B03 WCDMA Capacity Dimensioningxxxxxxx.ppt

B03 WCDMA Capacity Dimensioningxxxxxxx.ppt

1. March 7, 2025
WCDMA Capacity Dimensioning

2. Foreword
WCDMA is intrinsical Interference limited system
Coverage and capacity depend on the interference
WCDMA is a Soft Capacity system

3. Outline
 Radio Dimensioning Procedure
 Radio Link Budget
 Capacity Dimensioning
 Appendix
Stochastic Knapsack: Blocking Probability
Campbell’s Dimensioning Process
Interference Analysis

4. Capacity
Quality Coverage
Dependence among Capacity, Coverage and Quality
Capacity-Coverage-QualityCapacity-Coverage-Quality

5. Capacity vs. Coverage
― Cell Load   Coverage Range
― Cell Load   Subscribers
Capacity vs. Quality
― BLER   Capacity
― GoS   Capacity
Coverage vs. Quality
― BLER   Coverage Range
Capacit
y
Quality Coverage
Interference
Capacity-Coverage-QualityCapacity-Coverage-Quality

6. Independence among Capacity, Coverage and Quality of GSM
System
―Capacity: Timeslots and Carriers available and Reuse Mode
―Coverage Range: transmission Power on Uplink/Downlink (Link
Balance)
―Quality of Call: be ensured by network design to minimize interference
In GSM system, capacity, coverage and quality requirements can
be met by independently analysis and design
Frequency Planning is a crucial issue to GSM system
Capacity-Coverage-QualityCapacity-Coverage-Quality

7. Dependency among Capacity, Coverage and
Quality of WCDMA System
―WCDMA system is interference-limited.
―Capacity vs. Coverage
•Increase intended system loading will offer more
capacity while increasing intra-cell interference and
thus reduce coverage range (Application: Cell
breathing)
―Capacity vs. Quality
•System capacity can be achieved by relaxing quality
requirement for some connections (Application:
Reduce BLER target value by outer-loop power
control)
―Coverage vs. Quality
•Coverage range can be expanded by relaxing quality
requirement for some connections (Application: Slow
down data speed by AMRC to accommodate large
path loss)
Capacity
Quality
Coverage
Interference
Interference is the dominant
concern in capacity analysis
Capacity-Coverage-QualityCapacity-Coverage-Quality

8. Radio Dimensioning ProcedureRadio Dimensioning Procedure
• Network dimensioning is an iterative process
• Downlink analysis checks whether NodeB power is enough to cover the users

9. Outline
 Radio Dimensioning Procedure
 Radio Link Budget
 Capacity Dimensioning
 Appendix
Stochastic Knapsack: Blocking Probability
Campbell’s Dimensioning Process
Interference Analysis

10. Radio Link Budget: Purpose
Calculate the Maximum Path Loss
―EIRP
―Sensitivity of Receiver
―Minimum Required Signal Strength
Calculate the Cell Range
―Propagation Model
―Maximum Path Loss
―Antenna Height
―Carrier Frequency

11. rate data User
bandwidth Spreading
PG where
PG
1
x
No
Eb

No
Ec


BS Antenna Gain
Rx&Tx Cable Loss
P
ro
p
a
g
a
tio
n
L
o
s
s




Body Loss

Penetration Loss











No
Ec
NF N S thBS
Radio Link Budget: Sketch
TRX

12. EIRP
Sensitivity of Receiver
Minimum Signal Strength
Edge Coverage Probability
Propagation Model
Margin
Gain
Loss
Radio Link Budget: Important Parameters

13. Radio Link Budget: Margin, Gain, Loss
Margins
―Interference Margin
―Slow Fading Margin
―Fast Fading Margin
Gains
―Antenna Gain
―SHO Gain
Loss
―Body Loss
―Cable Loss
―Penetration Loss

14. Outline
 Radio Dimensioning Procedure
 Radio Link Budget
 Capacity Dimensioning
 Appendix
Stochastic Knapsack: Blocking Probability
Campbell’s Dimensioning Process
Interference Analysis

15. Capacity Dimensioning: Purpose
 Estimate Supported Subscribers
― Cell Resource
― Mixed Services
― Service Traffic
― Respective GoS
Estimate Site Number
― Site Number (Coverage)
― Site Number (Capacity)

16. Capacity Dimensioning: Difficulties
 Cell edge is continuously moving a
ccording to the traffic load
Mixed services: multiple data rates
Respective GoS Requirement

17. Capacity Dimensioning: main methods
 Campbell’s Theorem
 Stochastic Knapsack
 Fractional Load

18. Stochastic Knapsack: What is it ?
What is Knapsack ?
What is in the Knapsack ?
Knapsack for Uplink = ?
Knapsack for Downlink = ?

19. Stochastic Knapsack: Where is it from ?
 a Multi Service Traffic Model
 Used in ATM Multiplexer Dimensioning
 Simulate the Respective GoS of the Sup
ported Services

20. Stochastic Knapsack: Introduced into WCDMA
 Blocking Probabilities
 Shared Resource
 Simulate actual Traffic Behavior
 Uplink Resource: Cell Load (why)
 Downlink Resource: Power (why)
Modified to WCDMA Air Interface Dimensioning (why)

21.  Resource Shared
Time
C
o
n
s
u
m
e
d

R
e
s
o
u
r
c
e
Stochastic Knapsack: Resource Shared
Capacity: C
services

22. Stochastic Knapsack: Example
Users States for 2 services
Video Phone Call
Voice Call
• Knapsack?
• What is the user state in the
Knapsack?
• Which call will be blocked?
• Which call can access the Knapsack?

23. Stochastic Knapsack: Example
n1
n2
C

• the user state if not calls arrive?
• the user state if a voice call access?
• the user state if a video phone call access?
Voice
V
id
e
o

P
h
o
n
e
Users States for 2 services

24. n1
n2C
Cb1-
Two Services：Blocking Prob. for Service
1
Stochastic Knapsack: Example
Voice
V
id
e
o

P
h
o
n
e
the 4 states will be blocked for voice service.
Why?

25. n1
n2C
Cb2-
Two Services：Blocking Prob. for Service
2
Stochastic Knapsack: Example
Voice
V
id
e
o

P
h
o
n
e
the 7 states will be blocked for video phone service. Why?

26.  What is the advantage?
 What is the disadvantage?
Stochastic Knapsack: Questions

27. Campbell’s Theorem
Virtual Service
Video Phone Call
Voice Call

28. Campbell’s Theorem
 Multi services  one Virtual Service
 Virtual Service Load
 Virtual Service Traffic
 How to Calculate? (Appendix)
 One Service Calculation
 Erlang B Formula

29. Campbell’s Theorem
 What is the advantage?
 What is the disadvantage?

30. Fractional Load
 Traffic for each service
 Traffic/BH/Sub
 Supported Subscribers
 Channels needed for each service
 GoS requirement
 Erlang B
 Fractional Load for each service
 single link load
 channels
 Cell Load for all services
 accumulate all the fractional load

31. Fractional Load
Time
C
o
n
s
u
m
e
d

R
e
s
o
u
r
c
e

32.  What is the advantage?
 What is the disadvantage?
Fractional Load

33.  Stochastic Knapsack
Comparison of the Methods
 Complicated
 actual traffic behavior
 resource shared
 respective GoS for each service
 only one GoS for all services
 can guarantee all the GoS requirements?
 resource shared
 easy to calculate
 Fractional Load
 Campbell’s Theorem
 resource not shared
 easy to calculate

34.  only one service
 the same result (why)
 Fractional Load
 pessimistic
 more NodeB sites
 Campbell’s Theorem
 uncertain
 optimistic
e.g. 2% GoS for all services
 Stochastic Knapsack
 reasonable
Comparison of the Methods
Dimensioning Result:

35. Contents
Uplink capacity analysis
Downlink capacity analysis

36. Uplink capacity analysis
Single CS service
Single PS service
Mixed services

37. Single CS service
To single CS service, the uplink total received power
in BS can be calculated as:
0 0
1
(1 )
k
i b b
i
I w N w vE R f

  
meet Poisson arrival, and the mean value is:
1
k
i
i
v





38. 1
0
1 1
1
(1 )
1
K
i b b
i
v E R f
IW




 


0 0
0
0
1
1
1
(1 )
K
i b b
i
I IW
N
IW v E R f


 

  
Then：
So：
0
1
/ (1 )
/
K
i b
i
b
v E I f
W R


 


Single CS service

39. 1 0
/
(1 ) /
K
b
i
i b
W R
v
f E I




 

1 0
/
1 /
K
b
S i
i b
W R
N v
f E I


  


The number of uplink channel supported by system with co
rresponding η uplink loading is:
Single CS service

40. Defining:
Pblocking
/
N
/N!
n0
N
/u
n/n!
Then the soft-blocking formula based on interference of uplink is：
Single CS service
S
N N  

41. Uplink capacity analysis
Single CS service
Single PS service
Mixed services

42. Single PS service
• PS service model：

43. Then the soft-blocking formula based on interference
of uplink is：
0
/
1 /
b
s
b
W R
N
f E I


 
 
Single PS service

44. Single PS service
Because the blocking characteristic of PS service is determine
d by the acceptable delay, according to ErlangC formula, defini
ng the channel number: ，
the probability of call with delay can be calculated as:
So the probability with delay exceeding t(s) of any call is:





1
0
)!/()/1(!
]Pr[
N
k
NN
N
kANANA
A
delayed
t
H
AN
edelayedtdelay

 ]Pr[]Pr[


A
S
N N  

45. Single PS service
Mean delay is:
Mean throughput of Uplink is：
Here:A is the supported total traffic, N is the channel num
ber, H is the average duration per call of the service.
AN
H
delayedD

 ]Pr[
b
S R A 

46. Uplink capacity analysis
Single CS service
Single PS service
Mixed services

47. First, we assume these following variables :
the user number is X,
the number of service type is M,
the ration of other-cell to own-cell interference is f,
η is the cell loading.
mixed service

48. The traffic of specific service can be calculated as:
)
3600
Pr(_ 


ctorActivityFaBearerRate
Throughput
etrationServicePenoportionServiceErlangData
The loading factor of one service per user can be
calculated as:
0
1
/
1
( / )
i
b
b i i
A
W R
E N




mixed service

49. We convert all services to one virtual service and introduce
two variables C1 and C2:
( )
1
1 ( )
M
voice data i
i
C A X voiceErl A X dataErl i

      
2 2
( )
1
2 ( ) ( ) ( )
M
voice data i
i
C A X voiceErl A X dataErl i

      
And the loading factor of per virtual service is:
1
2
C
C
A
virtual

mixed service

50. Then the total traffic of virtual service that the cell can support
is:
virtual
A
C
VirtualErl
1

here：
voiceErl:traffic of voice service
)(idataErl
:traffic of the ith type data servic
e
VirtualErl:traffic of the virtual service
mixed service

51. According to the preconcerted cell loading η, the number
of virtual service channel N is:
(1 )
virtual
N Ns
f A
 
    
 
 
Based on ErlangB formulary:
Then:
From
virtual
A
C
VirtualErl
1
 we can get the number of users X.
mixed service


0
!
!
N
blocking N n
n
N
P
n
 
 



VirtualErl




52. Then we can get the uplink data throughput rate per
carry as following formulary:
 
i
iii
ctorActivityFaBearerRateErlangXCarrieroughputPerAverageThr
mixed service

53. Contents
Uplink capacity analysis
Downlink capacity analysis

54. Downlink capacity analysis
Single CS service
Single PS service
Mixed services

55. Single CS service
The total received interference of a specific UE is:
I0WN0WP1i
j2
M
Pji
here,is non-orthogonal factor, Pji
Eb
I0

iP1i
N0WP1i
j2
M
Pji

W
Rb
ith UE from jth site. Assuming the ratio of the dedicated power for
i
Then:
So:i
Eb/I0
W/Rb

j2
M
Pji
P1i

N0W
P1i

is the total received power of
UEi to total power is:

56. Defining:NoiseRise
N0WP1ij2
M
Pji
N0W
1
1
the ratio of other-cell interference is f :f
j2
M
Pji
P1i
i
Eb/I0
W/Rb

f
then:
The following equation must be satisfied:

i1
k
viicPcongestion
is the power of common channels
Pcongestion
c
Single CS service
is the corresponding power of RRM congestion threshold.

57. Then: NsPcongestionc
W/Rb
Eb/I0


f

Defining channel number:
NNs
Pblocking
/
N
/N!
n0
N
/u
n
/n!
Then we can get the downlink soft block traffic （soft block E
rlang）,and the downlink throughput rate per carry is:
From formula:
ctorActivityFaBearerRaterlangSoftBlockECarrieroughputPerAverageThr 
Single CS service

58. Downlink capacity analysis
Single CS service
Single PS service
Mixed services

59. NsPcongestionc
W/Rb
Eb/I0


f

The method of downlink single PS service capacity analysis is
similar with that of uplink:
Single PS service
NNs


A
Defining channel number:
and:

60. Single PS service





1
0
)!/()/1(!
]Pr[
N
k
NN
N
kANANA
A
delayed
t
H
AN
edelayedtdelay

 ]Pr[]Pr[
the probability of call with delay can be calculated by:
So the probability with delay exceeding t(s) of any call is:

61. Single PS service
AN
H
delayedD

 ]Pr[


RcS
Mean delay is:
Mean throughput of Uplink is：
Here: A is the supported total traffic, N is the channel num
ber, H is the average duration per call of the service.

62. Downlink capacity analysis
Single CS service
Single PS service
Mixed services

63. mixed service
First, we assume these following variables :
the user number is X,
the number of service type is M,
θis the non-orthogonal factor,
the ration of other-cell to own-cell interference is f,
η is the downlink cell loading.

64. The traffic of specific service can be calculated as:
)
3600
Pr(_ 


ctorActivityFaBearerRate
Throughput
etrationServicePenoportionServiceErlangData
The loading factor of one service per user can be
calculated as:

mixed service
iib
i
NE
RW
A


)/(
/
1
0

65. 


m
i
idatavoice
idataErlXAvoiceErlXAC
1
)(
)(1



m
i
idatavoice
idataErlXAvoiceErlXAC
1
2
)(
2
)()()(2
Defining:
1
2
C
C
A
virtual
virtualA
C
VirtualErl
1

mixed service
And the loading factor of per virtual service is:
Then the total traffic of virtual service that the cell can support
is:

66. According to the preconcerted cell loading η, the number
of virtual service channel N is:
Based on ErlangB formulary:
Then:
From
virtual
A
C
VirtualErl
1
 we can get the number of users X.
mixed service
]
)()1(
[][
virtualc Af
NsN




Pblocking
/
N
/N!
n0
N
/u
n
/n!


VirtualErl

67. Then we can get the downlink data throughput rate
per carry as following formulary:
 
i
iii
ctorActivityFaBearerRateErlangXCarrieroughputPerAverageThr
mixed service

68. According to the demand of traffic and terrain
characteristic, distinguish the planning region into
different areas, like as dense urban, urban, suburban,
rural and so on.
Different propagation models for areas.
Different user numbers and traffic models for each area.
The demands of QOS and GOS for each service and
area.
The carry demand for each area, one, two or more
carriers.

Capacity dimension input Capacity dimension input

69. According to the above inputs of capacity dimension, we
can get the number of site and site configuration based
on capacity demands.
Comparing the dimension result of capacity with that of
coverage, the limited result is proposed .
Capacity dimensionCapacity dimension

70.  a Rough Dimensioning Result
Rough Dimensioning Result
 a flat and homogenous landscape
 Digital database (Heights, Clutters, Vectors)
 Further Simulation
 homogeneous interference
 the propagation model
 simple propagation laws
 the actual traffic behavior
 regular hexagon pattern
 traffic growth expectation
 uniform traffic demand
 Simulation Tool
 Enterprise, etc

71. Outline
 Radio Dimensioning Procedure
 Radio Link Budget
 Capacity Dimensioning
 Appendix:
Stochastic Knapsack: Blocking Probability
Campbell’s Dimensioning Process
Interference Analysis

72. !n
a
)(G
!n
a
!n
a
!n
a
)(G)(
k
n
k
K
k
K
n
K
nn
kK
1
1
2
2
1
11
21


 n



n K
n
K
nn
!n
a
!n
a
!n
a
)(G
K

2
2
1
1
21
State Probability:
Stochastic Knapsack: Blocking Probability
k: the traffic of service k
kn: the connecting users of service k

73. Blocking Probabilities:
)(G
)(G
)n(B
k



k
Bn
B
k

Stochastic Knapsack: Blocking Probability
: the blocking probability for service k
k
B
: the blocked state for service kBk

74. Campbell’s Theorem




j
jj
j
jj
virtual
Erlang.A
Erlang.A
A
2
Virtual Load
virtual
j
jj
virtual
A
Erlang.A
Erlang

＝Virtual Traffic
: the load of a single user for service jjA
: busy hour traffic of a single user for service j
jErlang

75. Campbell’s Theorem
Virtual service channels
Total Traffic: using Erlang B Formula









virtual
UL
Af
N
.1

UL
















f
M
A
N
virtualDL
c


.
1
.1
DL

76. Campbell’s Theorem
Supported Subscribers in the cell
virtual
CelltheinTraffic
Erlang
Total

77. Interference Analysis
 Uplink Interference Analysis
 Downlink Interference Analysis

78. NotherownTOT PIII 
Uplink Interference Analysis
I
own
： interference caused by users of own cell
I
other：interference caused by users of other cells
P
N： equivalent noise input of the receiver

79.  Noise power of receiver: P
N
P
N = 10lg(KTW) ＋ NF
―K: Boltzmann Constant，= 1.38×10
-23
J/K
―T: temperature in degrees Kelvin
―W: Bandwidth of signal，3.84MHz for WCDMA
―NF: Noise figure of receiver
And
― 10lg(KTW) = -108dBm/3.84MHz
― NF = 3dB (typical value for Marco-cell)
― P
N
= 10lg(KTW) + NF = -105dBm/3.84MHz
Uplink Interference Analysis

80.  I
own : Own-cell interference
―Interference should be overcome by each user: I
TOT
- P
j
•P
j : desired signal power from user j received by NodeB
―With perfect power control:
―P
j can be estimated by：
―Own-cell interference: totally received power from all users
of own cell:
jjjTOT
j
j
vR
W
PI
P
EbNo
1



jjj
TOT
j
vR
W
EbNo
I
P
11
1 


N
jown
PI
1
Uplink Interference Analysis

81.  I
other : Other-cell interference
―Difficult to analyze theoretically, and depends on user
distribution, cell position, antenna patterns and so on
―Definition of i, the ratio of other-cell to own-cell
interference
own
other
I
I
i
Uplink Interference Analysis

82. 

N
N
jjj
TOT
N
N
j
NownNotherownTOT
P
vR
W
EbNo
I
i
PPiPIiPIII






1
1
11
1
1
1)1(
Define:
jjj
j
vR
W
EbNo
L
11
1
1


The total interference can be estimated:

N
N
jTOTTOT PLiII  
1
1
Uplink Interference Analysis

83. Then:


N
j
NTOT
Li
PI
1
11
1
Defining uplink load factor:
I
TOT reaches infinity while load factor equals 1
 


N
jjj
N
jUL
vR
W
EbNo
iLi
11 11
1
1
11
Uplink Interference Analysis

84. Noise Figure defined as follows:
UL
N
j
N
TOT
L
P
I
NoiseRise





1
1
1
1
1
50% load  3dB
60% load  4dB
75% load  6dB
Uplink Interference Analysis

85. Interference Analysis
Uplink Interference Analysis
Downlink Interference Analysis

86. NotherownTOT PIII 
I
own: Interference from BS of own cell
I
other: Interference form BSs of other cells
P
N: Equivalent noise input of the receiver
Downlink Interference Analysis

87. I
own
: Own-cell interference
―Individual channel distinguished by orthogonal OVSF code. The orthog
onality between channels can be achieved in static propagation environ
ment without multipath. Then there is no interference over each other i
n downlink.
―In multipath environment, not all paths of signal transmitted for a chann
el can be applied by RAKE and some energy adds to interference. It ca
n be modeled by the introduction of orthogonal factor :
―in the formula above, P
T is the total power transmitted by BS, in
cluding power of common channels and dedicated channels

j
T
jown
PL
P
jI 1)(

N
DCHCCHT jPPP
1
)(
Downlink Interference Analysis

88. I
other
: Other-cell interference
―Signals transmitted by BSs of other cells can cause
interference over the target cell. Due to different
scrambling codes, these signals are not orthogonal with
those of the target cell.
―Assuming uniformly distributed service and equal
powers transmitted by all BSs, if there are K of other
cells and the path loss from Kth BS to user j is PL
k,j,
then:

K
jk
Tother
PL
PjI
1
,
1
)(
Downlink Interference Analysis

89. 
N
K
jk
T
j
T
j
NotherownTOT
P
PL
P
PL
P
PIII



1 ,
1
1
With perfect power control, there is
jjTOT
jDCH
vR
W
jI
PLjP
jEbNo
1
)(
/)(
)( 
The required transmission power of DCH for user j is
jTOTj
j
jDCH
PLjIv
W
R
EbNojP  )()(
Downlink Interference Analysis

90. Since

N
DCHCCHT
jPPP
1
)(
Total transmission power can be estimated as follows:












































jN
K
jk
j
TTj
N
j
j
CCH
N
K
jk
T
j
T
j
N
jj
j
CCH
N
jTOTj
j
CCHT
PLP
PL
PL
PPv
W
R
jEbNoP
P
PL
P
PL
P
PLv
W
R
jEbNoP
PLjIv
W
R
jEbNoPP
1
,
1
1
,
1
1
1)(
1
1)(
)()(


Downlink Interference Analysis

91. P
T can be resolved as follows:

 
















N
j
j
jj
N
jj
j
NCCH
T
v
W
R
jEbNoi
PLv
W
R
jEbNoPP
P
1
1
)(11
)(

i
j is the ratio of other-cell to own-cell interference for
user j. and it is defined as follows:

K
jk
j
j
PL
PL
i
1 ,
Downlink Interference Analysis

92. Downlink load factor
―defined in common as the ratio of total transmission
power to maximal transmission power of the BS.
―The ratio of P
CCH to P
MAX is about 20%.
MAX
j
DCH
MAX
CCH
MAX
T
DL
P
jP
P
P
P
P


)(

Downlink Interference Analysis