GPRS optimisation and Network visualization Janne Myllyl T-110.456 - - PowerPoint PPT Presentation

Janne Myllylä T-110.456

GPRS optimisation and Network visualization

Janne Myllylä T-110.456

Topics

• What do we need to know?
• Different types of information available
• Basics of GPRS capacity optimisation

Janne Myllylä T-110.456

Planning

• The network elements:
• type specific information (e.g. family, radiation patterns)
• current settings
• Geographic information
• Land use
• Building height
• Statistics
• Models

Janne Myllylä T-110.456

Planning

• Using the information we can estimate:
• Network capacities in different areas
• Overall service quality
• Affect of changes in the network
• Problems:
• Models work in a perfect world
• Map information is never up-to-date or accurate
• Butterfly effect

Janne Myllylä T-110.456

Are there more accurate methods?

• Network performance can also be measured
• Field measurements
• Network measurements

Janne Myllylä T-110.456

Optimisation basics

Nokia NetAct Measure Analyse Optimise Provision

Janne Myllylä T-110.456

Measurement types

• Call/Session
• Radio Quality
• Volume

Janne Myllylä T-110.456

Field measurements

+

• Basically a modified cellural phone is driven on a route.
• Reliable information available without much traffic volume
• Vendor independent
• Can measure competitors network performance
• A lot of driving around needed.
• Measurement sample time is very limited

Janne Myllylä T-110.456

Network measurements

+

• Almost all possible events are measured.
• Measurements span over a longer timeperiod
• Not very standardized. Different vendors measure and collect

slightly different data.

• Moderate traffic volume is needed for reliable measurements.
• The total amount of data is huge.

Janne Myllylä T-110.456

Busy Hour

• The distribution of traffic is not even. During weekdays

there occurs peaks in the network usage.

• Radio networks don’t generally react well to traffic increase
• According to common sence:

Network behaviour during the busy hour is the weakest link.

• Heuristics can be used to identify the bh.

Janne Myllylä T-110.456

What is visualized

• Network static information
• Locations & directions
• Parameter values
• Relations between elements

Janne Myllylä T-110.456

What is visualized

• There are dozens of raw measurements (Performance Indicator)

that are related to GPRS performance.

• User wants to see the result of a preliminary analysis based on

the raw measurements (Key Performance Indicator).

Janne Myllylä T-110.456

KPI

• Traditional benchmarks ( BER, FER, CSR, HSR )
• (E) GPRS data related
• Reliability, max probability of erroneous RLC
• Throughput, amount of RLC payload
• Delay, measured time between SGSN and mobile
• (E)GPRS load, timeslots utilized by GPRS service
• And many more

Janne Myllylä T-110.456

Visualizing KPI

• Snaphot of network state:
• Performance of network on map
• List of elements not behaving within thresholds
• Trend of measurements
• Time based comparison between different

elements / measurements

• Performance animations on map

Janne Myllylä T-110.456

Network capacity balancing

• In GSM network the available capacity is defined by

timeslots dedicated for different services.

• It is possible to dimension timeslot usage between
• SDCCH
• CS
• PS

Janne Myllylä T-110.456

Visualization of timeslot usage

Janne Myllylä T-110.456

Visualizing service performance

Janne Myllylä T-110.456

Visualizing cell level performance

Janne Myllylä T-110.456

Effect of timeslot redimensioning

1 2 3 4 5 6 3.1 10.1 17.1 24.1 CS PS SDDCH

%

The relevant analysis of service performance need to be continuous, since without increase of total capacity timeslot dimensioning is always compromise.

Janne Myllylä T-110.456

Treatment classes

• Assigning GPRS capacity for different service classes
• PoC
• Streaming
• Corporate
• MMS
• Diverse DL/UL QoS requirements.

Janne Myllylä T-110.456

Capacity and QoS

Capacity offered for various services SMS Speech GPRS

TREC 0 TREC 1 TREC 2 TREC 3

• Priorisation

Capacity Balancing QoS Priorisation

Janne Myllylä T-110.456

Running out of capacity

1 2 3 4 5 6 7 3.1 10.1 17.1 24.1 CS PS SDDCH

Dimensioning can now

• nly be used to increase

CS performance. The only way to improve PS performance is to increase the total capacity.

Janne Myllylä T-110.456

How to increase capacity

• Some of the traffic volume could be redirected to other cells
• A new serving cell can be setup
• TRXs can be added for the current cell(s) to increase

total amount of timeslots

• Impact matrix

Janne Myllylä T-110.456

Impact matrix

• Also known as Interference matrix
• All cells whose signal has been measured in serving

cells dominance area

• Handover possibility
• Used to determine which cells could cause interference

with serving cell.

Janne Myllylä T-110.456

Interference basics

• The frequencies have traditionally been planned using

reuse patterns and propagation models

• In order to increase the traffic capacity, the channel re-use

becomes tighter

• Too tight use of the same and adjacent channels

causes a decline of C/I

BER and FER increase,

worse coding schemes

Janne Myllylä T-110.456

Interference without hopping

• When no hopping is used some timeslots will constantly

have more problems than others.

• After too much reuse performance deteriorates quickly

Janne Myllylä T-110.456

Too tight reuse on map

Janne Myllylä T-110.456

Averaging behaviour

• Frequency hopping may be used to average network

behaviour

• Main idea is to reduce continuous bad performance

between mobile and bss.

Janne Myllylä T-110.456

Averaged behaviour on map

Janne Myllylä T-110.456

Hopping mode: BB

• In BB hopping TRX frequencies don’t change,

but TRX serving the mobile phone does.

• Total amount of frequencies in BB hopping is the same as the

number of TRXs.

• Also BCCH timeslots 1-7 are included in the hopping.

Janne Myllylä T-110.456

Hopping mode: RF

• In RF hopping TRX serving the mobile phone doesn’t change,

but TRX frequencies do.

• In RF hopping an allocation list contains frequencies

that are used.

• BCCH TRX is not hopping.
• N channels enables 64*N different hopping sequences.
• MAIO offset has as many values as allocation list has channels
• HSN can be selected from 64 different sequences.

Janne Myllylä T-110.456

Hopping mode comparison

BB RF BB RF BB RF

TRX-3 TRX-2 TRX-1

Mobile hops the same frequency pattern in both modes

Janne Myllylä T-110.456

Measured performance

0.5 1 1.5 2 2.5 3 3.5 1 2 3 4 5 BB RF

EFL DCR Basically RF hopping enables a more tight channel reuse

Janne Myllylä T-110.456

Extreme channel reuse

• Two types of service areas inside cell:
• Normal with regular reuse patterns (overlay)
• Small with extreme reuse (underlay)
• The same underlay frequencies are used even in neighboring

cells.

• Cell tries to make as much as possible of the traffic volume

to use the underlay frequencies.

Janne Myllylä T-110.456

Extreme channel reuse

• The same traffic volume can be managed with less

frequencies.

• With this example situation 3 underlay TRXs

could free 6 frequencies. underlay

Janne Myllylä T-110.456

References

3GPP TS 25.215 V6.0.0 Physical layer – measurements 3GPP TS 23.107 V6.2.0 QoS concept and architecture Halonen, Romero, Melero: GSM, GPRS and EDGE performance