Advanced Digital Communications

Chapter II: Non-ideal Channels

Prepared by: Dr. TAMI Abdelkader

University of Saida Dr. Moulay Tahar 2020/2021

1
II. 1- Wireless channels, multipath,
noise, interference

II.2- Invariant and variant channels

II.3- Rayleigh Rice and fading

2
 II. 1- Wireless channels, multipath,
noise, interference

Introduction:

Different Radio-Mobil Generations

Rate Services
2nd Generation
(2G-2.5G-2.75G)
10-200kb/s Voice-Some basic Data
(GSM-CDMA(2G);
(GPRS),(EDGE)
3rd Generation
(3G)
(WCDMA:UMTS) Voice- Data
300kb-30Mb/s
3.5 G Video calling
(HSDPA/HSUPA)-
(1xEVDO)
Voice – Data
4th Generation
100kb-200Mb/s -HDTV (High Definition TV)
(LTE-Wimax)
- Online Gamming
Smart Home, e-Health, Drones,
5th Generation 10Gb/s Autonomous-cars, IoT, Real time, Virtual
reality, Reality Gaming

3
-Wireless channels:

LOS: Line of Sight

(Directed path)

NLOS No Line of Sight

(Reflected path,

Scattered path)

Wireless Channel

 Multiple component in a reception

 Multiple path propagation
 Constructive & destructive interference
𝐿−1
Channel Response: 𝑕 𝑡 = 𝑖=0 𝑎𝑖 𝛿 𝑡 − 𝜏𝑖

Signal output channel: 𝑦 𝑡 = 𝑕 𝑡 ∗ 𝑠(𝑡),

𝐿−1

𝑦 𝑡 = 𝑅𝑒 𝑎𝑖 𝑠𝑏 𝑡 − 𝜏𝑖 𝑒 −2𝑗 𝜋𝑓𝑐 𝜏 𝑖 𝑒 −2𝑗 𝜋𝑓𝑐 𝑡

𝑖=0

Complex signal base Carrier

band= 𝑦𝑏 𝑡 Complex
Received signal phase factor

4
 Wireless signal
𝑠 𝑡 = 𝑅𝑒 𝑠𝑏 (𝑡)𝑒 2𝑗 𝜋𝑓𝑐 𝑡

Pass band
Presentation
Complex base
band signal

Narrow band assumption:

𝑠𝑏 𝑡

𝑓
−𝑓𝑚 𝑓𝑚

If 𝑓𝑚 < 1 𝜏𝑖 for all 𝑖 (delay) → signal is a narrow band

𝑠𝑏 𝑡 − 𝜏𝑖 ≈ 𝑠𝑏 𝑡 ;
𝐿−1 −2𝑗 𝜋𝑓𝑐 𝜏 𝑖
So 𝑦𝑏 𝑡 = 𝑠𝑏 𝑡 𝑖=0 𝑎𝑖 𝑒 ; Delay is not significant at reception
𝐿−1

𝑦 𝑡 = 𝑅𝑒 𝑠𝑏 𝑡 𝑎𝑖 𝑒 −2𝑗 𝜋𝑓𝑐 𝜏 𝑖 𝑒 −2𝑗 𝜋𝑓𝑐 𝑡

𝑖=0

Fading coefficients:

𝑦 𝑡 = 𝑕 𝑡 ∗ 𝑠 𝑡 ; with 𝑠 𝑡 = 𝑅𝑒 𝑠𝑏 𝑡 𝑒−2𝑗𝜋𝑓𝑐 𝑡 and 𝑕 𝑡 = 𝐿−1

𝑖=0 𝑎𝑖 𝛿 𝑡 − 𝜏𝑖

Narrow band signal: 𝑠𝑏 𝑡 − 𝜏𝑖 ≈ 𝑠𝑏 𝑡 ⟹ 𝑦 𝑡 = 𝑅𝑒 𝑠𝑏 𝑡 𝑕 𝑒 2𝑗𝜋 𝑓𝑐 𝑡

−2𝑗𝜋 𝑓𝑐 𝜏 𝑖
For wireless system: 𝑦𝑏 𝑡 = 𝒉𝑠𝑏 𝑡 with 𝒉 = 𝐿−1
𝑖=0 𝑎𝑖𝑒

𝒉 is complex fading coefficients

for wireline system: 𝑦𝑏 𝑡 = 𝑠𝑏 𝑡

5
Analytical models:

Statistics of the fading coefficients:

𝒉 is a complex fading coefficient : 𝒉 = 𝑋 + 𝑗𝑌 = 𝑎𝑒 𝑗𝜑

If 𝑋, 𝑌 are two random variables their PDF (Probability Density Function)

are Gaussian, so 𝒉 has a PDF of Rayleigh

Random Variable (Random Variable)

+ 𝒋 (Random Variable)
(PDF Gauss) (PDF Gauss) (PDF Rayleigh)

𝐿−1 −2𝑗 𝜋𝑓𝑐 𝜏 𝑖 𝐿−1 𝐿−1

𝒉= 𝑖=0 𝑎𝑖 𝑒 = 𝑎𝑒 𝑗𝜑 = 𝑖=0 𝑎𝑖 cos 2𝑗𝜋𝑓𝑐 𝜏𝑖 − 𝑗 𝑖=0 𝑎𝑖 sin 2𝑗𝜋𝑓𝑐 𝜏𝑖 ,
𝐿−1 𝐿−1
With : 𝑋 = 𝑖=0 𝑎𝑖 cos 2𝑗𝜋𝑓𝑐 𝜏𝑖 ; 𝑌 = − 𝑖=0 𝑎𝑖 sin 2𝑗𝜋𝑓𝑐 𝜏𝑖

𝑋: Gaussian random variable, 𝑋 ⟶ 𝒩 𝜇𝑥 , 𝜎𝑥 2 where 𝜇𝑥 is average (mean) and 𝜎𝑥 2 is variance

𝑥 −𝜇 𝑥 2
1 −
PDF of 𝑋 is: 𝑓𝑋 𝑥 = 𝑒 2𝜎 𝑥 2
2𝜋𝜎𝑥 2

𝑌 : Gaussian random variable 𝑌 ⟶ 𝒩 𝜇𝑦 , 𝜎𝑦 2

2
𝑦 −𝜇 𝑦
1 −
2𝜎 𝑦 2
PDF of 𝑌 is: 𝑓𝑌 𝑦 = 𝑒
2𝜋𝜎𝑦 2

1 1
-For standard Norm: 𝑋 ⟶ 𝒩 0, , and 𝑌 ⟶ 𝒩 0,
2 2

1 2 1 2
𝑓𝑋 𝑥 = 2
𝑒 − 𝑥 ; 𝑓𝑌 𝑦 = 𝑒− 𝑦
2𝜋𝜎𝑥 2𝜋𝜎𝑦 2

1 𝑥 2 +𝑦 2
Joint random distribution is 𝑓𝑋,𝑌 𝑥, 𝑦 = 𝜋 𝑒 −

After change of variables and plan 𝑥, 𝑦 ⟶ 𝑎, 𝜑

𝑎 2
Joint random distribution: 𝑓𝐴,Φ 𝑎, 𝜑 = 𝜋 𝑒 −𝑎

Joint random distribution random variable of 𝐴, Φ

-Marginal distribution of the phase 𝚽:

∞ 1
𝑓Φ 𝜑 = 𝑓
0 𝐴,Φ
𝑎, 𝜑 𝑑𝑎; 𝑓Φ 𝜑 = 2𝜋 is uniform distribution

6
 𝑓Φ 𝜑

1
2𝜋

𝑓
−𝜋 𝜋

-Marginal distribution of the magnitude:

𝜋 2
𝑓A 𝑎 = 𝑓
−𝜋 𝐴,Φ
𝑎, 𝜑 𝑑𝜑; 𝑓A 𝑎 = 2𝑎𝑒 −𝑎 is Rayleigh Fading Density

𝑓A 𝑎

𝑎
Deep Fading

7
-BER for wireline and wireless communication system

a) BER for wireline communication system:

𝑦 = 𝑥 + 𝑛; 𝒉 = 1 and 𝑛 is AWGN noise with 𝒩 0, 𝜎𝑛 2

For BPSK modulation:

The bit: 1 ⟹ +1 × 𝑃; 𝑃 is transmitted power

The bit: 0 ⟹ −1 × 𝑃

If "0" is transmitted, the received signal is: 𝑦 = − 𝑃 + 𝑛 , if an error is committed

𝑦>0⟹𝑛> 𝑃
+∞ 1 2 2𝜎𝑛 2
Error probability: 𝑝(𝑛 > 𝑃) = 𝐸𝑏
e−𝑥 𝑑𝑥 ; Gaussian PDF of noise 𝑛
2𝜋𝜎𝑛 2

𝑥
Change of variable: 𝜎 = 𝑡 ⟹ 𝑑𝑥 = 𝜎𝑛 𝑑𝑡
𝑛

+∞
1 2 𝑃
𝑝 𝑛> 𝑃 = e−𝑡 2
𝑑𝑡 = 𝑄 =𝑄 𝑆𝑁𝑅 = 𝐵𝐸𝑅
2𝜋 𝜎𝑛 2
𝑃
𝜎𝑛 2

𝑄: Cumulative distribution function of standard Gaussian variable

+∞
1 2
𝑄 𝑥 = e−𝑢 2
𝑑𝑢
2𝜋
𝑥

8
b) BER for wireless communication system:
𝑦 = 𝒉𝑥 + 𝑛; 𝒉 = 𝑎𝑒 𝑗𝜑

2 𝑃𝑎 2
Received power 𝑃 × 𝒉 = 𝑃𝑎2 ; received SNR is : 𝑆𝑁𝑅 = 𝜎𝑛 2

𝑃𝑎 2
𝐵𝐸𝑅 = 𝑄 𝑆𝑁𝑅 = 𝑄 ; 𝑎2 is random variable so
𝜎𝑛 2

𝐵𝐸𝑅 is a function of 𝑎 so we take an average of BER,

+∞
Average of BER is : 𝐸 𝐵𝐸𝑅 = 0
𝐵𝐸𝑅 𝑎 𝑝(𝑎) 𝑑𝑡 ;

2 𝑃𝑎 2
where 𝑝 𝑎 = 𝑓𝐴 𝑎 = 2𝑎𝑒 −𝑎 ; 𝐵𝐸𝑅 𝑎 = 𝑄 𝜎𝑛 2

+∞ +∞
1 2 2
𝐵𝐸𝑅𝑎𝑣 = 𝐸 𝐵𝐸𝑅 = e−𝑥 2
𝑑𝑥 × 2𝑎𝑒 −𝑎 𝑑𝑎
0 𝑃𝑎 2 2𝜋
𝜎𝑛 2

Average BER of wireless communication system is:

1 𝑆𝑁𝑅 𝑃
𝐵𝐸𝑅 = 1− ; where 𝑆𝑁𝑅 = 𝜎 2
2 2+𝑆𝑁𝑅 𝑛

1 1 1 2
For high snr : 𝐵𝐸𝑅 = 1 − 1/ 1 + 2/𝑆𝑁𝑅 ≈ 1− ⟹
2 2 2 𝑆𝑁𝑅

1
𝐵𝐸𝑅 ≈
2𝑆𝑁𝑅

Remark: For the same transmitted power 𝐵𝐸𝑅(Wireline)<< 𝐵𝐸𝑅(Wireless)

For wireless system: 𝑦 = 𝒉𝑥 + 𝑛; 𝒉 = 𝑎𝑒 𝑗𝜑 fading coefficient

2
If the received power 𝑃 𝒉 < 𝜎𝑛 2 ; (𝜎𝑛 2 : noise power)
𝜎𝑛 2
𝑃𝑎2 < 𝜎𝑛 2 ⟹ 𝑎2 < = (𝑆𝑁𝑅)−1 Bad performance (Deep Fading)
𝑃

Probability of deep fade event:

9
 1/ 𝑆𝑁𝑅 2 2
𝑝 𝑎 < 1/ 𝑆𝑁𝑅 = 0
2𝑎e−𝑎 𝑑𝑎 = 1/𝑆𝑁𝑅; e−𝑎 ≈ 1

Introduction to diversity

For destructive interference, can improve performance by using spatial diversity .

Multiple Link Tx Rx
Tx Rx

Fading
No Fading

The diversity order is the number of independent link.

Multiple Antenna Maximum Ratio Combiner (MRC):

𝑕1 𝑦1

𝑥
𝑦2
𝑊 𝑥
𝑕𝐿

𝑦𝐿

𝑦1 𝑕1 𝑛1 𝑕1
⋮ ⋮ 𝐿−1 −2𝑗𝜋𝑓𝑐 𝜏𝑘
𝑦 = 𝒉𝑥 + 𝑛; ⇔ = ⋮ 𝑥+ where ⋮ is channel vector , 𝑕𝑖 = 𝑘=0 𝑎𝑘𝑒
𝑦𝐿 𝑕𝐿 𝑛𝐿 𝑕𝐿

. : means vector form

10
Beam forming:

𝑦1
𝐻
𝑥 = 𝑊 𝑦 = 𝑤1 ⋯ 𝑤𝐿 ∗ ∗ ⋮ = 𝑊 𝐻 𝑕𝑥 + 𝑊 𝐻 𝑛
𝑦𝐿

Beam-former
Signal Noise
Component
Component

2
𝑊𝐻 𝑕 𝑃 𝑕
𝑆𝑁𝑅 = , choose 𝑊 such maximizes the 𝑆𝑁𝑅 ⟹ 𝑊𝑜𝑝𝑡 =
𝜎𝑛 2 𝑊 𝐻 𝑊 𝑕

𝑃 2
𝑆𝑁𝑅 for MRC is : 𝑆𝑁𝑅 = 𝜎 2
𝑕
𝑛

BER with Diversity: Analysis of BER multiple antenna system:

𝑕1 𝑦1

𝑥
𝑦2
𝑊 𝑥
𝑕𝐿

𝑦𝐿

𝑃 2
At the reception after MRC receiver: 𝑆𝑁𝑅 = 𝜎 2
𝑕
𝑛

+∞ 𝑃
Average BER: 𝐵𝐸𝑅 = 𝐸 𝑄 𝑔𝑠𝑛𝑟 = 𝑄 𝑔𝑠𝑛𝑟 𝑓𝐺 (𝑔)𝑑𝑔: 𝑠𝑛𝑟 =
0 𝜎𝑛 2

2 2
𝑔= 𝑕 = 𝑕1 + ⋯ + 𝑕𝐿 2 ; (𝑔) is a Chi-Squared random variable with 2𝐿 degree of freedom

11
 1
𝑓𝐺 (𝑔) is distribution of gain (𝑔) with 𝑓𝐺 𝑔 = 𝑔𝐿−1 𝑒 −𝑔
𝐿−1 !

1−𝜆 𝐿 𝐿−1 𝐿+𝑙−1 1+𝜆 𝑙 𝑛! 𝑆𝑁𝑅

𝐵𝐸𝑅 = 𝑙=0 𝐶𝑙 ; where: 𝐶𝑘𝑛 = 𝑘! 𝑛−𝑘 ! and 𝜆 = ; 𝐿: received Antenna
2 2 2+𝑆𝑁𝑅

1−𝜆 1
≈ 1 1 𝐿
2 2 𝑆𝑁𝑅
At the high 𝑆𝑁𝑅 → 1+𝜆
; 𝐵𝐸𝑅 = 𝐶𝐿2𝐿−1 2𝐿 𝑆𝑁𝑅
≈1
2

1 1 1
If: 𝐿 = 1, 𝐵𝐸𝑅 𝛼 ;𝐿 = 2, 𝐵𝐸𝑅 𝛼 2
; 𝐿 = 3, 𝐵𝐸𝑅 𝛼
𝑆𝑁𝑅 𝑆𝑁𝑅 𝑆𝑁𝑅 3

So as the number of received antennas 𝐿 is increasing, the 𝐵𝐸𝑅 decreases at much faster

Spatial Diversity and Diversity order:

Multiple antenna diversity is also known as spatial diversity; For independent channel across

received antennas, the requires spacing is 𝜆/2 (wave length).

𝜆
Ex: for GSM: 𝜆 = 33,3 𝑐𝑚, 2 = 16,66 for phone set, it’s impossible

𝜆
For 3G, 𝑓𝑐 = 2,3𝐺𝑕𝑧 → 2 = 6,5 𝑐𝑚 for phone set, it’s possible to place multiple antennas

12
Diversity order: The BER in wireless system is given as: 𝑃𝑒 (𝑆𝑁𝑅)

𝑙𝑜𝑔 𝑃𝑒 (𝑆𝑁𝑅)
Diversity order 𝑑 = − lim𝑆𝑁𝑅→∞ 𝑙𝑜𝑔 𝑆𝑁𝑅

Diversity order is approximately related to the number of independent channel in the system.

1 1 𝐿
Ex: For 𝐿 antennas: and high SNR (𝑆𝑁𝑅 → ∞); 𝑃𝑒 = 𝐶𝐿2𝐿−1 2𝐿 𝑆𝑁𝑅

𝐿
1 1
𝑙𝑜𝑔 𝑃𝑒 𝑆𝑁𝑅 𝑙𝑜𝑔 𝐶𝐿2𝐿−1
2𝐿 𝑆𝑁𝑅
𝑑 = − lim = 𝑑 = − lim
𝑆𝑁𝑅→∞ 𝑙𝑜𝑔 𝑆𝑁𝑅 𝑆𝑁𝑅→∞ 𝑙𝑜𝑔 𝑆𝑁𝑅

1 1
𝐿 − log 𝑆𝑁𝑅 − log 𝐶𝐿2𝐿−1 (𝐶𝐿2𝐿−1
log⁡ )
= lim 2𝐿 = lim 𝐿− 2𝐿
𝑆𝑁𝑅→∞ 𝑙𝑜𝑔 𝑆𝑁𝑅 𝑆𝑁𝑅→∞ 𝑙𝑜𝑔 𝑆𝑁𝑅

1
(𝐶𝐿2𝐿−1
log⁡ )
= 𝐿 − lim 2𝐿 ≈ 𝐿 + 0 = 𝐿
𝑆𝑁𝑅→∞ 𝑙𝑜𝑔 𝑆𝑁𝑅

So for 𝑆𝑁𝑅 → ∞, 𝑑 = 𝐿

-Wireless channel and Delay Spread (Etalement temporel):

𝑕 𝜏
𝐿−1

𝑕(𝜏) = 𝑎 𝑖 𝛿 𝜏 − 𝜏𝑖
𝑖=0 𝑎0 𝑎1 𝑎𝐿−1

Signal
𝜏
Attenuation Component 𝜏0 𝜏1 𝜏𝐿−1

Max Delay Spread

𝛿𝜏

13
-Maximum Delay Spread:

𝛿𝜏 = 𝜏𝐿−1 − 𝜏0
-Average Delay Spread:
𝐿−1 2
𝑖=0 𝑎 𝑖 𝜏 𝑖
𝜏= 𝐿−1 𝑎 2 ;
𝑖=0 𝑖

2
𝑎𝑖 is power (gain) of 𝑖 𝑡𝑕 path

- Root Mean Squared Delay Spread (RMS delay spread) : is a delay of significant power

𝐿−1 2 2
𝑖=0 𝑎𝑖 𝜏𝑖 − 𝜏
𝜎𝜏 = 𝐿−1
𝑖=0 𝑎𝑖 2

𝜏
𝜎𝜏

Significant Power Lower Power

14
-Average RMS delay spread:

Average Power profile at delay 𝜏

∞ 2
2 Φ 𝜏 =E h τ
𝜎𝜏 = 𝜏𝑖 − 𝜏 𝑓 𝜏 𝑑𝜏
0

Deviation, Φ(𝜏)
Frational Power 𝑓 𝜏 = ∞
τ also equal to at delay 𝜏 0
Φ(𝜏) 𝑑𝜏
∞
𝜏= 𝜏𝑓 𝜏 𝑑𝜏 Total Power
0

-Coherence Bandwidth of wireless Channel 𝐵𝑐 :

Coherence Bandwidth of wireless Channel 𝐵𝑐 is the bandwidth where the channel response is

considered as constant.

15
The relation between coherence bandwidth and RMS delay is:
1
𝐵𝑐 =
2 𝜎𝜏

In time domain:

𝑇𝑠
𝑠0 𝑠2 𝑠4
𝑠1 𝑠3 LOS (Direct Path)

𝑇𝑠 ≫ 𝜎𝜏 ⟹ ∄ 𝐼𝐸𝑆
𝜎𝜏 𝑠0 𝑠2 𝑠4
𝑠1 𝑠3 NLOS (Non Direct Path)

𝑇𝑠
𝑠0 𝑠2 𝑠4
𝑠1 𝑠3 LOS (Direct Path)

𝑇𝑠 ≪ 𝜎𝜏 ⟹ ∃ 𝐼𝐸𝑆
𝜎𝜏 𝑠0 𝑠2 𝑠4
𝑠1 𝑠3 NLOS (Non Direct Path)

16
 II.2- Invariant and variant channels

-Inter Symbols Interference and Doppler in wireless communication:

Doppler: Doppler shift is the change in the frequency of the electromagnetic wave due to
the relative movement between transmitter and receiver.

Doppler shift:
𝑣 cos θ
𝑓𝑑 = 𝑓𝑐
𝑐
𝑣 is speed of mobile

𝑐 is speed of light

𝑓𝑐 is frequency carrier
𝑣 cos θ
Received frequency: 𝑓𝑟 = 𝑓𝑐 + 𝑓𝑑 = 1 + 𝑓𝑐
𝑐

Time selectivity (Time varying channel):

17
 𝒗

𝜽=𝟎

After 𝑡 , distance decrease by 𝑣 × 𝑡

𝑣𝑡
Delay after 𝑡 𝜏𝑖 (𝑡) = 𝜏0 − 𝑐 ; 𝜏0 is initial delay

𝑣𝑡 𝑡
𝜏 𝑖 𝑡 = 𝜏0 − 𝑐𝑜𝑠 𝜃 = 𝜏0 − 𝑓𝑑 ; 𝜏𝑖 𝑡 is not constant
𝑐 𝑓𝑐

𝐿−1 −2𝑗𝜋 𝑓𝑐 𝜏 𝑖 𝑡
Fading coefficient: 𝑕 𝑡 = 𝑖=0 𝑎𝑖 𝑒

18
 𝐿−1

𝑕 𝑡 = 𝑎𝑖 𝑒 −2𝑗𝜋 𝑓𝑐 𝜏 𝑖 × 𝑒 2𝑗𝜋 𝑓 𝑑 𝑡
𝑖=0

Time varying phase

Time varying channel

- Gain of a time selective channel

If : 𝑡 = 0 ⟹ 𝑕 0 = 𝑎𝑖 𝑒−2𝑗𝜋𝑓𝑐𝜏𝑖

1
If : 𝑐 ⟹ 𝑕 = 𝑗𝑎𝑖 𝑒−2𝑗𝜋𝑓𝑐 𝜏𝑖 = 𝑗 𝑕(0);
4𝑓 𝑑

1
𝑕(𝑡) is changing dramatically (from real to imaginy) at
4𝑓 𝑑

Coherent Time: 𝑇𝑐 is the time over which the channel is approximately constant
1 1
𝑇𝑐 = = ; 𝐵𝑑 is Doppler Spread ; 𝐵𝑑 =2𝑓𝑑
4𝑓 𝑑 2𝐵𝑑

19
 1 1 1 1
4𝑓𝑑 4𝑓𝑑 4𝑓𝑑 4𝑓𝑑

0 1 1 3
4𝑓𝑑 2𝑓𝑑 4𝑓𝑑

𝑇𝑐 𝑇𝑐 𝑇𝑐

-Doppler Spectrum: Doppler spectrum gives an intuition into rate of change of the wireless
channel.

Spectrum= TF [Correlation]

𝑆𝐻 𝑓 = 𝑇𝐹 Ψ Δ𝑡 = 𝑇𝐹 E 𝑎𝑖 𝑡 𝑎𝑖 ∗ t + Δ𝑡

𝑎𝑖 𝑡 = 𝑎𝑖 𝑒 −2𝑗𝜋 𝑓 𝑑 𝜏 𝑖 𝑒 2𝑗𝜋 𝑓 𝑑 𝑡 ; 𝑎𝑖 ∗ t + Δ𝑡 = 𝑎𝑖 𝑒 2𝑗𝜋 𝑓 𝑑 𝜏 𝑖 𝑒 −2𝑗𝜋 𝑓 𝑑 t+Δ𝑡

𝑚𝑎𝑥 𝑣
Ψ Δ𝑡 = E 𝑒 −2𝑗𝜋 𝑓 𝑑 t+Δ𝑡
if 𝑎𝑖 2
= 1 ⟹ Ψ Δ𝑡 = E 𝑒 −2𝑗𝜋 𝑓𝑑 𝑐𝑜𝑠 𝜃 Δ𝑡
; 𝑓𝑑𝑚𝑎𝑥 = 𝑓𝑑 𝑐

𝜋1 𝑚𝑎𝑥 1
Ψ Δ𝑡 = 0 𝜋
𝑒 −2𝑗𝜋 𝑓𝑑 𝑐𝑜𝑠 𝜃 Δ𝑡
𝑑𝜃 ; 0 ≤ 𝜃 ≤ 𝜋 ; (𝜃 is uniform distributed ; 𝑓𝜃 = 𝜋 );

Ψ Δ𝑡 = J0 2𝜋𝑓𝑑𝑚𝑎𝑥 Δ𝑡 ; J0 is Bessel function of the 0𝑡𝑕 order.

20
 Ψ Δ𝑡

Δ𝑡 = 𝑇𝑐

𝟎. 𝟓

Δ𝑡

Doppler spectrum:
𝑓
+∞ 1 𝑟𝑒𝑐𝑡 2𝑓 𝑑
𝑆𝐻 𝑓 = −∞
Ψ Δ𝑡 𝑒−2𝑗𝜋𝑓 Δ𝑡 𝑑 Δ𝑡 ⟹ 𝑆𝐻 𝑓 = 𝜋 𝑓𝑚𝑎𝑥
𝑑 2
𝑓
1− 𝑓𝑑

𝑺𝑯 (𝒇)

𝟐𝒇𝒅

−𝒇𝒅 𝒇𝒅
Jakes Model
21
Slow fading:

If Coherence Time (Tc) > Inter-Channel estimation Time

Tc

Channel Estimation Channel Estimation

Test 1
Test 2

26 Symbols of
Data Symbols Training
Sequence Data Symbols

156 Symbols

GSM Channel estimation

Fast fading:

If Coherence Time (Tc) < Inter-Channel estimation Time

Tc

Channel Estimation
Channel Estimation
Test 1 Test 2

22
 II.3- Gaussian, Rice fading

Gaussian channel:

When the channel does not vary over time and there is no direct path between the transmitter
and receiver, the real and imaginary parts of 𝒉 follow a Gaussian distribution.

Rice channel:

If the environment allows a direct path between the transmitter and the receiver, the channel then
includes a deterministic component 𝒉 and a random 𝒉:

𝒉=𝒉+𝒉

The Rice 𝑲𝒓𝒊𝒄𝒆 coefficient is given by the ratio between the power of the component deterministic
and that of the random component

𝟐
𝒉
𝑲𝒓𝒊𝒄𝒆 =
𝟐𝝈

With 𝝈 the standard deviation of the random component 𝒉. The module of 𝒉 will then follow a Rice
distribution:

2 2
𝒉 − 𝒉 +𝑎 𝒉𝑎
𝑝 𝒉 = 2 𝑒 2𝜎 2 𝐼0
𝜎 𝜎2

Where 𝑎 is the amplitude of the deterministic component and 𝐼0 is the Bessel function of the first
kind and of zero order. We notice that if 𝑎 is null, we obtain the Rayleigh distribution and if 𝑎
is great, we find a BBAG channel

23
 Temporal fading of a Rayleigh channel

Temporal fading of a rice channel with a 𝐾𝑟𝑖𝑐𝑒 = 10 𝑑𝐵

24
 𝐵𝑐

𝑇𝑐

Summary of the different selectivity of channel

25