequence


A good representation of digital signals are random binary sequences, which are sequences of pulses of constant duration Tb and random amplitudes Ak, which take values A or A with equal probability. The sequence of pulses is randomly positioned along the time axis.

d(t) A

Tb

Tb t

-A td 2Tb Tb

The statistical properties of the signal are as follows:

P [Ak = A] = P [Ak = A] =

(A2 E [Ak Ak +m ] = ) * 0 otherwise

1 2 if m = 0

- Equal probability of the pulse being A or -A - Two pulses are uncorrelated

td is uniformly distributed over the interval [0,Tb]


(P[e] probability of event e; E[x] Mathematical expectation of random variable x) SDSU Marko Vuskovic, 2004 McGraw-Hill The McGraw-Hill Companies, Inc., 2004

PSD of Random Binary Sequences


It can be shown that the PSD of the random binary sequence is:

1-28

A2Tb

" sin ( fTb ) # PSD[d (t )] = P ( f ) = A Tb $ % fT b & '


2

90% energy

95%

97%

3 2 1 1 2 3 0 Tb Tb Tb Tb Tb Tb As seen the PSD of random binary sequences is a continuous function, as opposed to PSD of periodic functions, which is discrete (see next slide).
SDSU McGraw-Hill Marko Vuskovic, 2004 The McGraw-Hill Companies, Inc., 2004

PSD of Random Binary Sequences (Cont.)


The PSD of a periodic sequence of pulses dPER(t) is a discrete function:

1-29

PSD[dPER (t )] = PPER ( f ) = A2Tb

( fTb )

A2Tb

1 3 5 f = , , ,... 2Tb 2Tb 2Tb

PSD[d (t )]

PSD[dPER (t )]
2 Tb 1 Tb 1 Tb 2 Tb

f
5 2Tb

SDSU McGraw-Hill

5 2Tb

3 2Tb

1 2Tb

1 2Tb

3 2Tb

Marko Vuskovic, 2004 The McGraw-Hill Companies, Inc., 2004

PSD of Random Binary Sequences (Cont.)


It is convenient to represent the PSD in decibels (dB)

1-30

" P (f ) # 10 log10 $ % & P (0) '

dB

3 Tb

2 Tb

1 Tb

1 Tb

2 Tb

3 Tb

3dB = 100.3 = 1.995 2, 3dB = 100.3 = 0.5012 1 2 10dB = 1, 10dB = 0.1, 20dB = 100, 1dB = 1.259
SDSU McGraw-Hill Marko Vuskovic, 2004 The McGraw-Hill Companies, Inc., 2004

PSD of Random Binary Sequences (Cont.)

1-31

Decibels are used to compare power. In diagram below the peak p1 is 13 dB higher than peak p2 meaning that the corresponding peak is 20 times higher in original diagram. Similarly 18 dB mans 63 times higher.
p1 p2 13 dB 18 dB p1

dB

p3

20 times higher

63 times higher

p2

3 T b

2 T b

1 T b

1 T b

2 T b

3 T b

f
3 Tb 2 Tb 1 Tb

p3

1 Tb

2 Tb

3 Tb

P1[dB ] = P2 [dB ] + 13dB + P1 = 20P2 P1[dB ] = P3[dB ] + 18dB + P1 = 63P3


SDSU McGraw-Hill Marko Vuskovic, 2004 The McGraw-Hill Companies, Inc., 2004

1-32

BPSK Modulator
Binary PSK (BPSK) modulation can be accomplished by simply multiplying the original signal d(t) (which is a binary random sequence) by the carrier signal, which is an analog sinusoidal oscillation. After multiplication a bandpass filter is required (see next slide).
Binary bit stream
, 1. 1/ . / . 1/ . / 0 11
, cos( c t ) . cos( t ) / c / C . . cos( c t ) / . / . cos( c t ) 1 / 0

d (t )
,1 .0/ . / .0/ . / 01 1

NRZ Encoding

s '(t ) = d (t )c (t )

Band Pass Filter

BPSK output

s (t )

c (t ) = C cos( ct )

s (t ) = C cos(ct + ), {0, 180}

Carrier signal

Crystal Oscillator

This is an idealized picture. In reality there are dynamic transitions between cos(t) and cos(t), and between -cos(t) and cos(t) etc., which make the signal s(t) much more complex. Therefore we need to use the concept of PSD, and PSD shifting by the carrier (see next slide).
SDSU McGraw-Hill Marko Vuskovic, 2004 The McGraw-Hill Companies, Inc., 2004

BPSK Modulator (Cont.)


PSD of BPSK signal
The modulated signal s(t) can be represented by a PSD that is shifted along the frequency axis by fc. In order to restrict the frequencies which go into the air (and may interfere with other channels), we need to apply a bandspass filter. The filter can be designed to pass the first lobe of the PSD, or the first two lobes, etc. The first lobe usually suffices as it carries 90% of energy.
.

1-33

The first lobe requires the bandwidth which twice the bit rate of the original signal: 2

B1 = 2Rb =

Base band

Tb

Pass band

2Rb band

PSD[s '(t )]

4Rb band

PSD[d (t )]

fc

3 2 1 Tb Tb Tb

1 Tb

2 Tb

3 Tb

2 fc Tb

fc

SDSU McGraw-Hill

1 Tb

fc

1 fc + Tb

fc +

2 Tb

Marko Vuskovic, 2004 The McGraw-Hill Companies, Inc., 2004

QPSK Modulator
, 1. 1/ 0 1

1-34

I-channel

, cos(ct ) sin(ct ) . cos( t ) + sin( t ) / = 0 c c 1 ,cos(ct + 135) , cos(ct + 90 + 45) 2. = 2. / / 0 cos(ct 135) 1 0cos(ct + 180 + 45) 1

NRZ Encoding
Binary bit stream
,1.1/ 01

(in phase)

cos(ct )
Crystal Oscillator
Phase Shift

, cos(ct ) . cos( t ) / c 1 0

d (t )
,1 .0 / . / = , 90 / .1 / . 0180 1 . / 01 1

Serial to Parallel Converter


,0 .1 / 0 1

+
, sin(ct ) . sin( t ) / c 1 0

Band Pass Filter

QPSK output

s (t )

90o
sin(ct )

NRZ Encoding

10 90o 00 o

, 1. 1/ 0 1

Q-channel
(quadrature phase)

11
180

k-th transmitted symbol:

sk (t ) = Ak cos(ct ) Bk sin(ct ), Ak , Bk {1, +1}


SDSU McGraw-Hill

01 270o

Marko Vuskovic, 2004 The McGraw-Hill Companies, Inc., 2004

1-35

Determining the phase and amplitude


A cos(ct ) B sin(ct ) = C cos(ct + )

C = A2 + B 2

= atan2(B , A)

C cos(ct + ) = C cos(ct ) cos() C sin(ct )sin() = = A cos(ct ) B sin(ct ) A = C cos() C sin() B + + = tan() = B = C sin() C cos() A

SDSU McGraw-Hill

Marko Vuskovic, 2004 The McGraw-Hill Companies, Inc., 2004

/4-Shifted QPSK Modulator

1-36

NRZ Encoding
Binary bit stream
,1.1/ 01

, 1. 1/ 0 1

, cos(ct ) . cos( t ) / c 1 0

,cos(ct + 1350 ) , cos(ct ) sin(ct ) / . cos( t ) + sin( t ) / = 2 . 0 c c 1 0 cos(ct 135 1

cos(ct )
Crystal Oscillator

even bits

d (t )
,1 .0/ , 0 . / = . 135 / .1 / 0 1350 1 . / 01 1

Serial to Parallel Converter


,0 .1 / 0 1

Band Pass Filter

QPSK output

s (t )

Phase Shift

90o
sin(ct )

odd bits

NRZ Encoding k-th transmitted symbol:

, 1. 1/ 0 1

, sin(ct ) C. / 0 sin(ct ) 1

10 135o
135o

00
45o 45o

sk (t ) = Ak cos(ct ) Bk sin(ct ), Ak , Bk {1, +1}


SDSU McGraw-Hill

11

01

Marko Vuskovic, 2004 The McGraw-Hill Companies, Inc., 2004

/4-Shifted QPSK Demodulator


X
Incoming PSL Signal

1-37

LPF

A cos(! ) B sin(! ) A

(A {1,1})
!
Controller X

Comparator

a {0,1}

c (t )

VCO
o

! s (t )
c (t )

90

)) AB cos(2(!

Parallel to serial converter

[a ,b ]
Data Signal

LPF

(B {1,1})
A sin(! ) + B cos(! ) B

Comparator b {0,1}

! s (t ) = A cos(ct +! ) B sin(ct +! )
c (t ) = 2cos(ct + ) c (t ) = 2sin(ct + )

VCO Voltage Controlled Oscillator LPF Low Pass Filter

! - Unknown phase shift of the signal during propagation - Estimation of !


Marko Vuskovic, 2004 The McGraw-Hill Companies, Inc., 2004

This is simplified diagram, carrier tracking and noise suppression omitted


SDSU McGraw-Hill

1-38

Trigonometric identities used in previous slide


sin(x ) cos(y ) = 1 2[sin(x + y ) + sin(x y )] sin(x ) sin(y ) = 1 2[ cos(x + y ) + cos(x y )] cos(x ) cos(y ) = 1 2[cos(x + y ) + cos(x y )] cos 2 (x ) sin 2 (x ) = cos(2x )

In addition:
A2 B 2 = 0 if A, B {1,1}

SDSU McGraw-Hill

Marko Vuskovic, 2004 The McGraw-Hill Companies, Inc., 2004

1-39

Simplified Representation of QPSK Transmitter/Receiver

fc d (t )
S/P NRZ

I/Q
[Ak , Bk ]
MOD BPF

s (t )

fc d (t )
P/S COMP

I/Q
LPF DEMOD

s (t )

SDSU McGraw-Hill

Marko Vuskovic, 2004 The McGraw-Hill Companies, Inc., 2004

1-40

Differential /4-Shifted QPSK (DQPSK)

BPSK, QPSK and p/4-shifted QPSK require coherent demodulation, which assumes exact knowledge of the frequency and phase of the incoming carrier wave. The synchronization of the local (receivers) oscillator can be done with an additional pilot carrier signal transmitted along with the data signal, or with a very complex circuitry that increase the space and energy requirements for the wireless device. Alternative is differential PSK which uses the phase changes instead of the absolute phase values for each symbol. This allows for noncoherent demodulation. The phase changes act as synchronous reference. Differential /4-Shifted QPSK (DQPSK) has slightly higher BER but requires simpler receiver circuitry. It is used in wireless LANs (802.11) and in second generation cellular telephony (IS-95).

SDSU McGraw-Hill

Marko Vuskovic, 2004 The McGraw-Hill Companies, Inc., 2004

1-41

Multi Carrier Modulation

This is a modulation technique which is recently used to fight multipath interference at very high data rates. The modulation is performed with a multitude of carriers (subcarriers) instead of with only a single carrier. The modulators in each carrier can be any of single carrier modulators considered so far (BPSK, QPSK, DQPSK, QAM). One of the important MCM methods is Orthogonal Frequency Division Multiplexing (OFDM).

SDSU McGraw-Hill

Marko Vuskovic, 2004 The McGraw-Hill Companies, Inc., 2004

Multipath Transmission and Intersymbol Interference (ISI)

1-42

It is a common situation (specially in urban areas) that receiver picks several signals from the same transmitter, which have propagated through different paths.

McGraw-Hill

The McGraw-Hill Companies, Inc., 2004

ISI (Cont.)

1-43

Since delays for different paths can be different, it is possible that the previous symbol interferes with the current symbol, specially if the difference in delay between two paths () is in the same order of magnitude as the symbol interval (Ts).

Path 1

Typical interpath delays are 100 ns, therefore the symbol rates of several Mbps and greater can be affected by ISI.

Symbol Symbol k-1 k

t
Path 2 Delay

Received signal

Ts

Symbol k

t
SDSU McGraw-Hill Marko Vuskovic, 2004 The McGraw-Hill Companies, Inc., 2004

1-44

Multi Carrier Modulation


In order to diminish the impact of ISI the baseband signal is split into N parallel signals, which are then modulated with different subcarriers. While the baseband signal has symbol rate of Rs = 1/Ts the subcarrier signals have N times smaller rate, R = Rs/N = 1/(NTs).

BPSK:

1010011110011110
16 symbols in time

1 0 1 0 10 10 01 11

0 1 1 1

1 0 0 1

1 1 1 0 10 01 11 10

4 symbols in time

QPSK:

1010011110011110
8 symbols in time
SDSU McGraw-Hill

2 symbols in time

Marko Vuskovic, 2004 The McGraw-Hill Companies, Inc., 2004

Multi Carrier Modulator


2Rs

1-45

d0 (t )
f
1/Ts

f0
MOD MOD

-1/Ts

d1 (t )
S/P converter

f1 f2

BPF BPF

The data rate of this signal is N times smaller, so is the effect of ISI

d (t )

d 2 (t )

......

MOD

BPF

s (t )

Symbol rate Rs=1/Ts

fN 1
MOD
N 1 k =0

dN 1 (t )

BPF

s (t ) = 2 Ak (t )cos(2fkt ) Bk (t )sin(2fk t )
......
2Rs/N

Symbol rate R = Rs/N = 1/(NTs)


SDSU McGraw-Hill

fN-1 f1 f0The McGraw-Hill Marko Vuskovic, 2004 Companies, Inc., 2004

Multi Carrier Demodulator


f0
DEM DEM

1-46

f1 f2

LPF LPF

d0 (t ) d1 (t ) d 2 (t )
P/S converter

s (t )
DEM

d (t )

LPF

.....

fN 1
DEM LPF

dN 1 (t )

Symbol rate Rs=1/Ts

Symbol rate R = Rs/N = 1/(NTs)


SDSU McGraw-Hill Marko Vuskovic, 2004 The McGraw-Hill Companies, Inc., 2004

1-47

OFDM
The multicarrier modulation solves the problem of ISI if there are enough parallel carriers (typically few thousand). However this method suffers from two serious problems: (1) High bandwidth demand (2) To complex circuitry (mainly because of bandpass filters) OFDM solves both problems by using orthogonal subcarriers, which allow spectrum overlapping and dont require bandpass filters for each subcarrier. The circuitry is still complex but with VLSI technology, the approach became feasible, specially with the mass production. As will be shown, the implementation makes use of highly efficient FFT (Fast Fourier Transform) and IFFT (Inverse FFT) algorithms. OFDM is used in IEEE 802.11a at bit rate 54 Mbps with 64 subcarriers.
SDSU McGraw-Hill Marko Vuskovic, 2004 The McGraw-Hill Companies, Inc., 2004

OFDM (Cont.)
fn 1 fn fn = n NTs
1 1 NTs NTs

1-48

OFDM MCM uses subcarrier frequencies which satisfy the following condition:

This has two consequences: (1) The shifted spectra overlap 50% (which help saving bandwidth) (2) The subcarrier waves are orthogonal Orthogonality is defined as:

2 NTs

NTs

!
0

i 2 j 2 2 sin( t )sin( t )dt = NTs NTs NTs


NTs

NTs

!
0

cos(

(1 if i = j i 2 j 2 t )cos( t )dt = ) NTs NTs *0 if i j

!
0

sin(

i 2 j 2 t ) cos( t )dt = 0 NTs NTs


Marko Vuskovic, 2004 The McGraw-Hill Companies, Inc., 2004

SDSU McGraw-Hill

OFDM (Cont.)
NTs

1-49

The separation of a particular component from the signal mix can be simply obtained by multiply-and-integrate procedure, no low pass filters needed:

2 NTs 2 NTs

!
0

" N 1 # $ 2 An cos(2fnt ) + Bn sin(2fnt ) % cos(2fmt )dt = Am & n =0 ' " N 1 # $ 2 An cos(2fnt ) + Bn sin(2fnt ) % sin(2fmt )dt = Bm & n =0 '
cos(m 2 t) NTs

NTs

!
0

s (t )

x x
sin(m

2 NTs 2 NTs
2 t) NTs

NTs

!
0

(...)dt

Am

NTs

! (...)dt
0

Bm

SDSU McGraw-Hill

Marko Vuskovic, 2004 The McGraw-Hill Companies, Inc., 2004

1-50

MCM with FDM

MCM with OFDM

Saving of the bandwidth f

Based on: Introduction to OFDM, TUD-TVS, by Dusan Matic

McGraw-Hill

The McGraw-Hill Companies, Inc., 2004

OFDM (Cont.)

1-51

After examining the modulation and demodulation equations it can be seen that they resemble discrete inverse Fourier transform and discrete Fourier transform. Therefore the implementation can use DSP chips for inverse FFT (IFFT) and FFT. Since these chips work completely in digital domain, digital to analog conversion is needed before outputting the signal to RF stage. Similarly analog to digital conversion needed after the signal is received from the RF input.

d (t )

s (t )

S/P

IFFT

P/S

D/A

LPF

d (t )

s (t )

P/S

FFT

S/P

A/D

LPF

Parallel digital data


SDSU McGraw-Hill

Serial digital data

Analog signal

Marko Vuskovic, 2004 The McGraw-Hill Companies, Inc., 2004

Vous aimerez peut-être aussi