Beruflich Dokumente
Kultur Dokumente
Version 5.0.2
© Copyright 2005 AIRCOM International Ltd
All rights reserved
ADVANTAGE, AIRCOM, ARRAY WIZARD, ASSET3g, CONNECT, DATASAFE,
ENTERPRISE, NEPTUNE, OPTIMA, QUALITA, RANOPT, TARGET and
WEBWIZARD are recognised trademarks of AIRCOM International.
Microsoft Word, Microsoft Office, Windows®, Windows 95™, Windows 98™,
Windows NT®, Windows XP® and MS-DOS™ are trademarks of the Microsoft
Corporation.
Other product names are trademarks of their respective companies.
This documentation is protected by copyright and contains proprietary and
confidential information. No part of the contents of this documentation may be
disclosed, used or reproduced in any form, or by any means, without the prior
written consent of AIRCOM International.
Although AIRCOM International has collated this documentation to reflect the
features and capabilities supported in the software products, the company makes no
warranty or representation, either expressed or implied, about this documentation,
its quality or fitness for particular customer purpose. Users are solely responsible for
the proper use of ENTERPRISE software and the application of the results obtained.
An electronic version of this document exists on our website.
This User Reference Guide finalised on 11 May 2005.
Refer to the Online Help for more information.
This User Reference Guide prepared by:
♦ ♦ ♦
Contents
Appendix A 2g and 2.5g Algorithms
Interference Table Algorithm 5
Interference and Connection Array Calculations 6
Worst Connection Array Calculation Method 7
Average Connection Array Calculation Method 8
Worst Interferer Array Calculation Method 8
Total Interference Array Calculation Method 9
Table of Default C/I BER Conversion Values 9
Frequency Hopping Algorithms 10
Synthesised Hopping Algorithm 12
Non-Frequency Hopping Algorithms 12
Automatic Frequency Planning (ILSA) 13
The Cost Function of the ILSA Algorithm 14
MAIO Planning Cost Function 14
GPRS and HSCSD Capacity Calculations 15
TRX Requirement - Circuit Switched Traffic and HSCSD 15
TRX Requirement - Circuit Switched, HSCSD and GPRS Traffic 15
Grade of Service and Data Rate 16
Channel Occupation Table 17
FCC Calculations 18
Frequency Calculations 20
Index
In This Section
Interference Table Algorithm
Interference and Connection Array Calculations
Frequency Hopping Algorithms
Non-Frequency Hopping Algorithms
Automatic Frequency Planning (ILSA)
MAIO Planning Cost Function
GPRS and HSCSD Capacity Calculations
FCC Calculations
Frequency Calculations
The values for area are obtained by averaging the probability of interference over the
region where A is the best server. The average is taken over all pixels in the
appropriate coverage array.
For traffic, the value to be averaged is the probability of interference x the traffic (in
mE) at that pixel. Thus it is necessary to have a traffic raster available to make this
calculation.
Note : The worst connection and the worst interferer calculations are the same in the
case of a non-frequency hopping network.
Where:
For each non-hopping carrier fi in the serving sub-cell, C/I(fi) is calculated.
For the hopping frequency group in the serving sub-cell, a single C/I(FH) is
calculated.
(2)
Where:
Note : The denominator in the equation above can never be zero ( and cannot both
be 0 at the same time). This is because ASSET3g does not allow you to set the total
number of TRX allocated to a sub-cell to zero, if at least one carrier layer is allocated.
If , then
If , then
Where:
For each (non-hopping) carrier fi in the serving sub-cell, C/I(fi) is calculated.
BERAV(serving cell) is calculated as the average BER(i) for all frequencies in the cell:
Where:
x Number of FH frequencies per TRX
mFH Number of FH frequencies/serving cell
nTRX Number of TRX/serving cell
BERAV(serving cell) is then converted back to dB to give C/I (FH)(serving cell).
Where:
and
PTOT = Average of all P(i) in the cell
Where:
= The adjacent channel interference caused on allocation i by allocation j (Units: 200*mE or 20,000*km2)
= The separation costs (from equipment, neighbours, exceptions or close separations) between
allocations i and j
= The handover count and intermodulation interference costs associated with allocation i
Where:
are sub-cells
is the C/I or C/A separation count for all TRX combinations on sub-
cells
The number of TS required ( ) for the CS traffic load ( ) given specified two
Grade of Services and a choice of Erlang table.
The number of TRX required is determined using the Channel to Transceiver Map by
increasing the number of TRX from 1 until the map’s is equal to or greater than
and is greater than or equal to .
The total number of TS required for CS and GPRS traffic ( ) can then be
determined using the average Circuit Switched TS requirement and the
channel occupation efficiency (e) as follows:
The number of TRX required and are determined using the channel to
transceiver map by increasing the number of TRX from the result of the previous
section until the number of available TS for traffic (NCS allocation) is equal to or
greater than .
Calculate the blocking for the CS traffic given the traffic load ( ) the
current allocation of TRX using the selected Erlang table.
HSCSD Blocking
Blocking is calculated from Erlang B or C using the number of HSCSD TS currently
allocated to the cell and the HSCSD load in timeslot Erlangs.
=timeslots allocated to CS
The GPRS data rate for the current allocation of TRX is determined by first
calculating the number of TS required for CS and HSCSD. The remaining TS are
available for GPRS. That is:
Where:
e is the efficiency from the Channel Occupation table determined from N
is the number of TS from the Channel Carrier Map for the current allocation of TRX
FCC Calculations
This section describes the algorithms used to calculate the data provided in the FCC
report.
Note : You will receive a warning if the Average Radial distance exceeds 40.2 km (79.1
km for Gulf of Mexico cells).
Used ERPS
This is the value of the transmitting ERP which is used in the calculations, it is the
Transmitting ERP subject to certain minima.
Used ERP is the maximum of:
• 0.1 W
• Maximum ERP/500
• Transmitting ERP for the radial
Frequency Calculations
Two frequency calculations are used when you create a Frequency Plan report.
Where:
REFF is the Effective Frequency Re-use for a subcell
NF is the total number of carriers available to hopping TRX on the subcell (note: this is
not the MA list length)
NTRX is the number of hopping TRX on the subcell
Frequency Load
The average frequency load is another approximate indication of the quality of the
hopping network.
It can be calculated for each subcell and also the average of these calculated to give a
figure for the network as a whole.
Where:
LFREQ is the Frequency Load of a subcell
UMTS Algorithms
This chapter describes the following topics:
In This Section
Notation for UMTS
List of Principal Symbols for UMTS
UMTS Basic Formulae
UMTS Uplink Noise Rise
UMTS Uplink Load
UMTS Frequency Re-Use Efficiency
UMTS Air Interface and User Bitrates
UMTS Shadow Fade Modelling
UMTS Power Control Error Modelling
UMTS Service Activity Modelling
UMTS Activity Factor Calculation For Packet Services (Web Model)
UMTS Transmit/Receive Diversity Modelling
UMTS Terminal Speed Modelling
UMTS Overview of a Snapshot
UMTS Scenario Prioritisation
UMTS Connection Evaluation
UMTS Blocking Probability
Uplink
Downlink
Pilot
Uplink (downlink) processing gain
Boltzmann constant
Pilot SIR
Temperature
Chip rate
(1)
(2)
(3)
(4)
(5)
(6)
(7)
(8)
(9)
(10)
(11)
(12)
This is expressed in dB in the Cell Uplink Report.
(13)
This is expressed as a percentage in the Cell Uplink Report.
(14)
This is expressed as a percentage in the Cell Uplink Report.
(15)
where is the standard deviation of the shadow fading at the pixel (in dB).
The above procedure is performed whenever a terminal is initialised at the beginning
of a snapshot. Fades for different terminals are uncorrelated, even if the terminals are
located in the same pixel.
Consumption of Resources
A successfully served circuit switched service will consume the same number of
resources regardless of the service activity factor. The number of resources in this case
depends only on the bearer used.
A successfully served packet switched service will consume a partial number of
resources depending on the service activity factor. For example, if a PS service is
served using a bearer that requires 2 resources and the activity factor is 1%, then 0.02
resources will be consumed.
Calculation of Throughput
The throughput of a successfully served service is calculated by multiplying the data
rate of the bearer used, by the service activity factor.
Calculation of Interference
Equations
Pαtotal
J = + Pαpilot
J + Pαcommon
J + Pαsync
J ( )
+ ∑ α ↓j + β ↓j Pαtraffic
Jj
j
(9)
(11)
Where:
= Average packet time period (s)
(16)
where is a random number taken from a normal distribution of zero mean and unit
variance.
Initialisation of Terminals
Gathering of Results
End of End of
Iteration 1 Iteration 3
End of End of
Iteration 5 Iteration 7
After the first iteration, the majority of “served” terminals fail to meet their
requirement. This is because terminals evaluated at the beginning of the first iteration
see little or no interference and so have their TX powers set to low values. By the end
of the first iteration, the noise in the system will have increased due to interference
from the newly served terminals. Hence terminals evaluated at the beginning of the
first iteration will no longer attain their desired by the end of the first iteration. In fact,
only the last terminal served is guaranteed to achieve its desired.
• of primary cell
• UL bearer
• DL bearer
The connection scenario being examined sets the candidate primary cell. This cell is
checked to see if it has a sufficient number of primary resources available, and to see
if it provides an adequate pilot SIR level at the terminal. If these conditions are met,
the cell is flagged as the primary cell of the candidate active set.
The remaining covering cells are evaluated to see if they can be handover cells. Cells
with a low downlink linkloss are checked before cells with a higher downlink
linkloss. A handover cell must have a sufficient number of handover resources
available, and provide an level that is within the handover margin of the level of the
primary cell. Each cell that satisfies these requirements is flagged as a handover cell
of the candidate active set unless the active set size limit specified by the primary cell
has been reached.
Check terminal power does not break noise rise limit on any
cells.
For each cell in the candidate active set, the terminal transmit power required to meet
the uplink is calculated. This lowest of these values is then quantised according to the
quantisation level specified for the terminal. We call the resulting power. The terminal
transmit power is temporarily set to, and the two best values on cells in the candidate
active set are calculated. The difference between these two values (in dB), together
with the terminal speed, allows the following quantities to be determined from the
tables that you supply in the Services dialog box
(17)
(18)
Iterate
until Calculate total achieved Eb N o at terminal assuming maximal
Eb/No
achieved ratio combining of links.
or not
changing
between
iterations
Increase/Decrease T if total achieved Eb N o at terminal is
too low/high.
Take a system with fixed capacity , and Poisson traffic with arrival rate users per
second and mean holding time seconds. The mean offered traffic is .
(19)
The probability that exactly C users are offered.
(20)
Lost Call Cleared: In an LCC system, blocked users do not try again.
(22)
Lost Call Held: In an LCH system, blocked users persistently retry until connected.
(23)
This is NOT a blocking probability and it should never be treated as one. The failure
rate can be an order of magnitude lower than both the LCC and LCH blocking
probabilities.
CDMA2000 Algorithms
This chapter describes the following topics:
In This Section
CDMA2000 Notation
List of Principal Symbols for CDMA2000
CDMA2000 Basic Formulae
CDMA2000 Uplink Noise Rise
CDMA2000 Uplink Load
CDMA2000 Frequency Re-Use Efficiency
CDMA2000 Air Interface and User Bitrates
CDMA2000 Shadow Fade Modelling
CDMA2000 Power Control Error Modelling
CDMA2000 Service Activity Modelling
CDMA2000 Activity Factor Calculation For Packet Services (Web
Model)
CDMA2000 Transmit/Receive Diversity Modelling
CDMA2000 Terminal Speed Modelling
PN Code Assignment Algorithm for CDMA2000
CDMA2000 Overview of a Snapshot
CDMA2000 Scenario Prioritisation
CDMA2000 Connection Evaluation
Calculation of Equivalent Control Overhead Factors for CDMA2000
CDMA2000 Blocking Probability
Uplink
Downlink
Pilot
Uplink (downlink) processing gain
,
Sector antenna gain
Boltzmann constant
Terminal TX power
Temperature
(2)
(3)
(4)
(5)
(6)
(7)
(8)
(9)
gain ( , ) is calculated by dividing the system chiprate by the air interface bitrate.
The User Bitrate is used purely to calculate traffic (data throughput) on a sector.
where is the standard deviation of the shadow fading at the pixel (in dB).
This procedure is performed whenever a terminal is initialised at the beginning of a
snapshot. Fades for different terminals are uncorrelated, even if the terminals are
located in the same pixel.
Consumption of Resources
A successfully served circuit switched service will consume the same number of
resources regardless of the service activity factor. The number of resources in this case
depends only on the bearer used.
A successfully served packet switched service will consume a partial number of
resources depending on the service activity factor. For example, if a PS service is
served using a bearer that requires 2 resources and the activity factor is 1%, then 0.02
resources will be consumed.
Calculation of Throughput
The throughput of a successfully served service is calculated by multiplying the data
rate of the bearer used, by the service activity factor.
Calculation of Interference
Equations:
(9)
and
(11)
Where:
= Average packet time period (s)
where is a random number taken from a normal distribution of zero mean and
unit variance.
s the number of first and second order neighbours with codes assigned
If the minimum code re-use distance is not selected in the parameters page then N
and NA are set to zero, the same applies to first and second order neighbours.
Initialisation of Terminals
Gathering of Results
End of End of
Iteration 1 Iteration 3
End of End of
Iteration 5 Iteration 7
After the first iteration, the majority of “served” terminals fail to meet their
requirement. This is because terminals evaluated at the beginning of the first iteration
see little or no interference and so have their TX powers set to low values. By the end
of the first iteration, the noise in the system will have increased due to interference
from the newly served terminals. Hence terminals evaluated at the beginning of the
first iteration will no longer attain their desired by the end of the first iteration. In
fact, only the last terminal served is guaranteed to achieve its desired .
• of primary sector
• UL bearer
• DL bearer
• UL radio configuration (CDMA2000 only),
• DL radio configuration (CDMA2000 only)
The connection scenario being examined sets the candidate primary sector. This sector
is checked to see if it has a sufficient number of primary resources available, and to
see if it provides an adequate level at the terminal. If these conditions are met,
the sector is flagged as the primary sector of the candidate active set.
Iterate
until Calculate total achieved Eb N t at terminal assuming
Eb/No
achieved
maximal ratio combining of links.
or not
changing
between
iterations
Increase/Decrease T if total achieved Eb N t at
terminal is too low/high.
If the downlink requirement can not be achieved, then the terminal fails to be
served, and all downlink powers are set to zero.
F1 Active Fundamental
Bearer
Inactive Fundamental F2
Bearer
α 1−α
(Active (Inactive
Period) Period)
The ratio of transmit powers for the active and inactive fundamental bearers
is given by the ratio of their ( ) requirements and processing gains as follows:
1α
α
(Active (Inactive
-
Period) Period)
1α
α
(Active (Inactive
-
Period) Period)
F1
Active Fundamental Bearer
α 1-α
(Active Period) (Inactive Period)
The ratio of transmit powers for the active and inactive fundamental bearers
is given by the ratio of their ( ) requirements and processing gains as follows:
α 1-α
(Active Period) (Inactive Period)
The ratio of fundamental powers to the power of the active supplemental bearer
and is given by the ratio of their ( ) requirements and
processing gains as follows:
Lost Call Cleared: In an LCC system, blocked users do not try again.
(21)
Lost Call Held: In an LCH system, blocked users persistently retry until connected.
(22)
This is NOT a blocking probability and it should never be treated as one. The failure
rate can be an order of magnitude lower than both the LCC and LCH blocking
probabilities.
HDR Algorithms
This chapter describes the following topics:
In This Section
HDR Notation
List of Principal Symbols for HDR
HDR Basic Formulae
HDR Uplink Noise Rise
HDR Uplink Load
HDR Frequency Re-Use Efficiency
HDR Air Interface and User Bitrates
HDR Shadow Fade Modelling
HDR Power Control Error Modelling
HDR Service Activity Modelling
HDR Transmit/Receive Diversity Modelling
HDR Terminal Speed Modelling
Overview of a HDR Snapshot
Scenario Prioritisation for HDR
HDR Connection Evaluation
Calculation of Uplink Equivalent Control Overhead Factor for HDR
HDR Notation
This list describes the notation symbols used in this section:
• A Greek subscript always indexes a carrier.
Uplink
Pilot
Uplink processing gain
Boltzmann constant
Terminal TX power
Temperature
Chip rate
(2)
(3)
(4)
(6)
(7)
(8)
processing gain. The uplink processing gain ( ) is calculated by dividing the system
chiprate by the air interface bitrate of the uplink bearer.
The User Bitrate of the uplink bearer is used purely to calculate traffic (data
throughput) on a cell.
where is the standard deviation of the shadow fading in the pixel (in dB).
This procedure is performed whenever a terminal is initialised at the beginning of
a snapshot. Fades for different terminals are uncorrelated, even if the terminals are
located in the same pixel.
where is a random number taken from a normal distribution of zero mean and
unit variance.
Initialisation of Terminals
Gathering of Results
HDR Iterations
An iteration involves sequentially evaluating the terminals (precisely once) to see if
they can make a connection to the network. After each terminal is evaluated, the noise
in the network is updated before moving on to evaluate the next terminal.
A terminal may connect to the network in a variety of different ways (connection
scenarios). For example, a terminal may have several different cells or carriers that it
may use. Each of the connection scenarios for a terminal is evaluated in turn until one
that allows a successful connection if found. If no scenario can produce a successful
connection to the network, the link powers for the terminal are set to zero and the
reasons for failure of the first scenario are recorded.
Terminals which fail to make a connection in an iteration are not removed from the
simulation, since success or failure in an iteration does not necessarily ensure the
same result in a subsequent iteration. In fact, the state (succeeded/failed) of a terminal
is determined purely by its state in the final iteration of a snapshot when convergence
has been achieved.
After the first iteration, the majority of served terminals fail to meet their
requirement. This is because terminals evaluated at the beginning of the first iteration
see little or no interference and so have their TX powers set to low values. By the end
of the first iteration, the noise in the system will have increased due to interference
from the newly served terminals. Hance terminals evaluated at the beginning of the
first iteration will no longer attain their desired by the end of the first
iteration. In fact, only the last terminal served is guaranteed to achieve its desired
.
Successive iterations produce increasingly accurate pictures of network noise and a
larger proportion of the terminals meet their requirement. By the seventh
iteration practically all the served terminals meet their requirement and the
system noise no longer changes significantly between iterations. The iteration have
converged to produce a plausible picture of served and failed terminals in the
network. Any remaining distribution in the achieved F values of served terminals is
largely due to quantisation of link powers or from specifying a non-zero power
control error standard deviation.
• Required
The rules for prioritising scenarios during connection evaluation are (in order of
decreasing importance):
• Higher (before lower) priority carriers (with respect to service)
In order to be served, the cell must provide an adequate level for the terminal.
The level determines the downlink bitrate that can be achieved by the
terminal. The mapping between and the downlink bitrate must be specified in
the HDR Downlink Parameters dialog box. If the level for the terminal is
lower than all the values specified in the table then the terminal will fail to be served.
The activity factor gives the proportion of time that the service is active. During an
inactive period, the terminal is assumed the transmit no power.
There is an uplink pilot channel whose strength is related to the uplink traffic channel
power.
T is the traffic channel power in the active period. P is the uplink pilot power in the
active period.
The average uplink power is given by:
Take a system with fixed capacity , and poisson traffic with arrival rate users per
second and mean holding time second. The mean offered traffic is .
(16)
(17)
(18)
Lost Call Cleared: In an LCC system, blocked users do not try again.
Lost Call Held: In an LCH system, blocked users persistently retry until connected.
(20)
This is not a blocking probability and it should never be treated as one. The failure
rate can be an order of magnitude lower than both the LCC and LCH blocking
probabilities.
where
m – gradient of the curve Eb/Nt vs (1-PER)
N – maximum number of reserved slots.
A uniformly distributed random number is drawn. If this is greater than p(ack) then
the physical layer packet is said to be successfully transmitted. If the random number
drawn is less that the probability of receiving an acknowledgement in the last
reserved slot then the entire physical layer packet is retransmitted.
The average IP packet transmission time, , can then be calculated using:
where
Slots- sum of all the slots used for transmission.
•
Hence the queueing delay depends on the IP packet arrival rate, departure rate and
the number of servers.
and Po is
The number of servers available to a service on a carrier will vary depending on the
number of services on that carrier. By using the above equations and the relation:
m=2
m=4
Where W is the maximum permitted waiting time in the queue. These equations are
solved for the maximum arrival rate which can be handled, , without exceeding the
maximum waiting time.
where
where
where
In This Section
Simulation Inputs for QoS Analysis
Traffic Generator for QoS Analysis
Preliminary Tests
Some conclusions can be deduced from the input data without running the simulation
at all. These are:
• 100% blocking on any service will result in delays building up to infinity
• Zero traffic on all services will result in zero delays
• Zero blocking on all services will result in zero delays
These results are immediately updated on the summary page of the QoS Analysis
dialog box.
The red line represents the mean number of users input from the Monte Carlo
simulation. The orange blocks represent the number of users varying over time. The
blue blocks represent the holding times of the packet sessions produced by the traffic
generator.
Little’s theorem gives us the relation between the arrival rate of packet sessions, the
mean number of users in the cell and their mean session holding time. Let
λ = mean session arrival rate
T = mean session holding time
Using the WWW traffic model, the mean holding time of a packet session is given
by:
T = (N pc − 1)D pc + N pc ( N d − 1)Dd
Packet Model
The traffic generator uses the session arrival and WWW models to produce a list of
packets for each service type, for each cell, for each carrier, lasting the duration of the
simulation. Each packet is stamped with its arrival time at the cell, and also keeps a
record of when it gets transmitted (its departure time), and its randomly generated
size. The packet service type lists are then merged and sorted in arrival time order, to
produce a single list of packets offered to the cell carrier:
In the diagram, the data contained in the packet boxes is the arrival time, the
departure time and the packet size. Initially, the packet’s departure time is set to be
the same as its arrival time. The departure time is updated each time step the packet is
queued, until it is successfully transmitted.
A histogram of the generated traffic is displayed for each service on each cell and
carrier in the graphs tab of the QoS Analysis dialog box.
In order to calculate the block size, the coding scheme allocated to each connection
needs to be input from the Monte Carlo simulation (a mean number of MS
connections per coding scheme, per bearer, per service type, per sub-cell array will be
required as input).
The block size can be inferred directly from the GPRS coding schemes, however, the
following mapping is used to calculate the block size for the first transmission attempt
for the link adaptation families:
• A – 592 bits
• B – 448 bits
• C – 352 bits
There are no default BLER versus C/I curves for MCS – 7, 8 and 9. In the
retransmission model, the lower bitrates of the link adaptation families are used.
Delay Class
GPRS has four different traffic classes. The following table shows that parameters that
specifies their QoS requirements:
Traffic Class Medium Application Data Rate (kbit/s) One-way Delay
Reliability Class
Applications can request different reliability classes, depending on their ability to
handle corrupt and duplicated blocks. The following table shows the reliability
classes that can be selected:
Reliability Class Lost Block Probability
1 10
2 10
3 10
where:
Throughput(C/I) = throughput in kb/s read off the throughput per timeslot graph for
the C/I achieved by the link
PeakDataRatePerSlot = peak rate per slot for the given coding scheme (the
asymptote of the throughput per timeslot graph
BLER(C/I) = block error rate for the C/I achieved by the link
The mean BLER over all the connections made per service type, per sub-cell is
required as an input from the Monte Carlo simulation, and is reported in the QoS
Analysis spreadsheet. Block errors also have implications for the retransmission
model. For more information, see Mean Retransmission Delay on page 102.
Throughput Class
Applications can request different mean and peak throughputs, in order to request
the desired throughput for bursty IP traffic. Peak throughput applies to short
intervals where the transfer rate is at a maximum. Mean throughput describes the
data transfer rate over an extended period of time, which could involve many idle
periods.
Peak throughput class Peak throughput (kb/s) Mean throughput class Mean throughput
(bytes per hour)
1 8 1 100
2 16 2 200
3 32 3 500
4 64 4 1 000
5 128 5 2 000
6 256 6 5 000
7 512*
8 1024* 17 20 000 000
9 2048* 18 50 000 000
*Data rate only reachable 31 Best Effort
with UMTS or EDGE
In GPRS, the peak throughput is determined by the peak data rate per slot achievable
by the coding scheme, and the number of timeslots for which the MS is enabled. The
peak throughput is calculated as follows:
PeakThroug hput = PeakDataRa tePerSlot * BlocksPerF rame * MaxNumberO fSlots
System Model
The call admission manager monitors the system's available capacity and
accommodates new packet transmission requests, at the same time ensuring the QoS
of existing connections. This may be situated at the BSC in a 2g network or the RNC
in a 3g network.
The steps of a connection admission procedure are:
• A new packet transmission request is received by the call admission manager
• The capacity of the destination cell is monitored
• The system either accepts or blocks the new connection
• If the QoS of an existing connection is degraded, it is dropped
Simulation Model
The simulation models the connection admission procedure by making the following
assumptions:
• The call admission manager monitors the cell capacity in every radio frame, that is
every 10ms
• The cell capacity for each service type is generated using the blocking probability
input from the Monte Carlo simulation
• The blocking decision is prioritised to accept new connections in the priority order
of their services
• The dropping of existing connections is not modelled
The cell capacity for each service is determined in each frame by generating a
uniformly distributed random number for each packet held in a queue. If the random
number is greater than the blocking probability, the packet starts transmission in that
frame. If the random number is less than of equal to the blocking probability, the
packet is delayed in the queue until the next frame.
The rule is then applied that if admissions for each service are considered in priority
order, and that if any higher priority packets remain queued, no lower priority
packets are admitted.
By the end of the simulation, the simulator will have produced a list of transmitted
packets, each stamped with its arrival and departure times from the cell.
A histogram of the queue length throughout the simulation is displayed for each
service on each cell and carrier in the graphs tab of the QoS Analysis dialog box.
k
Yi
Y =∑
i =1 k
σ2 =∑
k
(Y i−Y )
2
i =1 (k − 1)
2.σ
Hi =
m
Simulation Duration
This is calculated for each cell and carrier. The value depends on the parameters that
you have set for the services supported by that cell, and carrier, and the mean number
of users of those services input from the Monte Carlo simulation. Using the same
notation as the www traffic model section, plus the following definitions:
N req
= required number of packets
S req N req
= number of sessions required to generate packets
Treq S req
= time until the session arrives
D = recommended simulation duration
N pc .N d
Each session contains packets, so
N req
S req =
N pc .N d
(1)
The session arrivals are modelled as a Poisson process, and so the expected time until
S req
the session arrives is:
S req
Treq =
λ (2)
Substituting Little's law and equation (1) and (2),
N req .T
Treq =
N pc .N d .N
0 0.F C0
1 1.F C1
2 2.F C2
... ... ...
N n.F Cn
... ... ...
N N.F CN
N
TP = ∑ ci
i =0
∑c i
CP(n) = i =0
TP
∑ (F .n − D ) .P(n)
N
σ=
2
n =0
Standard deviation
⎛ BLER ⎞
Mean retransmis sion delay = 0.01.⎜⎜τ rt + 1⎟⎟seconds
⎝ (1 − BLER ) ⎠
References
The following are documents that have been referred to throughout this chapter:
• “Selection procedures for the choice of radio transmission technologies of the
UMTS” TR 101 112 v3.2.0, p.34
• “Quality of Service for Multimedia CDMA”, N. Dimitriou, R. Tafazolli, G. Sfikas,
IEEE Communications Magazine, July 2000
• “Simulating Computer Systems”, M.H. MacDougall, MIT Press, p.114
• “Introduction to Mathematical Statistics”, R.V. Hogg and A.T. Craig, Collier-
Macmillan Ltd, p.193
In This Section
Simulation Array File Formats
Live Traffic File Formats for 2g Networks
Live Traffic File Formats for 3g Networks
3gr Files
Advantages Disadvantages
The fact that the file contains everything from the simulation Because the file contains everything it is large
means you can load the file on a PC anywhere and run it, even
if it is from a completely different database
Can be loaded and saved from the Array Manager You can only have one 3gr file loaded at any one time
File Header
This table describes the header of the 3g archived array format (*.3ga):
Size Type Description Comments
4 Bytes int Magic Number should be 0x02121975
4 Bytes int Version Number Currently 4202
4 Bytes int Archive Name Length inc NULL terminator
char[ ] Archive Name User Visible Name
4 Bytes int Network Type Enumeration of one of the following:
NETWORK_UNKNOWN 0x00000000
NETWORK_UMTS 0x00000001
NETWORK_IS95 0x00000002
NETWORK_HDR 0x00000004
NETWORK_ALLTECHS 0xFFFFFFFF
4 Bytes int Region xMin
4 Bytes int Region xMax
4 Bytes int Region yMin
4 Bytes int Region yMax
4 Bytes int Resolution
4 Bytes int Memory Usage Mb
4 Bytes int Unique Name String Length inc NULL terminator
char[ ] Unique String GUID
4 Bytes time_t Date / Time
4 Bytes int User Name String Length inc NULL terminator
S
i
z
e
4 Bytes int Magic Number should be 0x21081970
4 Bytes time_t Date / Time
4 Bytes int Generic Name String Length inc NULL terminator
char[ ] Generic Name String
4 Bytes int Instance Number
4 Bytes int Carrier/Service Name Str Len inc NULL terminator
char[ ] Carrier/Service Name String
4 Bytes int Data Array Type currently 0 = float
4 Bytes int Data Array Num Elements
TYPE[ ] Data Array Type specified above
int[ ] Index Array
4 Bytes int String Array Num Elements
4 Bytes int String Length inc NULL terminator
char[ ] String
4 Bytes int Legend Colours Array Length
COLOREF Category Colour } Repeated
Int CategoryString Length } for each
UMTS
UMTS Cell Name (white-space) UMTS Traffic (Terminals)
CDMA 2000
CDMA 2000 Sector Name (white-space) CDMA 2000 Traffic (Terminals)
HDR
UMTS Cell Name (white-space) UMTS Traffic (Terminals)
TD-SCDMA
CDMA 2000 Sector Name (white-space) TD-SCDMA Traffic (Terminals)
The *.cbd file consists of a table of bearer traffic values for each cell.
A *.cbd file should contain only be one row per cell. If duplication is detected during
import then a warning message is given in the message log. For example: “CellXXX:
Skipping duplicate bearer traffic data.”
Whilst the traffic wizard is processing the *.cbd file, as multiple network bearers can
map to the same tool bearer, the application should generate a total for each tool
bearer.
If the import detects that values are missing, then a value of 0.00 Erlangs should be
assumed. A warning message should be added to the message log in those cases. For
example: “CellXXX: BearerXXX traffic not specified. Defaulting to 0.0”
If there are no data values for the bearers, then 0 should be assumed - but a warning
message in the message log indicates that this has occurred. For example: “CellXXX:
BearerXXX traffic not specified. Defaulting to 0.”
The cell matching criteria will vary according to the Naming Method specified in the
*.cbc file.
If no match occurs between the Cell ID and that in the database, a warning message is
displayed in the message log stating that. For example: “CellXXX: Failed to import
traffic. Cell does not exist.” (ID as appropriate to naming method)
As there can be up to 15,000 cells contained in the *.cbd file, a progress bar with an
Abort button appears if the processing contains more than XXXX cells. XXXX is
determined during testing to appear after 2 seconds.
The *.cbc file format allows a '-' instead of a Bearer ID to mean just ignore that
column.
Index S
Snapshots
algorithm • 32
T
A TETRA
algorithms • 5
Algorithms The main limitation on • 98
CDMA2000 • 43
FCC calculations • 18
Frequency hopping • 10 U
Frequency Re-use and Load • 20 UMTS
GPRS and HSCSD capacity • 15 algorithms • 23
HDR • 69
ILSA cost function • 13, 14
Interference arrays • 6
Interference Tables • 5
MAIO planning cost function • 14
Non-Frequency hopping • 12
Packet QoS • 89
UMTS • 23
C
CDMA2000
algorithms • 43
F
File formats
live traffic for 2g • 108
live traffic for 3g • 109
Simulation arrays • 105
G
GPRS
algorithms • 5
GSM
algorithms • 5
H
HDR
algorithms • 69
HSCSD
algorithms • 5
M
Monte Carlo, algorithm • 23, 32, 34, 35, 39
P
Packet Switched QoS algorithms • 89