tsks04&digital&communication - linköping university · 2016-01-22 ·...

19
TSKS04 Digital Communication Continuation Course Lecture 2 Complex Baseband Representation and Bandwidth Emil Björnson Department of Electrical Engineering (ISY) Division of Communication Systems

Upload: others

Post on 13-Jun-2020

2 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: TSKS04&Digital&Communication - Linköping University · 2016-01-22 · TSKS04&Digital&Communication Continuation(Course Lecture&2 Complex&Baseband&Representation&and&Bandwidth Emil&Björnson

TSKS04  Digital  CommunicationContinuation  Course

Lecture  2

Complex  Baseband  Representation  and  Bandwidth

Emil  Björnson

Department  of  Electrical  Engineering  (ISY)Division  of  Communication  Systems

Page 2: TSKS04&Digital&Communication - Linköping University · 2016-01-22 · TSKS04&Digital&Communication Continuation(Course Lecture&2 Complex&Baseband&Representation&and&Bandwidth Emil&Björnson

Complex  Baseband  Representation

Real-­valued  baseband  information  signals  (bandwidth  𝐵/2):𝑠%(𝑡) and  𝑠)(𝑡)

Real-­valued  passband  signal  (𝑓+ ≫ 𝐵):𝑠 𝑡 = 𝑠% 𝑡 2 cos 2𝜋𝑓+𝑡 − 𝑠) 𝑡 2 sin 2𝜋𝑓+𝑡

2016-­01-­22 TSKS04  Digital  Communication  Continuation  Course  -­ Lecture  2 2

In-­phase Quadrature-­phase

Complex  Baseband  Representation  or Complex  envelope�̃� 𝑡 = 𝑠% 𝑡 + 𝑗𝑠)(𝑡)

Relationships:𝑠 𝑡 = 𝑅𝑒 2�̃� 𝑡 𝑒:;<=>?

Page 3: TSKS04&Digital&Communication - Linköping University · 2016-01-22 · TSKS04&Digital&Communication Continuation(Course Lecture&2 Complex&Baseband&Representation&and&Bandwidth Emil&Björnson

Processing  in  Complex  Baseband

§ Signal  processing  at  baseband§ Use  general  purpose  hardware§ Sampling  frequency  depend  on  baseband  signal

§ Compact  descriptions  of  algorithms  and  analysis

2016-­01-­22 TSKS04  Digital  Communication  Continuation  Course  -­ Lecture  2 3

Create  complex  baseband  signal

Modulate   to  carrier  

frequencySend  over  channel

Demodulate  to  baseband

Sample  complex  baseband  signal

Page 4: TSKS04&Digital&Communication - Linköping University · 2016-01-22 · TSKS04&Digital&Communication Continuation(Course Lecture&2 Complex&Baseband&Representation&and&Bandwidth Emil&Björnson

Frequency  Domain:  Baseband  and  Passband

Fourier  transforms:   𝑆A 𝑓 = ℱ �̃� 𝑡

𝑆 𝑓 =12𝑆A 𝑓 − 𝑓+ + 𝑆A∗ −𝑓 − 𝑓+

2016-­01-­22 TSKS04  Digital  Communication  Continuation  Course  -­ Lecture  2 4

+𝑓+−𝑓+

0

𝑆A 𝑓

𝑆 𝑓

𝐵/2

𝐵

Page 5: TSKS04&Digital&Communication - Linköping University · 2016-01-22 · TSKS04&Digital&Communication Continuation(Course Lecture&2 Complex&Baseband&Representation&and&Bandwidth Emil&Björnson

Inner  Products  of  Complex  Signals

Definition:  Inner  product  of  two  complex  signals  𝑎 𝑡  and  𝑏 𝑡 :

< 𝑎, 𝑏 >  = L 𝑎 𝑡 𝑏∗ 𝑡 𝑑𝑡N

ON§ Property:  < 𝑎 + 𝑏 , 𝑐 >  =  < 𝑎, 𝑐 > +  < 𝑏, 𝑐 >

Parseval’s identity:

L 𝑎 𝑡 𝑏∗ 𝑡 𝑑𝑡N

ON= L 𝐴 𝑓 𝐵∗ 𝑓 𝑑𝑓

N

ON

Consequences:  

< 𝑎,𝑏 >  = L 𝐴 𝑓 𝐵∗ 𝑓 𝑑𝑓N

ON=  < 𝐴,𝐵 >

𝑎 ; =  < 𝑎, 𝑎 >  =  < 𝐴,𝐴 >  = 𝐴 ;

2016-­01-­22 TSKS04  Digital  Communication  Continuation  Course  -­ Lecture  2 5

Page 6: TSKS04&Digital&Communication - Linköping University · 2016-01-22 · TSKS04&Digital&Communication Continuation(Course Lecture&2 Complex&Baseband&Representation&and&Bandwidth Emil&Björnson

Orthogonality  between  I-­ and  Q-­parts

Two  passband  signals:𝑎 𝑡 = 𝑎% 𝑡 2 cos 2𝜋𝑓+𝑡 − 𝑎) 𝑡 2 sin 2𝜋𝑓+𝑡𝑏 𝑡 = 𝑏% 𝑡 2 cos 2𝜋𝑓+𝑡 − 𝑏) 𝑡 2 sin 2𝜋𝑓+𝑡

2016-­01-­22 TSKS04  Digital  Communication  Continuation  Course  -­ Lecture  2 6

Can  treat  I and  Q parts  separately< 𝑎,𝑏 >  =  < 𝑎%, 𝑏% > +  < 𝑎), 𝑏) >  = 𝑅𝑒 < 𝑎R, 𝑏S >

𝐸U = 𝑎 ; = 𝑎% ;+ 𝑎);= 𝑎R ; = 𝐸UR

Page 7: TSKS04&Digital&Communication - Linköping University · 2016-01-22 · TSKS04&Digital&Communication Continuation(Course Lecture&2 Complex&Baseband&Representation&and&Bandwidth Emil&Björnson

Recall:  Linearly  Modulated  Signal

Pulse-­amplitude  modulated  signal:

𝑆 𝑡 = V V𝑆W 𝑛 𝜙W(𝑡 − 𝑛𝑇 −Ψ)\

W]+

N

^]ON

2016-­01-­22 TSKS04  Digital  Communication  Continuation  Course  -­ Lecture  2 7

𝑛th  data  symbol(stochastic  as  before)𝑛 ∈ … , −1,0, +1, … = ℤ

𝑖th basis  function(not  necessarily  time-­limited)

Time  delayRandom  delay

How  to  select  𝑁 and  𝑇?

Page 8: TSKS04&Digital&Communication - Linköping University · 2016-01-22 · TSKS04&Digital&Communication Continuation(Course Lecture&2 Complex&Baseband&Representation&and&Bandwidth Emil&Björnson

Recall:  Sine-­Shaped  Basis  Functions  Same  frequency  (𝑁 = 2)

2016-­01-­22 TSKS04  Digital  Communication  Continuation  Course  -­ Lecture  2 8

Essentially  the  same  energy  spectra!90%-­bandwidth:

2/𝑇

Guard  bands  to  not  interfere  with  other  

systems

Page 9: TSKS04&Digital&Communication - Linköping University · 2016-01-22 · TSKS04&Digital&Communication Continuation(Course Lecture&2 Complex&Baseband&Representation&and&Bandwidth Emil&Björnson

Linear  Modulation  in  Complex  Baseband

Linear modulation  with sine-­shaped basis  functions:

𝑆 𝑡 = V 𝑆+ 𝑛2𝑇 cos 2𝜋𝑓+𝑡 𝐼[^f,^fgf]

N

^]ON+ 𝑆; 𝑛

2𝑇 sin 2𝜋𝑓+𝑡 𝐼[^f,^fgf]

Compare with:

𝑠 𝑡 = 𝑠% 𝑡 2 cos 2𝜋𝑓+𝑡 − 𝑠) 𝑡 2 sin 2𝜋𝑓+𝑡

𝑠% 𝑡 = V 𝑆+ 𝑛1𝑇 𝐼[^f,^fgf]

N

^]ON, 𝑠) 𝑡 = − V 𝑆; 𝑛

1𝑇 𝐼[^f,^fgf]

N

^]ON

2016-­01-­22 TSKS04  Digital  Communication  Continuation  Course  -­ Lecture  2 9

Pulse  in  PAM Pulse  in  PAM

Page 10: TSKS04&Digital&Communication - Linköping University · 2016-01-22 · TSKS04&Digital&Communication Continuation(Course Lecture&2 Complex&Baseband&Representation&and&Bandwidth Emil&Björnson

Recall:  Nyquist  Criterion  for  ISI-­Free  CommunicationSampling  of  𝑦 𝑡 = ∑ 𝑥 𝑘 𝑝(𝑡 − 𝑘𝑇)o at  time  𝑛𝑇:

𝑧 𝑛 = 𝑦 𝑛𝑇 =V𝑥 𝑘 𝑝(𝑛𝑇 − 𝑘𝑇)o

Inter-­symbol  interference  (ISI):  𝑧[𝑛] depends  on  𝑥[𝑘] for  𝑘 ≠ 𝑛.

Goal:  ISI-­free  communication!𝑧 𝑛 = 𝐾 ⋅ 𝑥 𝑛 for  constant  𝐾 ≠ 0.

2016-­01-­22 TSKS04  Digital  Communication  Continuation  Course  -­ Lecture  2 10

Achieved  when    𝑝 𝑛𝑇 = 𝐾𝛿 𝑛        ⟺        1𝑇

V 𝑃 𝑓 −𝑚𝑇 = 𝐾

N

x]ON

Nyquist ISI  criterion:  Constant!

Page 11: TSKS04&Digital&Communication - Linköping University · 2016-01-22 · TSKS04&Digital&Communication Continuation(Course Lecture&2 Complex&Baseband&Representation&and&Bandwidth Emil&Björnson

A  pulse with bandwidth 1/(2T) has  to  be  a  sinc to  be  Nyquist

2016-­01-­22 TSKS04  Digital  Communication  Continuation  Course  -­ Lecture  2 11

A  pulse  𝑝 𝑡  is  said  to  be  𝐍𝐲𝐪𝐮𝐢𝐬𝐭  if V 𝑃 𝑓 −𝑚𝑇

N

x]ON

   is  constant

Page 12: TSKS04&Digital&Communication - Linköping University · 2016-01-22 · TSKS04&Digital&Communication Continuation(Course Lecture&2 Complex&Baseband&Representation&and&Bandwidth Emil&Björnson

A  pulse with bandwidth more than 1/(2T)can have many shapes to  be  Nyquist

2016-­01-­22 TSKS04  Digital  Communication  Continuation  Course  -­ Lecture  2 12

A  pulse  𝑝 𝑡  is  said  to  be  𝐍𝐲𝐪𝐮𝐢𝐬𝐭  if V 𝑃 𝑓 −𝑚𝑇

N

x]ON

   is  constant

Page 13: TSKS04&Digital&Communication - Linköping University · 2016-01-22 · TSKS04&Digital&Communication Continuation(Course Lecture&2 Complex&Baseband&Representation&and&Bandwidth Emil&Björnson

Sine-­Shaped  Basis  Functions: A  bad  choice?

§ Pulse  shape:  𝑝 𝑡 = +f 𝐼[�,f]

§ Fourier  transform:  𝑃 𝑓 = − +f sinc 𝑓𝑇

§ Strictly  speaking:  Bandwidth  𝐵 = ∞§ Guard  bands  needed?

§ 90%-­energy  bandwidth:  1/𝑇§ 95%-­energy  bandwidth:  2/𝑇§ 99%-­energy  bandwidth:  10/𝑇

2016-­01-­22 TSKS04  Digital  Communication  Continuation  Course  -­ Lecture  2 13

Better  choices  from  TSKS01?• Sinc function• Raised  cosine  (𝛼 excess  bandwidth)

Smaller  if  we  increase  𝑇

Page 14: TSKS04&Digital&Communication - Linköping University · 2016-01-22 · TSKS04&Digital&Communication Continuation(Course Lecture&2 Complex&Baseband&Representation&and&Bandwidth Emil&Björnson

Sine-­Shaped  Basis  Functions    (1/3)  𝑁 different  frequencies

§ How  closely  can  we  place  sine-­shaped  basis  functions?

𝜙+ 𝑡 = 2/𝑇 cos 2𝜋𝑓+𝑡 , for  0 ≤ 𝑡 < 𝑇,

𝜙; 𝑡 = 2/𝑇 cos 2𝜋𝑓;𝑡 , for  0 ≤ 𝑡 < 𝑇,

𝜙� 𝑡 = 2/𝑇 sin 2𝜋𝑓;𝑡 , for  0 ≤ 𝑡 < 𝑇,

Orthogonal  if:

L 𝜙+ 𝑡f

�𝜙; 𝑡 𝑑𝑡 = L cos 2𝜋(𝑓++𝑓;)𝑡 + cos 2𝜋(𝑓+−𝑓;)𝑡

f

�𝑑𝑡 = 0

L 𝜙+ 𝑡f

�𝜙� 𝑡 𝑑𝑡 = L sin 2𝜋(𝑓++𝑓;)𝑡 + sin 2𝜋(𝑓+−𝑓;)𝑡

f

�𝑑𝑡 = 0

2016-­01-­22 TSKS04  Digital  Communication  Continuation  Course  -­ Lecture  2 14

2𝑓+𝑇, 2𝑓;𝑇 integers

𝑓+𝑇, 𝑓;𝑇 integers

Page 15: TSKS04&Digital&Communication - Linköping University · 2016-01-22 · TSKS04&Digital&Communication Continuation(Course Lecture&2 Complex&Baseband&Representation&and&Bandwidth Emil&Björnson

Sine-­Shaped  Basis  Functions    (3/3)  𝑁 different  frequencies

2016-­01-­22 TSKS04  Digital  Communication  Continuation  Course  -­ Lecture  2 15

Independent  basis  functions:Still substantial overlap in  frequency!

Φ+ 𝑓 ;Φ; 𝑓 ; = Φ� 𝑓 ;

Guard  band = 𝛽/𝑇

Distance between carriers:𝑓+ − 𝑓; = 𝑘/𝑇,  𝑘 = integer

Page 16: TSKS04&Digital&Communication - Linköping University · 2016-01-22 · TSKS04&Digital&Communication Continuation(Course Lecture&2 Complex&Baseband&Representation&and&Bandwidth Emil&Björnson

Sine-­Shaped  Basis  Functions    (3/3)  𝑁 different  frequencies

2016-­01-­22 TSKS04  Digital  Communication  Continuation  Course  -­ Lecture  2 16

Guard  band = 𝛽/𝑇(𝑁+ 1)/𝑇

Guard  band = 𝛽/𝑇

Total  bandwidth:

𝐵 =𝑁 + 1 + 2𝛽

𝑇

Page 17: TSKS04&Digital&Communication - Linköping University · 2016-01-22 · TSKS04&Digital&Communication Continuation(Course Lecture&2 Complex&Baseband&Representation&and&Bandwidth Emil&Björnson

Single-­Carrier  vs.  Multi-­Carrier  Signals

§ Available  passband  bandwidth:  𝐵

§ Single-­carrier  transmission§ One  carrier  frequency  𝑓+§ Use  raised  cosine  pulse  with  bandwidth  𝐵 = +g�

f§ Symbol  time  𝑇 = 1 + 𝛼 /𝐵

§ Multi-­carrier  transmission§ Use  𝑁 carrier  frequencies  𝑓+, 𝑓; ,… , 𝑓\§ Total  bandwidth:  𝐵 = \g+g;�

f§ Symbol  time  𝑇 = (𝑁 + 1 + 2𝛽)/𝐵 ≈ 𝑁/𝐵 if  𝑁 is  large

2016-­01-­22 TSKS04  Digital  Communication  Continuation  Course  -­ Lecture  2 17

Example:  OFDM  systems:

𝑁 ∝ 1000

Around  1/𝐵 complex  symbols/s  in  both  cases  (e.g.,  QPSK,  16-­QAM)

Page 18: TSKS04&Digital&Communication - Linköping University · 2016-01-22 · TSKS04&Digital&Communication Continuation(Course Lecture&2 Complex&Baseband&Representation&and&Bandwidth Emil&Björnson

Impact  of  Channel  Filter

Classification:§ Non-­dispersive:  𝑔� 𝑡 = 𝛿(𝑡)

§ Time-­dispersive:  Spreads  signal  in  time

2016-­01-­22 TSKS04  Digital  Communication  Continuation  Course  -­ Lecture  2 18

𝑔� 𝑡𝑠(𝑡) 𝑟 𝑡 = (𝑠 ∗ 𝑔�)(𝑡)

Delay  spread:Time  interval  where  𝑔� 𝑡 ≠ 0

Approximately  non-­dispersive:  Delay  spread  small  compared  to  symbol  time

Page 19: TSKS04&Digital&Communication - Linköping University · 2016-01-22 · TSKS04&Digital&Communication Continuation(Course Lecture&2 Complex&Baseband&Representation&and&Bandwidth Emil&Björnson

www.liu.se