Cellular Network Planning and Optimization Part9 PDF
Cellular Network Planning and Optimization Part9 PDF
Cellular Network Planning and Optimization Part9 PDF
General
Computation of theoretical WCDMA load is straightforward in uplink and downlink Load equations can be used to make semi-analytic coverage and capacity estimates. Load equations are semi-analytic since link level SIR performance needs to be simulated while system level performance is obtained analytically. As such load equations are very useful for quick system level evaluations. Load equations are also interesting since they provide a explicit mapping from link level performance to system level performance and link level results are usually more easily available than system level results.
2
r (t ) =
N own n =1
(t ) + z n (t ) +n(t )
n =1
N other
where first sum refers to signals that are coming from users that are connected to the considered cell, second sum refers to signals that are coming from users that are connected to other cells and last term refers to AWGN noise
3
E r (t )
}= I
n
total
= I own + I other + PN
E{y (t )
}, I
other
N other n=1
E{z (t )
n
Remark: Here E{ . } refers to expectation. Exercise: Go through all intermediate steps between (1) and (2)
4
I total NR = PN
our goal is to deduce a formula (load equation) that provides a connection between noise rise and link level parameters.
Hence, the minimum requirement for Eb/No is defined by the processing gain + power that is needed to overcome the interference from other users. Eb/No is an important variable since it actually maps the link level performance to the system level performance.
7
(Eb / N 0 ) j
Pj W = j R j I total Pj
where
W = System chip rate Pj = Signal power of user j
j =
Pj I total
1 j = 1 + W ( j R j (Eb / N 0 ) j )
(7) own
1 = j = j =1 j =1 1 + W ( j R j ( Eb / N 0 ) j )
Note: If there are e.g. 2 services used in the cell, then load equation is of the form
own =
N1 N2 + 1 + W ( 1 R1 (E b / N 0 )1 ) 1 + W ( 2 R2 (E b / N 0 )2 )
10
(9)
= (1 + )own
11
I total = (1 + ) Pj + PN = I total + PN
j =1
where last term is the AWGN noise power and we have used equation (2).
12
I total 1 NR = = PN 1
NRdB = 10 log(1 )
13
14
Activity factor :
Value 0.67 for speech (uplink recommendation) Value 1.0 for data
Bit rate R:
Depends on the service, usually up to 400-500 kbps
Give X-axis of the plot as a function of number of users. All sites in the network admit three sectors (cells) and user mobility is 3 km/h in the first and 120 km/h in the second case.
17
= (1 + )own
,
NRdB = 10 log(1 )
(Eb / N 0 )voice (3km / h) = 4.5dB, (Eb / N 0 )data (3km / h) = 1.5dB, (Eb / N 0 )voice (120km / h) = 5.5dB, (Eb / N 0 )data (120km / h) = 2.5dB, = 0.65
18
10
20
30 40 50 Number of users
60
70
80
12.2 kbps voice service (solid curve) and 128 kbps data service (dashed curve). Lower curves: 3km/h mobility, upper curves: 120 km/h mobility
19
20
W Pj R j Lm, j
1 (1 j ) P / Lm , j +
N cells n =1, n m
P/L
n, j
22
(1 j ) P / Lm, j
The second interference term contains interference coming N cells from other cells.
n =1, n m
P/L
n, j
23
Note: It is assumed that all base stations admit the same transmission power P
On orthogonality factor
In operational network, is continuously changing is estimated by base station based on UL multipath propagation According to experience, for typical WCDMA environments: 0,5 0,6 macro cells 0,8 0,9 micro cells (smaller cells, less multipath) Too optimistic can lead to coverage problems Too modest can lead to inefficient utilisation of DL performance
24
(Eb / N 0 ) j R j
W
j = 1,2,..., N own
We note that in downlink other-to-own cell interference is different for different users. We have N N
(15)
j =
n =1,n m
cells
P Lm , j P Ln , j
n =1,n m
cells
Lm , j Ln , j
25
P = P
j =1
N own
(Eb / N 0 ) j R j j
W
((1 j ) + j ) + PN
j =1
N own
(Eb / N 0 ) j R j j
W
Lm , j
From (16) we solve the required total base station transmission power
(17) P =
1
j =1
W N own (E / N ) R 0 j b j j
j =1
PN
N own
(Eb / N 0 ) j R j j
Lm , j
((1 j ) + j )
(Eb / N 0 ) j R j j
W
((1 j ) + j )
28
The factors that impact to the required transmission power in base station
AWGN noise (first term) Transmission power that is needed to serve own cell users (second term) Transmission power that is needed to overcome the interference (third term). Interference contains contribution from own cell (imperfect code orthogonality) and from other cells.
29
30
(20)
(Eb / N 0 ) j R j j
W
((1 ) + )
In ITU Vehicular A channel is round 0.5 In ITU Pedestrian A channel is round 0.9 In macro-cell deployment with omnidirectional antennas is round 0.55 In macro-cell deployment with 3-sector sites is round 0.65
31
(Eb / N 0 ) j R j j
W ((1 ) + )
(21)
PBS =
1
j =1
N own
(Eb / N 0 ) j R j j
W
Here N rf is the noise spectral density of the receiver front end. There holds
temperature in Kelvin and NF is receiver noise figure that is usually between 5 to 9 dB.
33
Example
Plot maximum allowed path loss as a function of number of users for 12.2 kbps voice service and for 64 kbps data service when
BS transmission power is 40W BS transmission power is 10W
All sites in the network admit three sectors (cells), users mean mobility is 3 km/h, average orthogonality factor is 0.5, average mobile noise figure is 7dB.
34
Example
Solution. We solve mean path loss from equation N own ( Eb / N 0 ) j R j j N rf W L N rf L N own (Eb / N 0 ) R W j =1 = PBS = N own ( Eb / N 0 ) j R j j N (Eb / N 0 ) R ((1 ) + ) 1 own ((1 ) + ) 1 W W j =1 and compute its numeric value for different number of users by substituting parameters N rf = -174+7dB, = 0.5, = 0.65, = 0.58, W = 3.84 Mcps,
R = 12.2 kbps, R = 64 kbps, Eb/No = 6.7 dB (voice), Eb/No = 5.3 dB (data), PBS= 0.85*40W, PBS = 0.85*10W (15% of BS power is spend on control channels)
Finally, we add 6 dB margin to achieved mean path loss. The resulting values are plotted in the figure of the following slide
35
Example
Observations: Number of downlink data users can be increased only slightly by increasing the base station transmission power Number of voice users can be significantly increased by increasing the base station transmission power
12.2 kbps voice
170
160
155
150
145
140
10
20
30 40 Number of users
50
60
70
64 kbps data
36
Example:
WCDMA capacity
70 number of simultaneous users per cell 60 50 40 30 20 10 0 UL loading 30% UL loading 50% UL loading 70%
The interference in the cell directly effects the cells capacity In WCDMA also the coverage is then effected
Voice
2,5 Cell radius [km] 2 1,5 1 0,5 0
CS 64 kbps
PS 64 kbps
PS 128 kbps
PS 384 kbps
Voice
CS 64 kbps
PS 64 kbps
PS 128 kbps
PS 384 kbps
38
28%
Traffic density: Amount of users increased by 25% homogeneously (other 39 conditions remained the same, indoors)
40%
Traffic asymmetry: the users moved from using 64/128 kbps PS RAB 40 to 128/128 RAB (other conditions remained the same, indoors)
As traffic increases, some sites might even prove to be decreasing the service probability (or need modification) => simulations with varying traffic density, distribution and asymmetry needed
Typically Iub throughput is selected as primary dimensioning (and financial) parameter, but support for enough cells/NodeB is always ensured
Enough HW channel elements (cards) are needed to support the offered traffic One HW channel typically support one AMR user and associated signalling (a 384kbps user can take up to 16 HW channels) The HW channels are organised as a pool common for all sectors Common channels need to be taken into account in the dimensioning
43
The most typical configuration in UMTS networks today is the 20W per sector multi-carrier PA. The problem with high-power MCPA is the possibility to lose code orthogonality when high interference is produced
44