GPRS&EDGE Network Planning and Optimization-Chapter5 Network Planning-20040527-A-1.0

Document number
Product name
Target readers
Product version
Edited by
Document version
GPRS/EDGE Network Planning

Prepared by
Date
Reviewed by
Date
Reviewed by
Date
Approved by
Date
Huawei Technologies Co., Ltd.

All rights reserved
Table of Contents
Chapter 5 GPRS/EDGE Network Planning.....................................................................................
5.1 Planning Principles...............................................................................................................
5.1.1 General......................................................................................................................
5.1.2 Effect of GPRS/EDGE Network against GSM Network..............................................
5.2 Traffic Model Establishment.................................................................................................
5.2.1 GPRS/EDGE Application...........................................................................................
5.2.2 Data Rate Calculation................................................................................................
5.2.3 Precautions................................................................................................................
5.3 Coverage Planning...............................................................................................................
5.3.1 Coverage Requirements............................................................................................
5.3.2 Carrier-to-Interference Ratio Requirement..............................................................
5.4 Frequency Planning...........................................................................................................
5.5 Capacity Planning..............................................................................................................
5.5.1 Planning Method......................................................................................................
5.6 Signaling Channel Planning...............................................................................................
5.6.1 Planning Method......................................................................................................
5.6.2 Conclusion...............................................................................................................
5.7 Parameter Planning...........................................................................................................
5.7.1 System Information Parameter Configuration..........................................................
5.7.2 Coding Scheme Conversion Parameters.................................................................
5.7.3 Configuration of Parameters Controlling GPRS/EDGE Cell Reselection.................
5.8 Dual-Band Network and Network Performance Indexes....................................................
5.8.1 Effect of GPRS/EDGE Service against Dual-Band Network and Solution
Strategies.........................................................................................................................
5.8.2 Network Performance Indexes.................................................................................
5.8.3 System Performance Indexes..................................................................................
5.8.4 Maintenance Indexes...............................................................................................
5.8.5 Reference Indexes...................................................................................................
5.8.6 Other GPRS/EDGE Indexes Deserved Special Attention........................................
Attachment 1 Discussions on GPRS/EDGE Traffic Model.......................................................
Attachment 2 Capacity Planning Examples..............................................................................
Attachment 3 Parameter Configuration for GPRS MS Cell Update..........................................
Attachment 4 GPRS System Information Description..............................................................
Attachment 5 Signaling Channel Planning Examples..............................................................
Chapter 5 GPRS&EDGE Network Planning
For internal use only
Chapter 5 GPRS/EDGE Network Planning

5.1 Planning Principles
5.1.1 General
The general principles of GPRS/EDGE network planning principles are as
follows:
Promote balance development of voice service and data service
Fully utilize the existing GSM network resources
Satisfy GPRS/EDGE network requirements while assure GSM network

quality
5.1.2 Effect of GPRS/EDGE Network against GSM Network

If a GPRS/EDGE network is introduced, it will bring the following effects against
the existing GSM network:
The GPRS/EDGE downlink power control mechanism is not mature, so the

effective downlink power control are not available in a GPRS/EDGE
network, which result in great downlink interference.
The transmission features and voice services implemented across the radio
interfaces in GPRS/EDGE network is different from that in GSM network.
The duty ratio of the PDCH radio signals is approaching 100%. Therefore,
compared with a TCH using DTX (discontinuous transmission), a maximum
interference of 3 dB might be introduced.
An extra interference of 0 dB to 3 dB will be generated depending on the

number and carries of the PDCH configured.
If the PDCH is configured on BCCH, no new interference will be added in

the downlink direction theoretically.
Currently, the GPRS/EDGE and GSM use the same cell selection and
reselection parameters. To meet GPRS/EDGE service quality, the radio
network capacity and coverage must be re-planned.
If the CCCH is used of GPRS/EDGE access, the load on CCCH will be

sharply increased.
GPRS/EDGE uses more flexible channel allocation strategies, so the radio

resource dispatch becomes more complicated.
As GPRS/EDGE service develops, subscriber service (including voice

service) will change. Therefore, both traffic model and overall planning need
to be adjusted accordingly.
2006-04-05
All rights reserved
Page 4 of 48
The compatibility and inheritance of the existed GSM/GPRS network are

fully considered in the EGPRS system, so the effect against core network is
light, and no change has occurred across the Gb, Gn/Gp, Gi, Gr, and Gs
interfaces of the existed GSM/GPRS network.
5.2 Traffic Model Establishment

5.2.1 GPRS/EDGE Application
GPRS/EDGE provides mobile subscribers with a kind of end-to-end packet data
transfer service, which can contribute to more efficient utilization of radio
resources than traditional circuit-switched data services can. GPRS/EDGE is
applicable to the services with the following characteristics:
Intermittent and bursting data transfer (The interval between two

consecutive transfers is far greater than the delay of the data transfer itself.)
Frequent transfer for a little amount of data (For example, several transfers
are performed within each minute, but the amount of data transferred is
even smaller than several K-bytes.)
Infrequent transfer for a large amount of data (For example, only several
transferred are performed within each hour, but the amount of data
transferred is greater than tens of K- bytes.)
Currently, the GPRS/EDGE network in China enables the following services:
Mobile phone and laptop online
WAP over GPRS/EDGE
Terminal installation-based services
Personal data assistant (PDA) terminal access
Short message over GPRS/EDGE
Private line access
Private network access
5.2.2 Data Rate Calculation

Generally, the data rate should be calculated according to the traffic model.
Hereunder provides assumptions and facts for data rate calculation.
The following assumptions can be made:
The SNDCP (SubNetwork Dependent Convergence Protocol) is not

compressed and decompressed, and is not segmented and reassembled.
(Such an IP packet is transferred as an LLC PDU across the LLC layer.)
LLC performs transfer with unacknowledged mode.
2006-04-05
All rights reserved
Page 5 of 48
The format of an LLC frame is: LLC header (9 bytes) + SNDCP header (4
bytes) + IP data + FCS (Frame Check Sequence, 3 bytes). Each data
packet seizes a RLC length indication byte.
RLC performs transfer with acknowledged mode and 10% of retransmission

rate is considered.
An IP packet is 200 bytes in length in average.
IP data flows continuously within 10 IP packets at least.
A radio block is transmitted every 20ms.
Hereunder lists the facts for the data rate calculation:
Generally, a RLC/MAC header seizes 3 bytes. If tail bits are not considered,
the LLC PDU bytes can be transmitted in each RLC data packet is 20 under
CS1 (Channel Coding Scheme), 30 under CS2, 36 under CS3, and 50
under CS4.
For the transfer performed with RLC acknowledged mode, the transfer of an
LLC PDU means a TBF establishment and release procedure. Generally,
during a TBF establishment and release procedure, the overhead of the
RLC/MAC control blocks account for 20% of the total radio blocks. This
overhead includes the time for TBF establishment and release.
The protocol headers of FR, NS, BSSGP, LLC, and SNDCP across the Gb
interface total 53 bytes.
In addition to the previous assumptions and facts, the following calculation

models must also be used:
M = A1/B
T = (M + M0.2 + M0.1) x 20
V_IP = A2/T
VGb = V_IP x (200 + 53)/200 = 1.265V_IP
Here,
M is the minimum number of RLC data blocks needed for the transfer of
N*LLC PDUs.
A1 is the total bytes of the N*LLC PDUs.
A2 is the total bytes of the N*IP packets.
B is the LLC PDU bytes that a RLC data block can bear.
T is the time needed for the transfer of N*LLC PDUs.
V_IP is the bearer rate of each PDCH estimated across the IP layer.
VGb is the bearer rate of each PDCH estimated across the Gb interface
physical layer.
Xindicates that X must be rounded up to the nearest integer.
Xindicates that X must be rounded down to the nearest integer.
Based on the previous assumptions, facts, and calculation models, the bearer
rate estimated across the IP layer under CS2 can be calculated as follows:
2006-04-05
All rights reserved
Page 6 of 48
A1 = (200 + 9 + 4 + 3 + 1) x 10 = 2170 (bytes)
M = 2170/30= 73 (blocks)
T = (73 + 73 x 0.2+73 x 0.1) x 20ms = 1880ms = 1.880s
V_IP = 200 x 10 x 8/1.88/1024 = 8.31 Kbps
VGb = 1.265V_IP = 10.51Kbps
Likewise, the bearer rate estimated across the IP layer under CS1, CS3, and
CS4 is 5.58 Kbps, 9.89 Kbps, and 13.95 Kbps respectively.
Generally, if unacknowledged RLC mode is used, the LLC layer will not use the
unacknowledged mode. In this case, the retransmission rate across the IP layer
must be considered. Since the retransmission rate across the LLC layer is higher
than that across the RLC/MAC layer in terms of wideband overhead,
unacknowledged RLC mode cannot enhance the bearer rate of PDCH across
the IP layer. For all the data rates, see the following table.
Bearer Rate
CS1 (kbps)
CS2 (kbps)
CS3 (kbps)
CS4 (kbps)
9.05
13.4
15.6
21.4
5.58
8.31
9.89
13.95
16
16
32
32
7.06
10.51
12.51
17.65
Um interface rata at
physical layer
Rate at IP layer
Abis interface rate at
physical layer
Gb
interface
at
physical layer
5.2.3 Precautions
The following items need special attention:
GPRS/EDGE and GSM shares the CCCH signaling channel. If the network
uses PCCCH access, the channel planning, parameter planning, and dualband network load strategies need to be adjusted greatly during network
planning.
The accurate estimation for GPRS/EDGE traffic model is unpractical. The

GPRS/EDGE traffic model cannot be described simply according to the Erl
of each subscriber, which is the basis of traffic model for voice service.
GPRS/EDGE
traffic
GPRS/EDGE
service
model
deserves
develops,
the
long-term
analysis
attention.
and
As
modification
the
for
GPRS/EDGE traffic model must be made accordingly, which makes the

adjustment of the methods and strategies for GPRS/EDGE network
2006-04-05
All rights reserved
Page 7 of 48
planning possible. In addition, the introduction of new services may risk an

abrupt change of the GPRS/EDGE traffic model.
The following contents are closely related to GPRS/EDGE traffic model:
Subscriber model (total volume of data service, ratio of different services,

distribution of service in terms of time, distribution of service in terms of
space)
Service model (packet length and packet gap distribution of single service)
Radio transmission model (CS distribution, signaling overhead)
5.3 Coverage Planning

Compared with GSM coverage, the GPRS/EDGE coverage has the following
features:
EIRP (Equivalent Isotropic Radiated Power) is the same.
The loss from transmission end to reception end is the same except body
loss.
GPRS/EDGE service is restricted by C/I (carrier-to-interference ratio)

instead of receiver sensitivity.
When CS1 is used, the C/I is required to be higher than 9 dB (for voice
service, the C/I is 9 dB). If it configured on BCCH carrier, the C/I can meet
the requirement.
The C/I varies greatly with transmission environment according to the data
defined in protocols. Even if in the same environment, however, the C/I varies
greatly with carriers. Therefore the poorest C/I is considered here. As defined in
the protocols, the lowest requirement on C/I (without considering USF, that is,
uplink state flag) is 13 dB under CS1 and 15 dB under CS2.
For the relationship between the conversation quality and C/I, see the following
table.
rxqual
C/I (dB)
23
19
17
15
13
11
According to previous analysis,
When the conversation quality level is 4 and the BLER (block error rate) of
CS1 is lower than 10%, the required C/I in GSM is the same as that
required in GPRS/EDGE.
When the conversation quality level is 3 and the BLER is lower than 10%,
the required C/I in GSM is the same as that required in GPRS/EDGE.
Therefore, the following conclusions can be reached:
2006-04-05
All rights reserved
Page 8 of 48
Under CS1 and CS2, GPRS coverage and GSM coverage are almost
equivalent.
Under the condition of existing network coverage and radio quality, CS2 is
mainly used.
5.3.1 Coverage Requirements

Hereunder lists the coverage requirements:
BLER must be controlled within 10%.
The coverage area under CS1 equals voice coverage area.
The coverage area under CS2 must reach 80% of the voice coverage area.
For link budget, the requirements on receiver sensitivity in GSM are as follows:
BTS: -104dBm
MS: -102dBm
The following three tables specify the requirements on BTS and MS sensitivity in
GPRS/EDGE of various channel types and modulation schemes.
For the signal level requirements of various channel types under GMSK
modulation scheme (common BTS), see the following table.
GSM900 MHz
Channel Type
Transmission Conditions
Static
Tu50 (no
Tu50
RA250
HT100(no
FH)
(ideal
(no FH)
FH)
FH)
PDTCH/CS-1
dBm
-104
-104
-104
-104
-103
PDTCH/CS-2
dBm
-104
-100
-101
-101
-99
PDTCH/CS-3
dBm
-104
-98
-99
-98
-96
PDTCH/CS-4
dBm
-101
-90
-90
USF/CS-1
dBm
-104
-101
-103
-103
-101
USF/CS-2 to 4
dBm
-104
-103
-104
-104
-104
PRACH/11 bits1)
dBm
-104
-104
-104
-103
-103
PRACH/8 bits1)
dBm
-104
-104
-104
-103
-103
GSM900 MHz
Channel Type
Static
TU50 (no
TU50
RA250
HT100(no
FH)
(ideal
(no FH)
FH)
FH)
2006-04-05
All rights reserved
Page 9 of 48
PDTCH/MCS-1
dBm
-104
-102.5
-103
-103
-102
PDTCH/MCS-2
dBm
-104
-100,5
-101
-100,5
-100
PDTCH/MCS-3
dBm
-104
-96,5
-96,5
-92,5
-95,5
PDTCH/MCS-4
dBm
-101,5
-91
-91
USF/MCS-1 to
dBm
-104
-102.5
-104
-104
-102,5
dBm
-104
-104
-104
-103
-103
dBm
-104
-104
-104
-103
-103
4
PRACH/11
bits2), 3)
PRACH/8 bits 1)
DSC1800 MHz
Channel Type
Static
TU50 (no
TU50
RA130
HT100 (no
FH)
(ideal
(no FH)
FH)
FH)
PDTCH/CS-1
dBm
-104
-104
-104
-104
-103
PDTCH/CS-2
dBm
-104
-100
-100
-101
-99
PDTCH/CS-3
dBm
-104
-98
-98
-98
-94
PDTCH/CS-4
dBm
-101
-88
-88
USF/CS-1
dBm
-104
-103
-103
-103
-101
USF/CS-2 to 4
dBm
-104
-104
-104
-104
-103
PRACH/11 bits1)
dBm
-104
-104
-104
-103
-103
PRACH/8 bits
dBm
-104
-104
-104
-103
-103
1)
DSC1800 MHz
Channel Type
Static
TU50 (no
TU50
RA130
HT100 (no
FH)
(ideal
(no FH)
FH)
FH)
2006-04-05
All rights reserved
Page 10 of 48
PDTCH/MCS-1
dBm
-104
-102,5
-103
-103
-101,5
PDTCH/MCS-2
dBm
-104
-100,5
-101
-100,5
-99,5
PDTCH/MCS-3
dBm
-104
-96,5
-96,5
-92,5
-94,5
PDTCH/MCS-4
dBm
-101.5
-90,5
-90,5
USF/MCS-1 to
dBm
-104
-104
-104
-104
-102,5
PRACH/11 bits
dBm
-104
-104
-104
-103
-103
PRACH/8 bits
dBm
-104
-104
-104
-103
-103
Note:
* indicates that the PDTCH/CS-4 and PDTCH/MCS-x cannot meet the

service standard required by the transmission condition.
TU50 stands for typical urban area, 50km/h; FH stands for frequency
hopping, RA250 stands for rural area, 250km/h; for HT, it is 100km/h.
For the signal level requirements of various channel types under 8-PSK
modulation scheme (common BTS), see the following table.
GSM900 MHz
Channel Type
Stati
Tu50 (no
Tu50
RA250
HT100(no
FH)
(ideal
(no FH)
FH)
FH)
PDTCH/CS-1
dBm
-104
-104
-104
-104
-103
PDTCH/CS-2
dBm
-104
-100
-101
-101
-99
PDTCH/CS-3
dBm
-104
-98
-99
-98
-96
PDTCH/CS-4
dBm
-101
-90
-90
USF/CS-1
dBm
-104
-101
-103
-103
-101
USF/CS-2 to 4
dBm
-104
-103
-104
-104
-104
PRACH/11 bits1)
dBm
-104
-104
-104
-103
-103
PRACH/8 bits1)
dBm
-104
-104
-104
-103
-103
GSM900 MHz
Channel Type
Stati
TU50 (no
TU50
RA250
HT100(no
FH)
(ideal
(no FH)
FH)
FH)
2006-04-05
All rights reserved
Page 11 of 48
PDTCH/MCS-5
dBm
-101
-96,5
-97
-96
-95
PDTCH/MCS-6
dBm
-99,5
-94
-94.5
-91
-91
PDTCH/MCS-7
dBm
-96
-89
-88.5
87**
86**
PDTCH/MCS-8
dBm
-93
-84
-84
-81,5**
PDTCH/MCS-9
dBm
-91,5
-80
-80
DSC1800 MHz
Channel Type
Stati
TU50 (no
TU50
RA130
HT100 (no
FH)
(ideal
(no FH)
FH)
FH)
PDTCH/MCS-5
dBm
-101
-95,5
-97
-96
-93
PDTCH/MCS-6
dBm
-99,5
-94
-94
-91
85,5
PDTCH/MCS-7
dBm
-96
-87
-86,5
-87**
PDTCH/MCS-8
dBm
-93
86,5**
86,5**
PDTCH/MCS-9
dBm
-91,5
-83**
-83**
Note:
** indicates that for service quality, BLER < 30%.

hopping, RA250 stands for rural area, 250km/h; for HT100, it is 100km/h.
For the signal level requirements of various channel types under 8-PSK
modulation scheme (MS), see the following table.
GSM900 MHz
Channel Type
Stati
Tu50 (no
Tu50
RA250
HT100(no
FH)
(ideal
(no FH)
FH)
FH)
2006-04-05
All rights reserved
Page 12 of 48
PDTCH/CS-1
dBm
-104
-104
-104
-104
-103
PDTCH/CS-2
dBm
-104
-100
-101
-101
-99
PDTCH/CS-3
dBm
-104
-98
-99
-98
-96
PDTCH/CS-4
dBm
-101
-90
-90
USF/CS-1
dBm
-104
-101
-103
-103
-101
USF/CS-2 to 4
dBm
-104
-103
-104
-104
-104
PRACH/11 bits1)
dBm
-104
-104
-104
-103
-103
PRACH/8 bits1)
dBm
-104
-104
-104
-103
-103
GSM900 MHz
Channel Type
Stati
TU50 (no
TU50
RA250
HT100(no
FH)
(ideal
(no FH)
FH)
FH)
PDTCH/MCS-1
dBm
-104
-102.5
-103
-103
-102
PDTCH/MCS-5
dBm
-98
-93
-94
-93
-92
PDTCH/MCS-6
dBm
-96
-91
-91,5
-88
-89
PDTCH/MCS-7
dBm
-93
-84
-84
-83**
PDTCH/MCS-8
dBm
-90,5
-83**
-83**
PDTCH/MCS-9
dBm
-86
-78,5**
-78,5**
DSC1800 MHz
Channel Type
Stati
TU50 (no
TU50
RA130
HT100 (no
FH)
(ideal
(no FH)
FH)
FH)
2006-04-05
All rights reserved
Page 13 of 48
PDTCH/MCS-5
dBm
-98
-93,5
-93,5
-93
-89,5
PDTCH/MCS-6
dBm
-96
-91
-91
-88
-83,5
PDTCH/MCS-7
dBm
-93
-81,5
-80,5
PDTCH/MCS-8
dBm
-90,5
-80**
-80**
PDTCH/MCS-9
dBm
-86
USF/MCS-5 to
dBm
-102
-99
-99
-100
-99
Note:

hopping, RA250 stands for rural area, 250km/h; for HT, it is 100 100km/h.
When the CS1 is used for data transmission, the uplink sensitivity is required to
be -101 dBm at least, and the receiver sensitivity is required to be greater than
-104dBm for GSM planning. However, since body loss is not considered for data
service, there is a gain of 3 dB, so the the coverage area under CS1 equals
voice coverage area, and the coverage area under CS2 can reach 80% of the
voice coverage area. However, as the GPRS/EDGE service develops fast, the
BTSs are aggressively constructed in most areas of China Mobile. As a result,
the radio signal strength can completely go with the service.
5.3.2 Carrier-to-Interference Ratio Requirement

The C/I required by GRPS/EDGE towards MS and BTS varies with channels and
modulation schemes. For the C/I requirements of various channel types under
GMSK modulation scheme, see the following table.
GSM900 MHz
Channel Type
TU3
TU3
TU50 (no
TU50
RA250 (no
(no FH)
(ideal
FH)
(ideal FH)
FH)
FH)
2006-04-05
All rights reserved
Page 14 of 48
PDTCH/CS-1
dB
13
10
PDTCH/CS-2
dB
15
13
14
13
13
PDTCH/CS-3
dB
16
15
16
15
16
PDTCH/CS-4
dB
21
23
24
24
USF/CS-1
dB
19
10
12
10
10
USF/CS-2 to 4
dB
18
10
PRACH/11 bits1)
dB
10
PRACH/8 bits1)
dB
GSM900 MHz
Channel Type
TU3
TU3
TU50
TU50
RA250
(no FH)
(ideal FH)
(no FH)
(ideal FH)
(no FH)
PDTCH/MCS-1
dB
13
9.5
10.5
9.5
10
PDTCH/MCS-2
dB
15
12
12.5
12
12
PDTCH/MCS-3
dB
16.5
16.5
17
17
19
PDTCH/MCS-4
dB
19
21.5
22
22
USF/MCS-1 to
dB
18
10
11
9.5
9.5
dB
10
dB
4
PRACH/11
bits2), 3)
PRACH/8 bits 1)
DSC1800 MHz
Channel Type
TU1,5
TU1,5
TU50
TU50
RA130
(no FH)
(ideal FH)
(no FH)
(ideal FH)
(no FH)
PDTCH/CS-1
dB
13
PDTCH/CS-2
dB
15
13
13
13
13
PDTCH/CS-3
dB
16
15
16
16
16
PDTCH/CS-4
dB
21
23
27
27
USF/CS-1
dB
19
10
USF/CS-2 to 4
dB
18
PRACH/11 bits1)
dB
10
PRACH/8 bits
dB
1)
10
10
10
DSC1800 MHz
2006-04-05
All rights reserved
Page 15 of 48
Channel Type
TU1,5
TU1,5
TU50
TU50
RA130
(no FH)
(ideal FH)
(no FH)
(ideal FH)
(no FH)
PDTCH/MCS-1
dB
13
9.5
10
9.5
10
PDTCH/MCS-2
dB
15
12
12
12
12
PDTCH/MCS-3
dB
16.5
16.5
17
18
19
PDTCH/MCS-4
dB
19
21.5
23
23
USF/MCS-1 to
dB
18
10
9.5
9.5
9.5
PRACH/11 bits
dB
10
PRACH/8 bits
dB
Note:

hopping, RA250 stands for rural area, 250km/h.
The correct training sequence requirement is

sequence;
2)
1)
stands for 3 training sequence;
stands for one training

3)
stands for that the
requirement is applied to CPRACH.
For the C/I requirements of various channel types under 8-PSK modulation
scheme (common BTS), see the following table.
GSM900 MHz
Channel Type
TU3
TU3
TU50 (no
TU50
RA250 (no
(no FH)
(ideal
FH)
(ideal FH)
FH)
FH)
PDTCH/MCS-5
dB
18
14.5
15.5
14.5
16
PDTCH/MCS-6
dB
20
17
18
17.5
21
PDTCH/MCS-7
dB
23.5
23.5
24
24.5
26.5**
PDTCH/MCS-8
dB
28.5
29
30
30
PDTCH/MCS-9
dB
30
32
33
35
DSC1800 MHz
Channel Type
2006-04-05
All rights reserved
Page 16 of 48
TU1,5
TU1,5
TU50
TU50
RA130
(no FH)
(ideal FH)
(no FH)
(ideal FH)
(no FH)
PDTCH/MCS-5
dB
18
14.5
15
15
16
PDTCH/MCS-6
dB
20
17
17.5
18
21
PDTCH/MCS-7
dB
23.5
23.5
26
26.5
27**
PDTCH/MCS-8
dB
28.5
29
25**
24.5**
PDTCH/MCS-9
dB
30
32
29**
29**
Note:

For the C/I requirements of various channel types under 8-PSK modulation
scheme (MS), see the following table.
GSM900 MHz
Channel Type
TU3
TU3
TU50 (no
TU50
RA250 (no
(no FH)
(ideal
FH)
(ideal FH)
FH)
FH)
PDTCH/MCS-5
dB
19,5
14,5
15,5
14,5
16,5
PDTCH/MCS-6
dB
21,5
17
18
17,5
21
PDTCH/MCS-7
dB
26,5
23,5
25
24,5
PDTCH/MCS-8
dB
30,5
23,5**
25,5**
25,5**
PDTCH/MCS-9
dB
25,5**
28**
30,5**
30,5**
DSC1800 MHz
Channel Type
2006-04-05
TU1,5
TU1,5
TU50
TU50
RA130
(no FH)
(ideal FH)
(no FH)
(ideal FH)
(no FH)
All rights reserved
Page 17 of 48
PDTCH/MCS-5
dB
19,5
14,5
15
15,5
16,5
PDTCH/MCS-6
dB
21,5
17
18
18,5
21
PDTCH/MCS-7
dB
26,5
23,5
27,5
28
PDTCH/MCS-8
dB
30,5
23,5**
29,5**
29**
PDTCH/MCS-9
dB
25,5**
28**
USF/MCS-5 to
dB
17
10,5
10
Note:

According to the data recommended in protocols, the data varies greatly with
transmission conditions. Even under the same transmission condition, the data
varies greatly with carriers. In the existing network, the distribution of voice
quality level can reflect GRPS/EDGE coverage distribution.
5.4 Frequency Planning

Frequency planning aims to decide frequency reuse pattern according to C/I
requirement, capacity requirement, and the available bandwidth.
At the initial stage, the requirement on GPRS capacity is not high. According to
coverage planning, in fact, the TCH C/I on BCCH is already meet GPRS
requirement. From the perspective of frequency planning, if the PDCH is
configured on BCCH carrier, consideration on multiple factors, such as 1*3, 1*1,
MRP, IUO, frequency hopping, and power control, are unnecessary.
The conditions of the existing network are as follows:
No TCH power control for BCCH carrier
No DTX
Frequency hopping or baseband frequency hopping is not used.
4*3 frequency reuse pattern is widely used.
Therefore, the following benefits can be obtained if PDCH is configured on

BCCH:
C/I can meet the current GPRS coverage requirement.
Data service will not add extra interference to GSM service
2006-04-05
All rights reserved
Page 18 of 48
If baseband frequency hopping is available on BCCH carrier, consider using

multislot for GPRS MS. In this case, each PDCH is required to have the
same MAIO (Mobile Allocation Index Offset) and HSN (Hopping Sequence
Number).
If the PDCH is configured on non-GPRS carriers, because power control,

aggressive frequency reuse, and IUO are applied to TCH, the followings are
present:
C/I may not meet the current GPRS coverage requirement.
Data service will add extra interference to GSM service and reduce the
coverage area of voice service.
If frequency hopping is used, each PDCH on a TRX is required to have the

same MAIO and HSN.
When GPRS capacity is planned to TRX level and the resources are inadequate
if all the TCHs on BCCH carrier are configured as PDCH, you can consider
configuring PDCH on other carriers. In this case, you can follow the suggestions
listed below:
Use frequency hopping technology to reduce the requirement of the MS on

C/I so as to prevent the voice coverage area from shrinking.
Since the occupation rate of the TCHs grows, if the network capacity is not
expanded, the congestion rate and the handover call drop rate will increase
for the base stations with a smaller configuration type. Therefore, you are
supposed to expand the capacity of the network.
5.5 Capacity Planning

Capacity planning aims to provide GPRS/EDGE services to a number of
GPRS/EDGE subscribers by configuring a suitable number of PDCHs. This
section provides the methods used for capacity planning and the configuration
reference calculated according to the current traffic model. If traffic model
changes, the configuration reference needs to be recalculated.
5.5.1 Planning Method

I. Theory Basis and Assumptions
GPRS/EDGE capacity is calculated similar to that of circuit service. That is, the
formula ERLANG-B is used to calculate the capacity due to the following
considerations.
If only one type of GPRS service is enabled (for example, if only WAP service is
enabled), the radio resource seized by the service is comparable to that of the
circuit service and can be estimated according to the formula ERLANG-B.
2006-04-05
All rights reserved
Page 19 of 48
If multiple services are enabled simultaneously, because traffic volume,

bandwidth requirement, and QoS vary with services, the accurate results can be
obtained through complicated mathematic formulas or emulations, but the actual
project needs a simple and effective method. Therefore, the applicable method is
to sum up multiple services to a standard service and then apply it to the
formula ERLANG-B. The related calculation and analysis towards the congestion
of various services prove that this method greatly contributes the project
planning.
In addition to standard service assumption, the assumptions on some important
parameters are provided in IVPDCH Number Calculation.
II. Subscriber Number Calulation

GSM subscriber number and GPRS subscriber penetration rate can be used to
calculate the number of GPRS subscribers in each cell. For details, GPRS
subscriber number under various conditions, see the following table.
TRX number
Available
TCH/PDCH
timeslots
Circuit
service
traffic (Erl)
GPRS subscriber
number
2.9
1.93
14
8.2
5.47
22
14.85
9.9
29
21
14
37
28.25
18.83
45
35.6
23.73
52
42.1
28.07
60
49.6
33.07
Note:
The figures listed in this table are based on the assumption that the GPRS
subscriber rate is 2%.
The assumed traffic volume of each GSM subscriber is 0.03 Erl.
III. Subscriber Throughput Conversion

In current traffic model, the bandwidth seized by each subscriber at the busy
hour is assumed to be 180 bps. In actual conditions, however, the subscriber is
quite sensitive to the on-time bandwidth used for data transmission. Considering
calculating traffic volume continently and flexibly, you are required to convert the
bandwidth used by the subscriber at busy-hour into the subscriber bandwidth
requirement and busy-hour traffic volume.
2006-04-05
All rights reserved
Page 20 of 48
If the GPRS subscriber needs a bandwidth of 1kbyte/s in average, that is, the
behavior of GPRS subscriber is abstracted as packet call, and each call seizes
a GPRS channel, the bandwidth is 1kbyte/s. Thus, the busy-hour traffic volume
of each GPRS subscriber is 0.025 Erl (that is, 180/1000/8=0.025).
IV. PDCH Number Calculation

Formula ERLANG-B is used to estimate the capacity requirement. Hereunder
details the parameters involved.
GOS = 2% (here it means that the bandwidth cannot meet requirements

when GPRS subscriber begins service.)
GPRS subscriber penetration rate = 2%
Bandwidth of each GPRS channel = 1kbyte/s
IP layer bandwidth of each PDCH = 1kbyte/s
The corresponding formula is as follows:
GPRS channel number = ERLANG-B (GPRS traffic volume, GOS)
PDCH number = GPRS channel number * bandwidth of each GPRS

channel/IP layer bandwidth of each PDCH.
For the calculation results in various situations, see the following table.
TRX
Available
Circuit
GPRS
GPRS
PDCH
number
TCH/PDCH
service
service
channel
number
timeslots
traffic (Erl)
traffic (Erl)
number
2.9
0.0483
1.3
1.3
14
8.2
0.1367
1.8
1.8
22
14.85
0.2475
2.1
2.1
29
21
0.35
2.4
2.4
37
28.25
0.4708
2.7
2.7
45
35.6
0.5933
52
42.1
0.7017
3.3
3.3
60
49.6
0.8267
3.5
3.5
V. PDCH Configuration Calculation

The figures listed in the following two tables are calculated according to the
formula ERLANG-B. Through calculating the probability of the TCHs that have
been seized, you can obtain the average available bandwidth of the dynamic
PDCH under various configurations, as listed in
2006-04-05
All rights reserved
Page 21 of 48
Table 1.1 Average available bandwidth of the PDCH under various configurations (static
PDCH = 1; dynamic PDCH = 1, 2, and 3)
Number
Available
Circuit
Probability for the number of idle TCHs is
Number of average
of TRXs
TCH/PD
service
available
CH
traffic
when the number
timeslot
(Erl)
of dynamic PDCHs
PDCHs
is M
N=0
N=1
N=2
N=3
M=1
M=2
M=3
2.9
4.68%
9.68%
16.69%
23.03%
1.95
2.81
3.50
14
8.2
3.48%
5.52%
8.08%
10.84%
1.97
2.88
3.70
22
14.85
2.95%
4.18%
5.63%
7.20%
1.97
2.90
3.77
29
21
2.77%
3.70%
4.75%
5.89%
1.97
2.91
3.80
37
28.25
2.67%
3.40%
4.22%
5.07%
1.97
2.91
3.81
45
35.6
2.57%
3.18%
3.84%
4.53%
1.97
2.92
3.82
52
42.1
2.51%
3.03%
3.60%
4.19%
1.97
2.92
3.83
60
49.6
2.44%
2.91%
3.40%
3.91%
1.98
2.92
3.83
Table 1.2 Average available bandwidth of the PDCH under various configurations (static
PDCH = 0; dynamic PDCH = 1, 2, and 3
Number
Available
Circuit
Probability for the number of idle
Number
of
of
TCH/PDCH
service
TCHs is N
average available
TRXs
timeslot
traffic
PDCHs when the
(Erl)
number
of
dynamic PDCHs is
M
N=0
N=1
N=2
N=3
M=1
M=2
M=3
2.9
1.90%
4.59%
9.50%
16.38%
0.98
1.92
2.76
14
8.2
2.00%
3.41%
5.41%
7.92%
0.98
1.93
2.82
22
14.85
1.95%
2.90%
4.10%
5.52%
0.98
1.93
2.84
29
21
1.97%
2.72%
3.63%
4.66%
0.98
1.93
2.85
37
28.25
2.00%
2.62%
3.33%
4.13%
0.98
1.93
2.85
2006-04-05
All rights reserved
Page 22 of 48
45
35.6
2.00%
2.52%
3.12%
3.77%
0.98
1.93
2.86
52
42.1
1.99%
2.46%
2.97%
3.53%
0.98
1.94
2.86
60
49.6
1.98%
2.40%
2.85%
3.33%
0.98
1.94
2.86
When the number of the dynamic PDCHs is M, the number of the average
available PDCHs is the sum of the followings:
The number of the static PDCHs
If only one dynamic PDCH is available, the probability is the figure in the
previous table when N = 1.
If only two dynamic PDCHs are available, the probability is the figure in the
previous table when N = 2.
If M * dynamic PDCHs are available, the probability is the sum of figures in

the previous table when N M.
The following figure shows the TCH distribution calculated according to the
formula ERLANG-B when a cell is configured with 5 TRXs (for your reference).
According to this figure, two conclusions can be obtained:
The peak value of TCH seizure probability appears at the average traffic
volume, stepping down at the two sides.
The sum of the thee values at the most right is smaller than 10%. That is,
three idle TCHs are present in at least 90% of the time.
Therefore, when the number of the configured PDCHs is far smaller than the
total number of TCHs, the GPRS service can be applied in most cases.
According to IVPDCH Number Calculation and Table 1.1 and Table 1.2, you can
obtain the PDCH configuration, as listed in
2006-04-05
All rights reserved
Page 23 of 48
Table 1.3 PDCH configuration under various conditions

TRX number
Available
Circuit
PDCH
PDCH
TCH/PDCH
service
number
configuration
timeslots
traffic (Erl)
2.9
1.3
1+1; 0+2
14
8.2
1.8
1+1; 0+2
22
14.85
2.1
1+2; 0+3
29
21
2.4
1+2; 0+3
37
28.25
2.7
1+2; 0+3
45
35.6
1+3; 0+4
52
42.1
3.3
1+3; 0+4
60
49.6
3.5
1+3; 0+4
5.6 Signaling Channel Planning

Signaling channel planning aims to judge if the capacity expansion is necessary
for the network according to the configuration and load of the network and the
increase of the signaling load after GPRS/EDGE is introduced.
5.6.1 Planning Method

Similar to location area division, the routing area is configured according to the
signaling flow of the radio common channels. Hereunder provides a calculation
method for your reference. For details, see the attachments in later parts.
1)
Calculate the capacity of the signaling channel according to network

configuration.
2)
During GPRS/EDGE signaling procedure and data transfer, calculate the

signaling channel load increased by each WAP subscriber after the
introduction of GPRS/EDGE according to GPRS/EDGE traffic model.
3)
During GPRS/EDGE signaling procedure and data transfer, calculate the

signaling channel load increased by each Internet subscriber after the
introduction of GPRS/EDGE according to GPRS/EDGE traffic model.
4)
Obtain the WAP subscriber number and the Internet subscriber number
according to the results of capacity planning.
5)
Calculate the total increase of the signaling channel load after the
introduction of GPRS/EDGE.
6)
Judge if CCCH needs to be expanded according to network load.
The following table lists the capacity of the existing network.
2006-04-05
All rights reserved
Page 24 of 48
CCCH
RACH
AGCH
configuratio
capacity
assignment
(channel
message/second
request:
message/s
econd
Combined
capacity
(immediate
from
Support
Not
dynamic
signaling
support
signaling
configuration
BSC:
PCH capacity (paging message from

BSC: message/second
dynamic
Support
Not
channel
dynamic
signaling
signaling
configuration
channel
channel
configurati
configurati
on
on
support
dynamic
channel
57
0.6
Configure
2.1
12
Configure
messages/
messages
1AGCH
messages
messages
1AGCH
messages/
second
/second
and
/second
/second
and
second
2PCHs
2PCHs
No-
108
1.8
Configure
2.1
36.3
Configure
34
combined
messages/
messages
1AGCH
messages
message/
1AGCH
messages/
second
/second
and
/second
second
and
second
8PCHs
8PCHs
Configure
4.2
Configure
30
2AGCHs
messages
2AGCHs
messages/
and
/second
and
second
7PCHs
7PCHs
Configure
6.3
Configure
25
3AGCHs
messages
3AGCHs
messages/
and
/second
and
second
6PCHs
6PCHs
The following table lists the increased amount after the introduction of
GPRS/EDGE.
TRX
Inc_Num_Pac
Inc_Num_Pa
Inc_Num_Pack
Inc_Num_Packet
number
ketChannReq
cketPaging_
etDLImmass_P
Immass_AGCH_
_RACH_Total
PCH_Total
CH_Total
Total
(message/sec
(message/se
(message/seco
(message/secon
ond)
cond)
nd)
d)
0.06
0.11
0.16
0.32
0.3
0.58
0.42
0.85
2006-04-05
All rights reserved
Page 25 of 48
0.57
1.13
0.7
1.41
0.85
1.7
The PCHs cannot not be monitored by the GPRS/EDGE subscribers (PDP

active) during data transfer will result in circuit-switched paging retransmission.
And these PDCHs must be considered in the increased number of paging
channels.
5.6.2 Conclusion
As far as the existing network is concerned, the RACH and PCH needs not to be
expanded, but you need to consider configuring extended BCCH for AGCH
based on the actual load. To support GRPS/EDGE service and avoid AGCH
congestion, you need to expand the AGCH. In addition to that, the following
methods are also recommended:
Configure multiple non-combined CCCHs
Adopt PCCCHs at the early stage
Enable the support of Gs interface
Currently, the GSM network uses the non-combined CCCH configuration. At the
early construction stage of the GPRS/EDGE network, the network size and
subscriber number is small, so you do not have to expand the CCCH
immediately after the GPRS/EDGE service is introduced into the GSM network.
As the size of the GPRS/EDGE network grows, you can increase the number of
CCCHs gradually according to network load.
The size of the routing area is the same as that of the location area during
planning stage. As the number of GPRS/EDGE subscriber increases, however,
you need to reduce the size of the routing area and packet pages to reduce PCH
load.
5.7 Parameter Planning

The planning towards GPRS/EDGE parameters can be divided into two parts,
namely, the part defined by protocols and the part defined by each carrier. The
part defined by protocols is reflected by the setting of the GPRS/EDGE system
parameters, which are detailed in the later attachments. The part defined each
carrier involves radio resource allocation and utilization strategies, power control,
and so on, which varies greatly with carriers.
2006-04-05
All rights reserved
Page 26 of 48
5.7.1 System Information Parameter Configuration

The system information (SI) related to GPRS/EDGE are SI13 and PSI13 (packet
system information). SI13 is transmitted on BCCH, and PSI13 is transferred on
PDCH and PACCH. The configuration of system information parameters has
effects against MS behavior and network performance. Therefore, you are
required to pay special attention to system information parameters. Hereunder
details several parameters.
DRX_TIMER_MAX
The MS can keep in NONDRX state for a period of time after entering idle state
from packet transfer state. In this case, an immediate assignment message for
downlink TBF establishment does not have to be sent on PCH after paging group
calculation. Instead, it can be send on the AGCH directly. The period of the time
for the MS to keep in NONDRX state depends on the minimum value of the DRX
parameter and the parameter DRX_TIME_MAX, the former of which is carried in
an ATTACH message sent by the MS. If the parameter DRX_TIME_MAX is set to
a larger value, the time for TBF establishment will be shortened. In addition, the
standby time of the MS will also be shortened. Through adjusting the parameter
DRX_TIME_MAX, you can also adjust the signaling load on AGCH and PCH.
T3192
Upon the release of downlink TBF, the MS will start T3192. Before T3192
expires, the network side can send downlink immediate assignment messages
on the PACCH directly if receiving new downlink data. If T3192 is set to a larger
value, you can shorten the TBF establishment time at special occasions. In this
case, a waste of radio resources may be caused. Through adjusting T3192, you
can also adjust the signaling load on PDCH and AGCH.
PAN_DEC, PAN_INC, and PAN_MAX
When transferring uplink data, the MS will start T3182 to wait for the
acknowledgement from the network. If T3182 expires, the counter N3102 will
reduce by PAN_DEC. If the MS receive the acknowledgement from the network,
the counter N3102 will increase by PAN_INC. If counter N3102 reduces from
PAN_MAX to 0, the MS will perform cell reselection. Through adjusting the three
parameters, you can adjust the sensitivity of the MS to radio transmission quality.
5.7.2 Coding Scheme Conversion Parameters

I. Defaulted Coding Scheme
For the areas with great carrier-to-interference ratio (C/I), you are suggested to
default the coding scheme to CS-2. For the areas with low C/I, you are
suggested to default the coding scheme to CS-1.
2006-04-05
All rights reserved
Page 27 of 48
Generally, the C/I will be in good terms if PDCH is configured on BCCH carrier. In
this case, you are suggested to set the coding scheme to CS-2. For the cells
where coverage is poor, you are suggested to set the coding scheme to CS-1.
If throughput is considered, according to theoretical analysis and the simulation
data defined in the protocols made by ETSI (European Telecommunications
Standards Institute), the throughput under CS-2 (when BLER > 20%) must be
lower than that under CS-1. In actual engineering, therefore, you are can adopt
CS-2 by force, and then judge whether to adopt CS-1 through analyzing the
average block retransmission rate under CS-2.
II. Coding Scheme Coversion Threshold

If coding scheme conversion threshold is adjustable, you can set the threshold
according to the methods described in the pervious section.
5.7.3 Configuration of Parameters Controlling GPRS/EDGE Cell

Reselection
At present, only a few parameters are available for controlling GPRS/EDGE cell
reselection, so the current ability for the GPRS/EDGE system to manage and
control congestion is relatively poor.
Therefore, you are required to make a good planning in terms of coverage and
parameters for the areas in which no strong cell is present. In some occasions,
the parameter GMMREADY can be used to solve the problem if it is set as
follows:
GMMREADY timer is set to about 1 minute.
For the areas in which no strong cell is present, the cell reselection penalty
time is longer than that of the GMMREADY timer.
For details, see the later attachments.

For cell reselection, the relationship between traffic model and RLC data
throughput can be obtained according to the data recorded for the existing
network.
5.8 Dual-Band Network and Network Performance

Indexes
5.8.1 Effect of GPRS/EDGE Service against Dual-Band Network and
Solution Strategies
GPRS/EDGE service has the following effects against the dual-band network:
If the traffic guide strategy of the dual-band network enables the 1800 MHz
network to take the priority to absorb the traffic, the GPRS/EDGE service
2006-04-05
All rights reserved
Page 28 of 48
will concentrates at the 1800 MHz network, which will cause network
congestion. However, the dual-band network cannot guide the traffic load to
the 900MHz network similarly by means of the handover used for GSM
voice service.
The indoor coverage of 1800MHz network is relatively poor, which affects

the QoS of packet service, but the dual-band network cannot take the
initiative to enable the 900MHz network to help the 1800MHz network with
its indoor coverage.
The reasons for the previous problems are that the GPRS/EDGE MS and the
GSM MS use the same strategies for cell selection and reselection. At present,
however, the GPRS/EDGE does not support network-controlled cell reselection.
In addition to improving capacity planning, the following methods can also be
used to solve GPRS/EDGE congestion problems:
Configuration dynamic allocation PDCH for the 1800MHz network.
When GPRS/EDGE congestion occurs at the 1800MHz network, trigger the

voice service seizing the dynamic allocation PDCH to hand over to the
900MHz network.
Apply the dynamic allocation PDCH released by the previous handover to

the GPRS/EDGE service.
If the GPRS/EDGE MS attaches to the 1800MHz network and the service

problems are caused by poor coverage, you need to solve the problem by
improving the coverage planning for the 1800MHz network or changing the
priority of the 1800MHz network.
5.8.2 Network Performance Indexes

The performance measurement indexes concerning the daily maintenance for
the GPRS/EDGE network can be divided into three categories, which are
detailed hereunder:
System measurement indexes
System measurement indexes reflect the processing capability and data

throughput capability of the system.
Maintenance indexes
Maintenance indexes record the abnormal conditions, such as request rejection,

block retransmission rate, occurring in the system. Through monitoring the
maintenance indexes, you can remove network failures in time and guide
network optimization. As for the values of the indexes, they must be collected,
analyzed, and summarized during network operation.
Reference indexes
2006-04-05
All rights reserved
Page 29 of 48
Reference indexes, such as the average packet length and the overhead of each
subscriber at busy-hour, are related to traffic model. These parameters help you
accumulate more data and experience on how to run the GPRS/EDGE network
and can be applied to the future optimization and expansion of the GPRS/EDGE
network.
At the early implementation of the GPRS/EDGE service, the traffic model
adopted is not authenticated by the network. In addition, the introduction of new
services may result in rapid change of the traffic model. Therefore, you need to
monitor the key indexes to prevent the change of GPRS/EDGE traffic model from
impacting the network planning.
5.8.3 System Performance Indexes

Hereunder lists the system performance indexes:
PDCH number in service state
RLC data block throughput (by uplink, downlink, and coding schemes)
Uplink packet service requests (by CCCH and PACCH)
Uplink TBF establishment successes
Packet paging requests
LLC PDU packets, throughput (by uplink and downlink)
5.8.4 Maintenance Indexes

Hereunder lists the maintenance indexes:
Retransmission rate of RLC data block (by uplink, downlink, coding

schemes)
Change rate of coding scheme (by uplink, downlink, and coding schemes)
Uplink TBF establishment rejects (by CCCH and PACCH)
TBF establishment failure rate (by uplink and downlink)
TBF interruption rate (by uplink, downlink, and cause)
Downlink LLC PDU retransmissions
Number of downlink LLC PDUs discarded due to time out of life cycle
5.8.5 Reference Indexes

Hereunder lists the reference indexes:
PDCH utilization rate: PDCHs in service state/available PDCH
RLC data block utilization rate (by coding scheme)
Ratio of MAC blocks to RLC data blocks (by uplink and downlink)
Ratio of downlink TBF establishment during PCH, AGCH, PACCH, and

T3192
Ratio of uplink TBF establishment during AGCH and PACCH
2006-04-05
All rights reserved
Page 30 of 48
Mean throughput and mean duration of TBF (by uplink and downlink)
Mean number of TBF occupied PDCHs (by uplink and downlink)
Mean length of LLC PDU (by uplink and downlink)
5.8.6 Other GPRS/EDGE Indexes Deserved Special Attention

The following two indexes are quite important for capacity planning. The system
performance will deteriorate sharply if they are not properly planned.
Mean length of LLC PDUs (by uplink and downlink)
Uplink and downlink TBF overhead on CCCH
The following two indexes can work as reference for network expansion.
PDCH utilization rate
Uplink TBF establishment rejects
2006-04-05
All rights reserved
Page 31 of 48
Attachment 1 Discussions on GPRS/EDGE Traffic Model

I. GPRS/EDGE Traffic Model Analysis
The GPRS/EDGE traffic model cannot be defined simply according to the traffic
volume (Erl) of each subscriber. From the perspective of radio network planning,
two parameters, packet mean length and packet signaling overhead, deserve
special attention.
II. Packet Mean Lenght

From the perspective of packet length probability distribution function, ETSI
provides three models, Funnet model, Mobitex model, and Railway model. The
three models are from the existed packet data network.
The Funnet model is obtained from the record of the e-mails from the Finland
University and the e-mails researching computer networks, as shown in Figure
1.1. According to this figure, the probability distribution function can be
approximately taken as the truncated Cauchy distribution of a maximum of 10
kbps (the maximum value is 0.8kbytes/s).
Figure 1.1 Probability distribution function under Funnet model

The railway model is obtained from the abstraction of the record of the amount of
the mobile data service for European railways, as shown in Figure 1.2. According
to this figure, the maximum value is 1kbps.
2006-04-05
All rights reserved
Page 32 of 48
Figure 1.2 Probability distribution function under Railway model

The Mobitex model is abstracted from the amount of the service of the Sweden
Mobitex mobile data communication system. The amount of service is
3015bytes on uplink and 11557bytes on downlink.
According to pervious examples, the packet length distribution of data packet
varies greatly with environments (fixed network, railway, and sea) and services
(e-mail and group communication).
Hereunder defines two marks:
The average length of IP packet on downlink: DlAveIpPacketLen
The average length of IP packet on uplink: UlAveIpPacketLen
At present, the average length of each packet is set to 200 bytes for
GPRS/EDGE radio network planning, but it must be adjusted according to actual
network data after the network is put into practical use.
2006-04-05
All rights reserved
Page 33 of 48
III. Packet Signling Overhead
Figure 1.1 Packet signaling overhead

As shown in this figure, the arrival of a packet is characterized by discontinuity. If
multiple packets arriving continuously form a packet call, which is called
temporary block flow in GPRS/EDGE. When CCCH access mode is used, the
establishment and release of each TBF will add signaling load to the CCCH. For
details, see the following table.
Item
RACH
AGCH
PCH
Remark
TBF
Packet
Packet
establishm
channel
uplink
ent
request*1
immediate
TBF
Packet
Generally, paging groups need
establishm
downlink
to be calculated for downlink
ent
immediate
immediate assignment
on
uplink
assignment
*1
on
downlink
assignment*1
(GMMREA
DY, DRX)
2006-04-05
All rights reserved
Page 34 of 48
TBF
Packet
The MS keeps in NON-DRX
establishm
downlink
state for a period of time after
ent
immediate
releasing the TBF. In this case,
downlink
assignment
the
(GMMREA
*1
assignment message can be
on
DY
downlink
immediate
sent on AGCH.
NONDRX)
TBF
Packet
In GMMSTANDBY state, the
establishm
paging*1
network delivers the paging
ent
on
message first. Upon receiving
downlink
the message, the MS will make
(GMMSTA
a response to the message.
NDBY)
After that, the MS will enter

GMREADY state.
Others
If the downlink and uplink TBF

is interleaving with each other,
the new TBF is established on
the PACCH on which the old
TBF is established other than
on the CCCH. Upon the release
of the downlink TBF and before
T3192
expires,
the
new
downlink TBF is not established

on CCCH.
The signaling overhead parameters and the marks of data packets on CCCH are
defined as follows for the convenience of engineering calculation.
DlImmAssOnAgchNumPerKbyteDl
It indicates the packet downlink immediate assignments on AGCH per 1000

bytes on the downlink.
DlImmAssOnPchNumPerKbyteDl
It indicates the packet downlink immediate assignments on AGCH per 1000

bytes on the downlink.
PageNumPerKbyteDl
It indicates the packet page number per 1000 bytes on the downlink.
ChReqNumPerKbyteUl
It indicates the packet request number per 1000 bytes on the uplink.
2006-04-05
All rights reserved
Page 35 of 48
UlImmAssPerKbyteUl
It indicates the packet uplink immediate assignments per 1000 bytes on the
uplink.
Hereunder provides some assumptions for the estimation of signaling overhead.
The assumptions are listed in the following:
On the downlink, the TBF must be established once for each 750 bytes. The
750 bytes, indicating the average length of the data for a WAP page, are
provided by the equipment provider.
For each establishment of TBF, the ratio for the network to page the MS is
0.. At the early stage, GET service will take the majority. In this case, the
interval between the MS sending service requests to the network and the
network feeding back the data is often shorter than READY timer. The
signaling overhead of each GRPS/EDGE message of the MT is smaller than
that of the GSM message of the MT, which has already been considered in
CCCH load calculation for GSM.
The ratio of uplink TBF establishment times to downlink TBF establishment

times is 1:1. Based on GET theory, each transfer of downlink data can be
considered triggered by uplink data transfer.
In this case, the following results can be obtained:
DlImmAssOnAgchNumPerKbyteDl =0
DlImmAssOnPchNumPerKbyteDl =1.33
PageNumPerKbyteDl =0
ChReqNumPerKbyteUl =5.33
UlImmAssPerKbyteUl =5.33
According to protocols, the MS can keep in NONDRX state after leaving

TRANSFER state for IDLE state. In this case, the immediate assignment
message for the downlink TBF establishment needs not account the paging
group and can be sent on AGCH. The time for the MS to keep in NONDRX state
depends on the smallest value of the parameter DRX_TIME_MAX carried in
system information and the parameter DRX carried in the ATTACH message. In
actual engineering, the parameter DRX_TIME_MAX can be set to a larger value.
If the equipment supports the messages sent on AGCH, PCH can be seized. In
this case, the CCCH utilization rate will become higher and the standby time of
the MS will become shorter.
IV. Methods to Extract Traffic Model from Runing Network

The GPRS/EDGE network has not been put into commercial use, so great
deviation is allowed for the assumptions of its traffic model. In this case, you are
required to measure the related parameters in actual running network for the
2006-04-05
All rights reserved
Page 36 of 48
purpose of correcting traffic model. The average IP packet length is measured

within a period of busy-hours in the unit of cell or BSS.
DlBssgpPacketNum (it is the downlink BSSGP packet number)
DlBssgpDataVol (it is the downlink BSSGP data volume)
UlBssgpPacketNum (it is the uplink BSSGP packet number)
UlBssgpDataVol (it is the uplink GSSGP data number)
The average packet length can be expressed by the following equations:
DlAveIpPacketLen = DlBssgpDataVol/DlBssgpPacketNum-BSSGP header

overhead
UlAveIpPacketLen = UlBssgpDataVol/UlBssgpPacketNum-BSSGP header

overhead
Here, BSSGP header overhead is about 53 bytes in length.

Hereunder provides the methods to measure the IP packet CCCH signaling
overhead within a period of busy-hour in the unit of cell or BSC.
DlBssgpDataVol (it is the downlink BSSGP data volume)
UlBssgpDataVol (it is the uplink BSSGP data volume)
DlImmAssOnAgchNum (it is the downlink immediate assignments sent on

SGCH)
DlImmAssOnPchNum (it is the downlink immediate assignments sent on

PCH)
PageNum (it is the CCCH packet paging number)
ChReqNum (it is the CCCH packet channel request number)
UlImmAss (it is the CCCH packet uplink immediate assignments)
If the BSSGP volume is converted into IP data volume, the following equations
can be obtained:
DlIpDataVol
DlBssgpDataVol*DlAveIpPacketLen/(BSSGP
header
overhead + DlAveIpPacketLen)
UlIpDataVol = UlBssgpDataVol*UlAveIpPacketLen/(BSSG header overhead

+ UlAveIpPacketLen)
If the CCCH packet signaling measurement result is divided by the IP data

volume, the following equations can be obtained:
DlImmAssOnAgchNumPerKbyteDl=DlImmAssOnAgchNum/DlIpDataVol
DlImmAssOnPchNumPerKbyteDl=DlImmAssOnPchNum/DlIpDataVol
PageNumPerKbyteDl=PageNum/DlIpDataVol
ChReqNumPerKbyteUl=ChReqNum/UlIpDataVol
UlImmAssPerKbyteUl=UlImmAss/UlIpDataVol
2006-04-05
All rights reserved
Page 37 of 48
Attachment 2 Capacity Planning Examples

This section provides examples on WAP/WWW Service
Hereunder analyzes the network QoS according to the examples on WAP/WWW
services. The following assumptions aim to analyze the block of WAP/WWW
services when the PDCH number is 2 and 4.
For WAP service, 750 bytes must be transferred within 1 second. That is, a
bandwidth of 6000 bps is needed. For WWW service, a bandwidth of 16K
bps is needed.
If the subscriber gives up the access when failing to get adequate

bandwidth, you should use the formula ERLANGB to calculate the GOS.
The GPRS subscribers are divided into pure WAP subscribers and pure
WWW subscribers. The total subscribers account 2% of the GSM
subscriber. The busy-hour bandwidth is 180bps.
The following table lists the results when PDCH = 2.

TRX Number
WAP Service
WAP GOS
Volume
WWW
WWW GOS
Service
Volume
1 (7)
0.0580
0.014%
0.0218
2.129%
2 (14)
0.1640
0.177%
0.0615
5.794%
3 (22)
0.2970
0.735%
0.1114
10.021%
4 (29)
0.4200
1.629%
0.1575
13.607%
5 (37)
0.5650
3.111%
0.2119
17.483%
6 (45)
0.7120
5.002%
0.2670
21.073%
7 (52)
0.8420
6.920%
0.3157
23.998%
8 (60)
0.9920
9.334%
0.3720
27.114%
Analysis:
According to previous sections, when the TRX number = 1 and 2, 2 PDCHs must
be configured. In this case, GOS = 2%. The maximum difference between the
GOSs listed line 1 and 2 in above table and 2% is about 4%, so the error is
limited.
The following table lists the results when PDCH = 4.
2006-04-05
All rights reserved
Page 38 of 48
TRX Number
WAP Service
WAP GOS
Volume
WWW
WWW GOS
Service
Volume
1 (7)
0.0580
0.000%
0.0218
0.023%
2 (14)
0.1640
0.000%
0.0615
0.178%
3 (22)
0.2970
0.001%
0.1114
0.555%
4 (29)
0.4200
0.003%
0.1575
1.060%
5 (37)
0.5650
0.013%
0.2119
1.818%
6 (45)
0.7120
0.037%
0.2670
2.736%
7 (52)
0.8420
0.080%
0.3157
3.650%
8 (60)
0.9920
0.166%
0.3720
4.801%
Analysis:
According to previous sections, when TRX number = 6, 7, and 8, 4 PDCHs must
be configured. In this case, GOS = 2%. The maximum difference between the
GOSs listed line 6, 7, and 8 in the above table and 2% is about 2%, so the error
is limited.
Attachment 3 Parameter Configuration for GPRS MS Cell

Update
If a GPRS MS performs the cell reselection when it is in GMMREADY state, it
must upload a packet for the cell update. After cell update, the GMMREADY
timer will restart. Therefore, for the GPRS MS in GMMREADY state, if the
interval for cell reselection cannot be controlled within GMMREADT timer, the
MS will repeat the cell update. In this case, the standby time of the MS will
reduce and the overhead at the network side will increase.
Two reasons can explain the phenomenon:
The MS is in an area where no strong cell can be found. In this case, the
MS will repeat the cell update.
When the MS keeps on moving, it will repeat the cell update when
performing even once data service (including routing area update).
To solve the former problem, you can set the cell reselection penalty time to a
value larger than GMMREADY timer.
2006-04-05
All rights reserved
Page 39 of 48
To sole the second later problem, you can set the GMMREADY timer to a value
smaller than the time for the MS to move across the cell. For example, if the
GPRS MS moves at a speed of 15 kilometers per hour, it will take 2 minutes for
the MS to move across of cell with a diameter of 500m. Therefore, you can set
the GMMREADY timer to 1 minute for the existing network at the first-stage
construction.
Generally, a timer judging and responding to the expiry is configured at the
GPRS application layer. To reduce the paging times (especially when the MS
takes GET service as the major service), you should set the GMMREADY timer
to a value larger than this timer.
Attachment 4 GPRS System Information Description

I. NMO (2)
NMO stands for network mode of operation.
NMO is related with paging channels used in a GPRS system. It is defined by Gs
interface and PCCCH.
When Gs interface is configured in the GPRS network, NMOI is required. In

this case, the MS can receive the packet paging and circuit-switched paging
through listening to only paging channel. When the MS is idle, it listens to
PCCCH or CCCH. When the MS is busy, it listens to PACCH (CS).
When no Gs interface and no PCCCH are configured in the GPRS network,
NMOII is required. In this case, the MS can receive the packet paging or
circuit-switched paging through listening to one CCCH despite it is idle or
busy.
When no Gs interface but a PCCCH is configured in the GPRS network,

NMOIII is required. In this case, the MS listens to the packet paging on
PCCCH and listens to circuit-switched paging on CCCH despite it is idle or
busy.
II. ACCESS_BURST_TYPE (1)

This parameter is broadcast in SI13 and PSI13. It decides if the PACKET
CHANNEL REQUEST message sent by the MS is in the format of 8-bit or 11-bit.
For details, see GSM046011.5.
III. CONTROL_ACK_TYPE (1)

This parameter sets the default format used by an MS in a CtrlAck message.
When it is 0, it represents 4 access bursts. When it is 1, it represents a
RLC/MAC control block. When the CtrlAck message works as a response
towards the Polling message, the Polling message must specify the CtrlAck type
2006-04-05
All rights reserved
Page 40 of 48
to be used; otherwise the related settings in the system information are to be

used. If the timing advance of the MS does not exist in the system and the type
of the CtrlAck message is 4 access bursts, the network side set the Polling bit
with the delivery of the downlink assignment message. In this case, the MS will
send the CtrlAck message in terms of 4 access bursts to export the timing
advance.
IV. BS_CV_MAX (4)

This parameter is the binary representation of the parameter BS_CV_MAX,
which is used by an MS to calculate the Countdown Value (CV). The initial value
of T3198 is set to BS_CV_MAX block period.
V. PAN_DEC (3), PAN_INC (3), PAN_MAX (3)

PAN_DEC sets the value for N3102 of the MS. When T3182 of the MS expires,
the N3102 will reduce by PAN_DEC. When the MS starts T3182 and receives a
packet uplink ack message, the N3102 will increase by PAN_INC. PAN_MAX is
the maximum value of N3102.
VI. ACC_CONTROL_CLASS (16)

This parameter stands for access control class. If an MS of Class N is accessforbidden, the corresponding bit is set to 0. Bit 11 means emergency call.
Bits: 16151413121110987654321
ClassN: 1514131211EC9876543210
VII. MAX_RETRANS (2)*4

This parameter defines the maximum times of radio priority class 14 of MS
respectively to retransmit a Packet Channel Access Request. Level 1 is the
highest radio priority level and level 4 the lowest. Each radio priority uses doubledigit to set the maximum retransmission times. The mapping relationship
between the retransmission times and the double-digit is as follows:
00-1; 01-2; 10-4; 11-7
VIII. S (4)
The parameter S calculates the minimum timeslot interval between two
continuous Packet Channel Access Request messages. For details, see
GSM046012.14.
IX. TX_INT (4)

The parameter TX_INT indicates the number of extended transmission timeslot
of random access.
2006-04-05
All rights reserved
Page 41 of 48
X. PERSISTENCE_LEVEL (4)*4
This parameter stands for the access persistence level and is broadcast in SI1.
Each radio access consists of 4 bits, indicating the access persistence level of
this radio priority level and the values ranges from 0 to 14. If this parameter is not
found, each radio access level is defaulted to 0.
This parameter defines the P (i) value of an MS whose radio priority is 1. PL2
defines the P (i) value of an MS whose radio priority is 2, and so on.
When network resource is limited, some MSs are restricted from sending
channel request message. In this case, the MS cannot perform packet service.
To be specifically, when an MS sends a Packet Channel Request message, it
will define a random number from {0, 1, 215}. If this random number is higher
than the PERSISTENCE_LEVEL, channel request is allowed; otherwise it is
forbidden.
XI. ALPHA (4)

This parameter calculates the PCH value of MS output power on each PDCH.
For open loop power control, the coding 1.0 is used.
This parameter is coded according to the following table.
bit
4321
0000
= 0.0
0001
= 0.1
0010
= 0.2
1001
= 0.9
1010
= 1.0
All other values are reserved.
XII. T_AVG_W (5)

It is a power control parameter, which is carried in SI13 and PSI13. It sets the
signal strength filter period for power control in packet idle mode. Its value is
2(k/2)/6multiframes, k=0, 1, 2...25. All other values are reserved.
2006-04-05
All rights reserved
Page 42 of 48
XIII. T_AVG_T (5)

It is a power control parameter, which is carried in SI13 and PSI13. It sets the
signal strength filter period for power control in packet transfer mode. Its value is
2(k/2)/6multiframes, k=0, 1, 2...25. All other values are reserved.
XIV. PAGE_MODE
This parameter is broadcast in PSI1, PSI3, and SI13 and is used to control the
paging modes, including normal paging, extended paging, paging recombination,
and Sameasbefore. It cannot be set at the operation and maintenance console,
for the system will control the paging mode at this time.
XV. PC_MEANS_CHAN (1)

It is a power control parameter carried in PSI13 and SI13. It indicates the
measurement power level received on a channel that is used for uplink power
control (0-BCCH, 1-PDCH)
XVI. N_AVG_I (4)

It sets the conflicted signal strength filter period for power control. Its value is
2(k/2), k=0, 115.
XVII. RAC (8)

This parameter stands for routing area code. It is allocated by carriers according
to network planning and is broadcast in PSI13 and SI13. RAI (routing area
identifier) consists of MCC, MNC, LAC, and RAC, in which RAC is 1 byte.
XVIII. SI13_LOCATION (1)

This parameter is the data field of PSI3, indicating the logical channels broadcast
in SI13.
0: SYSTEMINFORMATIONTYPE13messageissentonBCCHnorm
1: SYSTEMINFORMATIONTYPE13messageissentonBCCHext
XIX. NETWORK_CONTROL_ORDER (2)

This parameter is the information field of PSI13 and SI13. It decides how the MS
sends measurement reports in MM Ready state.
If the NETWORK_CONTROL_ORDER parameter = NC0, the MS controls

cell reselection but not send measurement repots to the network.
If the NETWORK_CONTROL_ORDER parameter = NC1, the MS controls

cell resection and sends measurement reports.
2006-04-05
All rights reserved
Page 43 of 48
If the NETWORK_CONTROL_ORDER parameter = NC2, the MS sends

measurement reports to the network without cell reselection. The network
controls the cell reselection
The above network control orders function only when the MS is READY state.
When the MS is in Standby state, only NC0 functions. For details, see
GSM050810.1.4.
This parameter is coded according to the following table.
Bit
21
00
NC0
01
NC1
10
NC2
11
Reserved
Attachment 5 Signaling Channel Planning Examples

The signaling channels are configured in Combined mode and Non-Combined
mode in the network.
Hereunder lists the assumptions for existing network capacity.
Immediate assignment messages/second is 5% that of the paging

messages/second.
A paging message can page 2 MSs.
Overload coefficient is 50%.
If the previous assumptions are present, for combined mode:
RACH capacity = 27TDMAframes/235ms = 114.7frames/s. That is, 114.7

channel request messages are sent by the MS per second, so the overload
coefficient is 114.7*50%=57.8 (messages/second).
For AGCH and PCH, there are 3 blocks per 235ms.
If the system supports dynamic signaling channel, it can page 24.2 MSs per
second (that is, 2* 95%*(3/0.2354) = 24.2). In this case, the overload
coefficient = 24.2*50 =% 12.1. That is, the BSC can deliver 12.1 messages
per second. Therefore, the BSC can deliver 0.6 immediate messages per
second (that is, 12.1*5% = 0.6).
If the system does not support dynamic signaling channel and AGCH: PCH
= 1:2, it can page 17 MSs per second (that is, 2*(2/0.2354) = 17). In this
case, the overload coefficient = 17*50% = 8.5. That is, the BSC can deliver
8.5 paging message per second. In this case, the BSC can deliver 4.2
immediate assignment messages per second (that is, 1/0.2354 = 4.2), and
2006-04-05
All rights reserved
Page 44 of 48
the overload coefficient = 4.2*50%=2.1. Therefore, AGCH capacity = 2.1 (8.5*5%) = 1.7 messages/second.
In Non-Combined mode:
RACH capacity = 51TDMAframes/235ms = 216.6frames/s. That is, 216.6

channel request messages are sent by the MS per second, so the overload
coefficient is 216.6*50%=108.3 (messages/second).
For AGCH and PCH, there are 3 blocks per 235ms.
If the system supports dynamic signaling channel, it can page 72.6 MSs per
second (that is, 2*95%*(9/0.2354)=72.6). In this case, the overload
coefficient = 72.6*50%=36.3. That is, the BSC can deliver 36.3 messages
per second. Therefore, the BSC can deliver 1.8 immediate messages per
second (that is, 36.3*5%=1.8).
If the system does not support dynamic signaling channel and AGCH:
PCH=1: 8, it can page 68 MSs per second (that is, 2*(8/0.2354) = 68). In
this case, the overload coefficient = 68*50%=34. That is, the BSC can
deliver 34 paging message per second. In this case, the BSC can deliver
4.2 immediate assignment messages per second (that is, 1/0.2354 = 4.2),
and the overload coefficient = 4.2*50%=2.1. Therefore, AGCH capacity =
2.1-(34*5%) = 0.4 messages/second.
Hereunder discusses the signaling amount increased by GPRS.

The followings are a series of assumptions:
According to traffic model, busy-hour attaches times of each subscriber are

0.5 times per hour; busy-hour detach times of each subscriber are 0.5 times
per hour; busy-hour PDP context activation times of each subscriber is 1
time per hour; busy-hour deactivation times of each subscriber is 1 timer per
hour; busy-time RA updates of each subscriber are 1.5 times per hour; cell
updates are 3 times per hour.
Num_Retrans_Channel_Req is 10%; Num_Retrans_ULImmass_AGCH is

10%;
Num_Retrans_DLImmass_AGCH
is
10%;
Num_Retrans_DLImmass_PCH is 10%.
The ratio of downlink immediate assignments on AGCH to that on PCH is

0%.
Each attach procedure, detach procedure, PDP context activation

procedure, PDP context deactivation procedure, and routing area update
procedure use one channel request, one uplink immediate assignment on
AGCH, and one downlink immediate assignment on AGCH respectively.
One cell update procedure uses one channel request and one uplink
immediate assignment on AGCH.
Signaling flow is still present if the subscriber receives short messages through
using either GPRS or GSM, so the signaling flow generated by short messages
2006-04-05
All rights reserved
Page 45 of 48
can be ignored here. In this case, the following conclusions can be obtained
based on the assumptions above:
Number
of
packet
channel
(0.5+0.5+1+1+1.5+3)*(1+10%)
request
=
8.25
messages
on
messages/hour
RACH
=
0.0023
messages/second.
Number of packet paging messages = 0 messages/second
Number of packet downlink immediate assignment messages on AGCH =

(0.5+0.5+1+1+1.5+3)*(1+10%)
8.25
messages/hour
0.0023
messages/second.
Number of packet uplink immediate assignment messages on AGCH =

(0.5+0.5+1+1+1.5+3)*(1+10%)
8.25
messages/hour
0.0023
messages
AGCH
messages/second.
Number
of
immediate
assignment
on
0.0023+0.0023=0.0046 messages/second.
Due to the increased signaling flow caused by data transfer, you can make the
following assumptions:
DlImmAssOnAgchNumPerKbyteDl = 1.33 messages/second
DlImmAssOnPchNumPerKbyteDl = 0 messages/second
PageNumPerKbyteDl = 0 messages/second
ChReqNumPerKbyteUl = 5.33 messages/second
UlImmAssPerKbyteUl = 5.33 messages/second
Num_Retrans_Channel_Req = 10%
Num_Retrans_Paket_Paging = 10%
Num_Retrans_ULImmass_AGCH = 10%
Num_Retrans_DLImmass_AGCH = 10%
Num_Retrans_DLImmass_PCH = 10%
Downlink rate of each WAP subscriber at busy-hour = 180bps
Downlink rate of each Internet subscriber at busy-hour = 1kbyte/s
Ratio of the number of Internet subscribers to that of WAP subscribers = 2%
Ratio between uplink data flow and downlink data flow = 1:4
If these assumptions are present, you can obtain the following conclusions:
Number
of
packet
channel
request
messages
on
RACH
5.33*((180/8)*0.25/1000)*(1+10%) = 0.033 messages/second
Number of packet downlink immediate assignment messages on PCH = 0

messages/second

1.33*((180/8)/1000)*(1+10%)=0.033 messages/second

5.33*((180/8)*0.25/1000)*(1+10%)=0.033 messages/second
2006-04-05
All rights reserved
Page 46 of 48
Number of immediate assignment messages on AGCH = 0.033+0.033 =

0.066 messages/second
As a result, the following conclusions can be obtained for each Internet

subscriber:
Number
of
packet
channel
request
messages
on
RACH
5.33*(1*0.25)*(1+10%)=1.47 messages/second

messages/second

1.33*1*(1+10%)=1.46 messages/second

5.33*(1*0.25)*(1+10%)=1.47 messages/second
Number of immediate assignment messages on AGCH = 1.46+1.47 = 3

messages/second
Therefore, the total increase amount of signaling flow for each WAP subscriber is
as follows:
Number
of
packet
channel
request
messages
on
RACH
0.033+0.0023=0.0353 messages/second

messages/second
Number
of
immediate
assignment
messages
on
AGCH
And the total increase amount of signaling flow for each Internet subscriber is as
follows:
Number
of
packet
channel
request
messages
on
RACH

messages/second
Number of immediate assignment messages on AGCH = 3 + 0.0046 = 3

messages/second
In this case, you can make further assumptions as follows:
There are 5 TRXs in a cell; the call loss rata is 2% at busy-hour; and the
traffic volume is 30 Erl.
The traffic volume of each subscriber at busy-hour is 0.025 Erl.
GPRS subscribers account for 2% of the GSM subscribers.
The attach rate of GPRS subscribers is 90%.
2006-04-05
All rights reserved
Page 47 of 48
The PDP context deactivation rate of the GPRS subscriber is 50%.
If these assumptions are present, the number of GPRS subscribers with PDP
activated in the system is 10.8. (30/0.025)*2%*90%*50% = 10.8). That is, 10.8
subscribers are using WAP service. Therefore, the number of Internet
subscribers is 0.21 (10.8*2%=0.2).
Some GPRS subscriber with PDP activated cannot monitor the PCH during data
transfer. In this case, 50% of the paging messages will be retransmitted.
Therefore, the total percentage of increased paging amount is 0.45%
(2%*90%*50%*50%=0.45%).
In this case, the total increased amount of the signaling flow within the system is
as follows:
Number
of
packet
channel
request
messages
on
RACH
10.8*0.035+0.21*1.48=0.7 messages/second
Paging messages generate an increase of 45% of the signaling flow.

messages/second
Number
of
immediate
assignment
messages
on
AGCH
10.8*0.07+0.21*3=1.4 messages/second
According to the previous analysis, the following conclusions can be obtained:
In combine mode, if the load of the existing network is lower than 114
messages per second (114.7-0.7=114), the RACH needs not be expanded.
In non-combined mode, if the load of the existing network is lower than
215.9 messages per second, the RACH needs not to be expanded.
If the system supports the configuration of dynamic signaling channel, you

can judge if the expansion is necessary by bounding the AGCH and PCH. In
this case, you should check the actual load of the network to see whether
the network can still accommodate the signaling flow generated by GPRS
service.
If the system does not support the configuration of dynamic signaling

channel, you should judge whether to expand the network capacity
according to the configuration and load of the existing network.
2006-04-05
All rights reserved
Page 48 of 48

GPRS&amp;EDGE Network Planning and Optimization-Chapter5 Network Planning-20040527-A-1.0

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

GPRS&amp;EDGE Network Planning and Optimization-Chapter5 Network Planning-20040527-A-1.0

Uploaded by

Copyright:

Available Formats

Document number

GPRS/EDGE Network Planning

Huawei Technologies Co., Ltd.

Chapter 5 GPRS&EDGE Network Planning

For internal use only

Chapter 5 GPRS/EDGE Network Planning

Promote balance development of voice service and data service

Fully utilize the existing GSM network resources

Satisfy GPRS/EDGE network requirements while assure GSM network

5.1.2 Effect of GPRS/EDGE Network against GSM Network

The GPRS/EDGE downlink power control mechanism is not mature, so the

An extra interference of 0 dB to 3 dB will be generated depending on the

If the PDCH is configured on BCCH, no new interference will be added in

If the CCCH is used of GPRS/EDGE access, the load on CCCH will be

GPRS/EDGE uses more flexible channel allocation strategies, so the radio

As GPRS/EDGE service develops, subscriber service (including voice

All rights reserved

Chapter 5 GPRS&EDGE Network Planning

For internal use only

The compatibility and inheritance of the existed GSM/GPRS network are

5.2 Traffic Model Establishment

Intermittent and bursting data transfer (The interval between two

Currently, the GPRS/EDGE network in China enables the following services:

Mobile phone and laptop online

WAP over GPRS/EDGE

Terminal installation-based services

Personal data assistant (PDA) terminal access

Short message over GPRS/EDGE

Private line access

Private network access

5.2.2 Data Rate Calculation

The SNDCP (SubNetwork Dependent Convergence Protocol) is not

LLC performs transfer with unacknowledged mode.

All rights reserved

Chapter 5 GPRS&EDGE Network Planning

For internal use only

RLC performs transfer with acknowledged mode and 10% of retransmission

An IP packet is 200 bytes in length in average.

IP data flows continuously within 10 IP packets at least.

A radio block is transmitted every 20ms.

Hereunder lists the facts for the data rate calculation:

In addition to the previous assumptions and facts, the following calculation

VGb = V_IP x (200 + 53)/200 = 1.265V_IP

A1 is the total bytes of the N*LLC PDUs.

A2 is the total bytes of the N*IP packets.

T is the time needed for the transfer of N*LLC PDUs.

Xindicates that X must be rounded up to the nearest integer.

Xindicates that X must be rounded down to the nearest integer.

All rights reserved

Chapter 5 GPRS&EDGE Network Planning

For internal use only

A1 = (200 + 9 + 4 + 3 + 1) x 10 = 2170 (bytes)

T = (73 + 73 x 0.2+73 x 0.1) x 20ms = 1880ms = 1.880s

V_IP = 200 x 10 x 8/1.88/1024 = 8.31 Kbps

VGb = 1.265V_IP = 10.51Kbps

The accurate estimation for GPRS/EDGE traffic model is unpractical. The

GPRS/EDGE traffic model must be made accordingly, which makes the

All rights reserved

Chapter 5 GPRS&EDGE Network Planning

For internal use only

planning possible. In addition, the introduction of new services may risk an

Subscriber model (total volume of data service, ratio of different services,

Radio transmission model (CS distribution, signaling overhead)

5.3 Coverage Planning

GPRS&EDGE Network Planning and Optimization-Chapter5 Network Planning-20040527-A-1.0

GPRS&EDGE Network Planning and Optimization-Chapter5 Network Planning-20040527-A-1.0