Date post: | 03-Apr-2018 |
Category: |
Documents |
Upload: | tahir-muneer |
View: | 218 times |
Download: | 0 times |
of 34
7/28/2019 02-White Paper - Analysis of GPRS Services Rev 1.1
1/34
MM/ACTIX/20040628/01 Copyright Actix 2004
White paper Analysis of GPRS services Page 1 of 34
White paperAnalysis of GPRS services
Date: 01/07/2004
Revision: 1.1
Reference:
Actix Limited,200 Hammersmith RoadHammersmith, LondonW6 7DL, UKTel: +44 (0) 208 735 6333
Fax: +44 (0) 208 735 6301www.actix.com
COPYRIGHT Actix Ltd 2004
7/28/2019 02-White Paper - Analysis of GPRS Services Rev 1.1
2/34
White paper Analysis of GPRS services Page 2 of 34
Document Control Information
Document Details:
Filename White paper on GPRS services
Revision 1.1
Release Date
AuthorsJ arkko Lapinlampi,
Massimiliano Mannelli
Classification External
Revisions of This Document:
Date Reference Revision Authors Comment
28/06/2004 MM/ACTIX/20040628/01 1.0 J L, MM Initial Version
01/07/2004 NR/ACTIX/20040701/01 1.1 NJ R Edited Version, changed title
Referenced Documents:
Date Reference Revision Authors Comment
7/28/2019 02-White Paper - Analysis of GPRS Services Rev 1.1
3/34
White paper Analysis of GPRS services Page 3 of 34
Contents
1 INTRODUCTION............................................................................................................... 4
2 SERVICE ANALYSIS ....................................................................................................... 5
3 AGGREGATION LEVELS AND TRACKERS .................................................................. 5
3.1 Subscriber level.................................................................................................................. 73.2 Context level ....................................................................................................................... 73.3 Service level ....................................................................................................................... 73.4 Task level ............................................................................................................................ 73.5 Packet level ......................................................................................................................... 8
4 TRIGGER POINTS AND KEY PERFORMANCE INDICATORS...................................... 8
5 ANALYSIS EXAMPLES ................................................................................................. 12
5.1 Statisti cal analysi s of the traces ......................................................................................125.1.1 Sample Gb file......................................................................................................... 125.1.2 Sample Gn file......................................................................................................... 16
5.2 FTP Transfer ......................................................................................................................195.3 HTTP Session ....................................................................................................................24
APPENDIX A .............................................................................................................................. 27
APPENDIX B .............................................................................................................................. 32
7/28/2019 02-White Paper - Analysis of GPRS Services Rev 1.1
4/34
White paper Analysis of GPRS services Page 4 of 34
1 Introduction
For many operators, GPRS-based services are a key element of their strategies to boost
average revenue per user. The convergence of wireless networks and IPnot having
been originally conceived as a wireless data protocolbrings a unique challenge tooperators of GPRS networks. As new services and devices for transmitting data are
launched onto the market, and traffic increases, more pressure will be placed on
monitoring and optimizing GPRS networks.
In the market there are many products that allow users to monitor radio or network
performance (drive test tools and protocol analysers, although with poor KPIs creation
and troubleshooting capability), but none that combine that with a clear vision of the user
perception of services with the availability of indicators to dig down into the service
behaviour. Actix Solutions (SVS and IVS) for GPRS and EDGE now support KPIs and
detailed messaging, thereby offering this functionality.
This White Paper gives an insight into the structure of the new trackers, attributes and
events. It starts with an overview of the different levels of investigation, and goes on to
document the definition of the single parameters and processes. A full set of practical
real-life examples will enable users to become confident with the new capabilities.
The final aim is to offer a level of detail that gives an unrivalled flexibility on the type of
analysis. Users will be able, in the same tool, to have a very high-level overview of
service utilisation and user behaviour, as well as to be able to do a very low-level drill
down on messages belonging to a specific service or TCP/IP session.
7/28/2019 02-White Paper - Analysis of GPRS Services Rev 1.1
5/34
White paper Analysis of GPRS services Page 5 of 34
2 Service analysis
Performance of data services depends on the interaction of many entities and different
protocol levels. Flexibility in data filtering and aggregation possibilities is necessary, so
Actix solutions allow:
- Data aggregation per single users, PdP Contexts, service sessions, single IP
connections and many others.
- Application of generic KPIs to different interfaces and services, thereby
standardising queries and reports.
- Analysis from the service (HTTP, MMS etc.1) to TCP/IP transport layer and
further down to GPRS specific protocols.
In the following sections, these elements will be analysed in more detail.
3 Aggregation levels and trackers
The aggregation of measurements to provide metrics can be done in two different ways:
- From a network point of view (so per network elements)
- From a 'user experience' point of view
The first case applies when we want to maximise the network performance and
efficiency, the second when we want to achieve an optimal user experience (QoS as
perceived by the customers). Of course, the best optimisation needs to find the trade-off
between the two.
This section specifies the trackers (i.e. identifiers attached to every message belonging
to the particular session) used to aggregate performance indicators, thus measuring the
user perception of services.
1As from IVS release notes, the services that have predefined KPIs are: FTP, HTTP, MMS, WAP and
ICMP. SMTP and POP3 are decoded and listed in summary queries but no KPI information is currentlyavailable.
7/28/2019 02-White Paper - Analysis of GPRS Services Rev 1.1
6/34
White paper Analysis of GPRS services Page 6 of 34
The following table provides the list of all the available trackers and their applicability:
Tracker Um Gb Gn Gi Server
Subscriber Id - Yes Yes - -
Mobile IP Address Yes Yes Yes Yes Yes
PdP Context Id No Yes Yes - -
Service Id Yes Yes Yes Yes Yes
Task Id Yes Yes Yes Yes Yes
Packet Session Id(TCP/UDP/WTP)
Yes Yes Yes Yes Yes
IP Session Id Yes Yes Yes Yes Yes
Table 1 Trackers per interface
IMPORTANT: the trackers are available in traces containing IP packets:
- IP sniffer files, collected on any interface from R (between TE and MT) to the
server side2
- Protocol analyser files3
2The traces outside the GPRS network must be based on a protocol stack similar to the Gi interface.
3For the complete list of protocol analysers and interfaces supported see IVS release notes or contact
Actix support.
7/28/2019 02-White Paper - Analysis of GPRS Services Rev 1.1
7/34
White paper Analysis of GPRS services Page 7 of 34
3.1 Subscr iber level
The actual user represents the highest aggregation level. In fact, this is the focal point to
define the customer experience. The identifier is given by the IMSI (mapped to the
attribute Subscriber_Session_Id in Actix solutions). It is valid from the access to the
network (attach procedure) till the detach (detach procedure).
3.2 Context level
This level is triggered when the user accesses the services. The Context_Session_Id
starts when an attached user opens a context (PdP Context activation) to access
services and ends with the context deactivation.
3.3 Service level
Every service entity is defined as the self-consistent and usable result of an action
performed by the user. In practical terms it can be:
- a connection to an FTP server to trigger the download or upload of a group files
- the download of a web or WAP page
- the transfer of a picture message or other MMS transactions
- a complete e-mail
The Service_Session_Id is created for the scope.
3.4 Task level
This level defined the single elements defining a service. In some cases, the two levels
can coincide (for a single object WAP page or a PING the task coincides with the
service entity).
Examples of tasks (Task_Session_Id) are:
- the single file downloads triggered by a retrieve
- the single objects of a web page (text, icons and pictures)
- the text and the attachments of an e-mail
- an object of an MMS session
7/28/2019 02-White Paper - Analysis of GPRS Services Rev 1.1
8/34
White paper Analysis of GPRS services Page 8 of 34
3.5 Packet level
Single packets from transport protocols constitute this level. This is too detailed for KPI
definition, but is useful for troubleshooting and optimization (when drill-down is required).
The protocols that are implemented for this level are:
- TCP
- WTP
- ICMP
UDP is currently not implemented.
4 Trigger points and Key Performance Indicators
Every entity defined above can be subdivided in a setup phase, an activity phase and a
release phase, and that applies to any investigation level.
For example, for the user level, the setup corresponds to the attach procedure, the
activity phase is the time where the user remains attached, and the release phase
corresponds to the detach procedure. At the lower level, the setup is, for instance, the
signalling phase of the HTTP protocol for a single object of a web page and the activity
phase is the actual object download.
The following diagrams (Figure 1, Figure 2 and Figure 3) define the:
- Interactions among service, task and packet levels.
- Trigger points for the different events (set-up, data transmission and release
phase, when applicable).
- Measurements and KPIs available in the different phases (signalling times,
success and failure events, data volumes and throughput, retransmissions,
durations).
7/28/2019 02-White Paper - Analysis of GPRS Services Rev 1.1
9/34
White paper Analysis of GPRS services Page 9 of 34
Figure 1 - Session levels and attributes
7/28/2019 02-White Paper - Analysis of GPRS Services Rev 1.1
10/34
White paper Analysis of GPRS services Page 10 of 34
Figure 2 - Service Session durations (all attr ibutes have a prefix Service_)
Figure 3 - Packet Session durations (all attributes have a prefix Packet_)
The attributes are organised in groups (see Figure 4):
- General attributes (like the session identifiers).
- Key Performance Indicators (like overall throughput with and without
retransmissions, retransmission percentage, initialisation time).
- Measurements (like delta and cumulative packets, bytes, throughput with and
without retransmissions, times).
- Events and events time (also failure causes are included here).
7/28/2019 02-White Paper - Analysis of GPRS Services Rev 1.1
11/34
White paper Analysis of GPRS services Page 11 of 34
Figure 4 - Applications attribute tree
Every level sets similar names for these parameters (the prefix distinguish them). That
gives a broad, comprehensive and easy-to-understand set of KPIs or attributes that can
be aggregated in KPIs. Reports and queries can be designed on top of them. Appendix
A gives details for every single indicator definition, while in Appendix B the formula of
every single aggregated value is provided.
7/28/2019 02-White Paper - Analysis of GPRS Services Rev 1.1
12/34
White paper Analysis of GPRS services Page 12 of 34
5 Analysis examples
This section provides example analyses using the indicators and trackers presented in
the previous sections. The aim is to show how it is possible to use the information
described in the previous sections to:
- Provide a statistical view of the content of the traces (users, traffic, events etc.).
- Identify quickly those cases with poor performance.
- Drill down into the causes and circumstances that generated the problems.
5.1 Statistical analysis of the traces
In this section, we will analyse a Gb file and a Gn file. A similar approach could be
followed with traces collected from other interfaces (compatible with the trackers
available, see Table 1).
The scope of this example is to analyse the services. For an analysis of the network
level (MM, SM, LLC or BBSGP procedures), please refer to the IVS 1.6 release notes.
5.1.1 Sample Gb file
Loading a Gb file and importing the queries included in Subscriber and Context
queries.aqf4, it is possible to make a statistical analysis of the attached users and the
active ones in terms of contexts and applications.
The queries presented below can be also used to generate a summary report, since
they provide information on all levels:
- Subscriber
- PdP Context- Service
- Task
- Packet (normally too detailed for a statistical analysis)
4The queries can be found under: C:\Program Files\Actix\Analyzer\Queries\Subscriber and Service
Analysis (IVS release 1.6)
7/28/2019 02-White Paper - Analysis of GPRS Services Rev 1.1
13/34
White paper Analysis of GPRS services Page 13 of 34
The first query gives statistics of the attached subscribers (124 unique IMSIs5 are in the
trace):
Figure 5 Statistics of attached subscribers
Clicking on the query Active subscribers per application, we can see that only 5 of
them are active6 in the trace (1 using FTP, generating most of the data volume, and the
others doing WAP):
Figure 6 Active subscribers per application
5Sometimes the traces contain only parts of user sessions (for example they do not include the attach or
the PdP Context activation); IVS tracks the sessions anyway but assigns progressive internal identifiers,substituting the IMSI if it is found at some point in the trace, otherwise leaving the internal ID.6
A subscriber or other dimension is defined Active when at least one packet is transmitted during thesession.
7/28/2019 02-White Paper - Analysis of GPRS Services Rev 1.1
14/34
White paper Analysis of GPRS services Page 14 of 34
With the query Active Context per trace, we get the list of active PdP Contexts:
Figure 7 Active context per trace
Any of these queries can be used to filter on a specific quantity (a subscriber or a
context) for more detailed analysis.
Using the queries activated by the scenarios, it is also possible to extract the
subscribers in separate sub-streams. See the IVS release notes and online help for
details.
A second step in the analysis is given by the queries on service, task or packet level.
The service summary shows the events and data throughput and volumes. It is clear
from it that most of the sessions, although generating traffic, have been abnormally
terminated (see statistic #Service abort):
Figure 8 Service summary
7/28/2019 02-White Paper - Analysis of GPRS Services Rev 1.1
15/34
White paper Analysis of GPRS services Page 15 of 34
We need then to have a better insight on the causes, selecting the query Subscriber
and service report:
Figure 9 Subscriber and service report
From that we can appreciate not only the causes of the aborts, but also KPIs like the
time taken to connect to the server, the total duration of the connection, or the part
dedicated only to data transfer (thus cutting out the idle time and signalling), so that
usage patterns can be identified.
Scrolling on the right, we can see all the other KPIs (throughput values, data volumes
and retransmissions, in number of packets and percentage):
Figure 10 Other KPIs
7/28/2019 02-White Paper - Analysis of GPRS Services Rev 1.1
16/34
White paper Analysis of GPRS services Page 16 of 34
The throughput values represent the throughput average of the single tasks. The
Service summary query includes also the average of all the single throughput
measurements during the service (so the longest tasks have a higher influence on the
resulting value, whilst in the first case they all contribute in the same way).
In this case, the task level is not adding a lot, because there is a single download per
connection to the FTP server (that is not true in case of mget or mput with multiple files),
and the cause of abort is still the Packet Data Session Ended:
Figure 11 Service summary
We have seen how to summarise in a single click the content of the trace, and then to
refine the statistics to a higher level of detail.
There are some problems related to the services used in the file so the investigation canproceed with a drill-down. The follow-up of the investigation (possible in the same
session) will be performed in the next sections (5.2 and 5.3).
5.1.2 Sample Gn file
Let us repeat the statistical analysis, but this time with a Gn file. The important thing is
that the same queries and the same approach can be used because we are addressing
information belonging to the IP protocol layer or higher, which are common across the
interfaces.
As previously said, the queries are applicable seamlessly on every interface, so the look
and feel is the same as the one analysed for Gb.
7/28/2019 02-White Paper - Analysis of GPRS Services Rev 1.1
17/34
White paper Analysis of GPRS services Page 17 of 34
There are 224 IMSIs attached:
Figure 12 Subscribers per trace
of which 33 are active (31 on HTTP and 2 doing Pings, next picture). The logged trace
is quite short; one effect of this is that there are many sessions, of which the start ofsignalling and data transfer is tracked, but not the conclusion (that explains the
discrepancy between the counters).
Figure 13 Active subscribers per application
7/28/2019 02-White Paper - Analysis of GPRS Services Rev 1.1
18/34
White paper Analysis of GPRS services Page 18 of 34
From the PDP Context query, we can see that there several open contexts for other
services, but they are not active in the trace (no packets are tracked):
Figure 14 PDP Context
If we analyse the WAP sessions, we can see that 3 of them are aborted because of a
connection redirect:
Figure 15 Service report
For a drill-down, Packet report query can be selected and a single WAP session
filtered:
Figure 16 Packet repor t
7/28/2019 02-White Paper - Analysis of GPRS Services Rev 1.1
19/34
White paper Analysis of GPRS services Page 19 of 34
Then tables using WSP and WTP attributes and the Protocol Stack Browser can be
used:
Figure 17 Protocol Stack Browser
5.2 FTP Transfer
This example describes an analysis of a file containing some FTP data transfer. We will
see how to use KPIs and queries introduced in the previous sections to separate
successful cases from unsuccessful cases, and how to analyse unsuccessful cases step
by step, finding the root cause of the problem.
To figure out what is going on, we open a crosstab query (Service summary) which
shows summary information of traffic dimensioned by Protocol type and ServiceSession. That query (Figure 18) shows only an overview of Service Sessions, namely
Service Start and End events and KPIs.
7/28/2019 02-White Paper - Analysis of GPRS Services Rev 1.1
20/34
White paper Analysis of GPRS services Page 20 of 34
Figure 18 - FTP example, Service summary
We can see that there are two Service Session Starts, but only one Service Session
End. Also, we find that Service Abort has happened. When we open individual FTP
Sessions (Figure 19), we find two Service Sessions. The first Service Session ends
correctly, but the second one has no Service End but a Service Abort. Lets focus on the
second Service Session.
Figure 19 - FTP example, Service Abort
We can use the filter functionality of the Statistic Explorer to narrow down the analysis to
the messages of the second Service Session. Lets use the Service report query to
investigate that Service Session. From that query, we can see that the cause of abort
was Packet Data Session Ended7 (Figure 20).
Figure 20 - FTP example, Service details
7Another possible cause for abort is Task Session Timeout.
7/28/2019 02-White Paper - Analysis of GPRS Services Rev 1.1
21/34
White paper Analysis of GPRS services Page 21 of 34
The next step is to investigate individual tasks, and find out which task caused the
problem. From the Task report (Figure 21) view, we see that there are three FTP tasks,
and one of these is aborted. Also, other information on tasks is presented, although we
will not consider it in this analysis.
Figure 21 - FTP example, Task level
We will go further down to the packet level using a Packet report query. We can see that
the initiator of the abort was the FTP Server (Figure 22).
Figure 22 - FTP example, Packet level
Using Internet level attributes and the Protocol Stack Browser, we can investigate at a
more detailed level what really happened (Figure 23). The FTP Client has asked to store
a file (FTP_Event = STORE), and a new TCP Data connection was opened.
TCP_Event_ID attribute describes how TCP connection proceeds. Here we can see that
there has been normal data transfer, one retransmission and then the FTP server sends
an abort. Checking that packet from the Protocol Stack Browser, we can see that the
TCP Reset bit is set. By this way, it is possible to investigate all aspects of TCP level,
and if necessary, go still further down to IP level.
7/28/2019 02-White Paper - Analysis of GPRS Services Rev 1.1
22/34
White paper Analysis of GPRS services Page 22 of 34
Figure 23 - FTP Example, TCP level
The event engine uses internal Event Diagrams to generate application level information
and events. Diagrams calculate part of attributes and events, but they are also a visual
aid to internal troubleshooting8. For example, to find out how FTP Session proceeds,
FTP Diagram shows main states and transitions between states (Figure 24). There are
diagrams for Service and Task Sessions too, so one can start the investigation usingthese diagrams, and proceed to lower levels if necessary. In this example, one may
notice that an Abort happens from the Transfer state to the End State, which means that
whole FTP Session is aborted. If the transition was from the Transfer state to the Ready
8 Currently not released in the product.
7/28/2019 02-White Paper - Analysis of GPRS Services Rev 1.1
23/34
White paper Analysis of GPRS services Page 23 of 34
state, only that particular task was aborted (e.g. the user cancelled the current file
download).
Figure 24 - FTP Example, FTP Diagram
7/28/2019 02-White Paper - Analysis of GPRS Services Rev 1.1
24/34
White paper Analysis of GPRS services Page 24 of 34
5.3 HTTP Session
This example shows an analysis of HTTP Sessions. We start the investigation from the
Service Report query, which shows us that there are two HTTP Sessions, which both
start and end gracefully (Figure 25). We can see also the number of tasks and service
initialization time. Initialization time (Service_Duration_Session_Initialization) is an
important attribute to see how fast a service provider responds to a service request. In
the HTTP case, it is the time it takes to create a first TCP connection to the server.
HTTP task is defined as a download or upload of a particular object of web page, so it
starts when HTTP Client sends a GET method (or any other method specified in
RFC2616) to the server.
Figure 25 - HTTP Example, Service report
When scrolling the same query to the right, we can check the throughput values, both
including retransmissions and without retransmission. We can see that there is somedifference between these values, and if we check retransmission percentages, we see
that the first HTTP session has 0.30 % of packets retransmitted (Figure 26). The second
session is free of retransmissions.
Figure 26 - HTTP Example, Retransmissions
After that, we can use Task level query to find out which tasks have had problems. Then
we can open a chart with the attribute Packet_Evt_Packet_Retransmitted. There are 10
7/28/2019 02-White Paper - Analysis of GPRS Services Rev 1.1
25/34
White paper Analysis of GPRS services Page 25 of 34
retransmissions, but most of them occurred on the signalling phase, so they are not
taken into account when calculating the number of Service Packets and Bytes. Lets
zoom to the task that we found had retransmissions (Figure 27). Using tables, charts
and the Protocol Stack Browser, we can see when retransmissions happened, in which
session and task, and also investigate particular IP packets if necessary.
Figure 27 - HTTP Example, Chart
In this case there were only three retransmissions, which do not cause any significant
quality degradation to the end user. Anyway, in this way it is possible to focus on the
problematic tasks and analyse these only.
Another useful attribute is to use round trip time measurements:
Packet_RTT_Server_Side and Packet_RTT_Mobile_Side (Figure 28).
7/28/2019 02-White Paper - Analysis of GPRS Services Rev 1.1
26/34
White paper Analysis of GPRS services Page 26 of 34
Figure 28 - HTTP Example, RTTs
From this chart, we can see what the response times are. Since the logfile is captured
on Mobile Site, response times are short. From Server Side RTT we can see quite
constant response time, with some peaks, which can be due network delay or the server
has been busy. Every user can define some ideal threshold for the parameters and use
them as benchmarking to highlight a critical performance.
Although this specific trace shows no big problems, the process shown can be applied
to spot abnormal patterns and drill down to the root cause.
7/28/2019 02-White Paper - Analysis of GPRS Services Rev 1.1
27/34
White paper Analysis of GPRS services Page 27 of 34
Appendix A
Attribute Description Unit Direction
Subscriber Session
Subscriber_Session_ID A unique ID for this Subscriber Session Identifier
Event
Subscriber_Evt_Start_Session Generated on the first packet for a Subscriber Session Event
Subscriber_Evt_End_Session Generated when the Mobile or Server indicate the SubscriberSession is over
Event
Event Times
Subscriber_Time_Start_Session The time of Subscriber_Evt_Start_Session Relativemillisec
Subscriber_Time_End_Session The time of Subscriber_Evt_End_Session Relativemillisec
Context Session
Context_Session_ID A unique ID for this Context Session Identifier
Context_Duration_Session The duration of the Context Session Time
Event
Context_Evt_Start_Session Generated on the first packet for a Context Session Event
Context_Evt_End_Session Generated when the Mobile or Server indicate the ContextSession is over
Event
Event Times
Context_Time_Start_Session The time of Context_Evt_Start_Session Relativemillisec
Context_Time_End_Session The time of Context_End_Session Relativemillisec
Service Session
Service_Session_ID A unique ID for this Service Session Identifier
Service_Protocol_Type The service type, eg HTTP, FTP, POP3, SMTP, WAP or ICMP Protocol
Key Performance Indicators
Service_Duration_Session_Initialization
The time it takes for the server to be ready to serve a mobile'srequest for data transfer
Millisec
Service_Perc_Packets_Retr_UL The percentage of packets that have been retransmitted Percent Uplink
Service_Perc_Packets_Retr_DL The percentage of packets that have been retransmitted Percent Downlink
Service_ThrPut_Fin_IncRetr_UL Final throughput including retransmissions. Calculated only atthe end of the service
Kilobits persecond
Uplink
Service_ThrPut_Fin_NoRetr_UL Final throughput not including retransmissions. Calculated onlyat the end of the service
Kilobits persecond
Uplink
Service_ThrPut_Fin_IncRetr_DL Final throughput including retransmissions. Calculated only atthe end of the service
Kilobits persecond
Downlink
Service_ThrPut_Fin_NoRetr_DL Final throughput not including retransmissions. Calculated onlyat the end of the service
Kilobits persecond
Downlink
Measure
Service_Duration_Data_Transfer_De
lta
The time the task was active Millisec
Service_Duration_Data_Transfer_Cum
The sum of Service_Duration_Data_Transfer_Delta Millisec
Service_Duration_Session_Delta The time since the last task was complete or the Service wasinitialized. This can be used as interarrival time between tasks.
Millisec
Service_Duration_Session_Cum The time since Service was Initialized. Calculated at the end ofTask and at the end on Service.
Millisec
Service_Packets_Delta_IncRetr_UL The number of packets transmitted in the completed task. This
includes retransmitted packets
Count Uplink
Service_Packets_Cum_IncRetr_UL The sum of packets transmitted in the completed tasks. Thisincludes retransmitted packets
Count Uplink
7/28/2019 02-White Paper - Analysis of GPRS Services Rev 1.1
28/34
White paper Analysis of GPRS services Page 28 of 34
Attribute Description Unit Direction
Service_Packets_Delta_NoRetr_UL The number of packets transmitted in the completed task. Thisdoes not include retransmitted packets
Count Uplink
Service_Packets_Cum_NoRetr_UL The sum of packets transmitted in the completed tasks. Thisdoes not include retransmitted packets
Count Uplink
Service_Bytes_Delta_IncRetr_UL The number of bytes transmitted in the completed task. Thisincludes retransmitted packets
Bytes Uplink
Service_Bytes_Cum_IncRetr_UL The sum of bytes transmitted in the completed tasks. This
includes retransmitted packets
Bytes Uplink
Service_Bytes_Delta_NoRetr_UL The number of bytes transmitted in the completed task. Thisdoes not include retransmitted packets
Bytes Uplink
Service_Bytes_Cum_NoRetr_UL The sum of bytes transmitted in the completed tasks. This doesnot include retransmitted packets
Bytes Uplink
Service_Packets_Delta_IncRetr_DL The number of packets transmitted in the completed task. This
includes retransmitted packets
Count Downlink
Service_Packets_Cum_IncRetr_DL The sum of packets transmitted in the completed tasks. Thisincludes retransmitted packets
Count Downlink
Service_Packets_Delta_NoRetr_DL The number of packets transmitted in the completed task. This
does not include retransmitted packets
Count Downlink
Service_Packets_Cum_NoRetr_DL The sum of packets transmitted in the completed tasks. Thisdoes not include retransmitted packets
Count Downlink
Service_Bytes_Delta_IncRetr_UL The number of bytes transmitted in the completed task. Thisincludes retransmitted packets
Bytes Downlink
Service_Bytes_Cum_IncRetr_DL The sum of bytes transmitted. This includes retransmittedpackets
Bytes Downlink
Service_Bytes_Delta_NoRetr_DL The number of bytes transmitted in the completed task. Thisdoes not include retransmitted packets
Bytes Downlink
Service_Bytes_Cum_NoRetr_DL The sum of bytes transmitted. This does not includeretransmitted packets
Bytes Downlink
Event
Service_Evt_Start_Session Generated on the first packet for a Service Session Event
Service_Evt_Incomplete_Start_Session
Generated when a start of the Service Session was not found Event
Service_Evt_Initialize_Session Generated when the Server has acknowledged that it is readyto service any request from the Mobile
Event
Service_Evt_End_Session Generated when the Mobile or Server indicate the ServiceSession is over
Event
Service_Evt_Abort_Session Generated when the Mobile or Server abnormally terminate the
Service Session, or a timeout occurs in the Application Layer
Event
Service_Cause_Abort The reason for abnormal termination of the Service Session Packet DataSessionEnded/TaskSessionTimeout
Event Times
Service_Time_Start_Session The time of Service_Evt_Start_Session Relative
millisec
Service_Time_Initialize_Session The time of Service_Evt_Initialize_Session Relativemillisec
Service_Time_End_Session The time of Service_Evt_End_Session Relativemillisec
Task Session
Task_Session_ID A unique ID for this Task Session Identifier
Key Performance Indicators
Task_Perc_Packets_Retr_UL The percentage of packets retransmitted. Calculated only at
the end of the task
Percent Uplink
Task_Perc_Packets_Retr_DL The percentage of packets retransmitted. Calculated only atthe end of the task
Percent Downlink
Task_ThrPut_Fin_IncRetr_UL Final throughput including retransmissions. Calculated only atthe end of the task
Kilobits persecond
Uplink
Task_ThrPut_Fin_NoRetr_UL Final throughput not including retransmissions. Calculated only
at the end of the task
Kilobits per
second
Uplink
7/28/2019 02-White Paper - Analysis of GPRS Services Rev 1.1
29/34
White paper Analysis of GPRS services Page 29 of 34
Attribute Description Unit Direction
Task_ThrPut_Fin_IncRetr_DL Final throughput including retransmissions. Calculated only atthe end of the task
Kilobits persecond
Downlink
Task_ThrPut_Fin_NoRetr_DL Final throughput not including retransmissions. Calculated onlyat the end of the task
Kilobits persecond
Downlink
Measure
Task_Duration_Session The duration of the Task Session Time
Task_Time_Delta_UL Time since last Application Data. This can be used as
interarrival time between packets
Time Uplink
Task_Time_Delta_DL Time since last Application Data. This can be used asinterarrival time between packets
Time Downlink
Task_Time_Cum_UL The sum of Task_Time_Delta Time Uplink
Task_Time_Cum_DL The sum of Task_Time_Delta Time Downlink
Task_Packets_Cum_IncRetr_UL The sum of packets including retransmissions Count Uplink
Task_Packets_Cum_NoRetr_UL The sum of packets excluding retransmissions Count Uplink
Task_Bytes_Delta_IncRetr_UL The number of bytes sent in this packet includingretransmissions
Bytes Uplink
Task_Bytes_Delta_NoRetr_UL The number of bytes sent in this packet excludingretransmissions
Bytes Uplink
Task_Bytes_Cum_IncRetr_UL The sum of bytes sent including retransmissions Bytes Uplink
Task_Bytes_Cum_NoRetr_UL The sum of bytes sent excluding retransmissions Bytes Uplink
Task_Packets_Cum_IncRetr_DL The sum of packets including retransmissions Packets Downlink
Task_Packets_Cum_NoRetr_DL The sum of packets excluding retransmissions Packets Downlink
Task_Bytes_Delta_IncRetr_DL The number of bytes sent in this packet includingretransmissions
Bytes Downlink
Task_Bytes_Delta_NoRetr_DL The number of bytes sent in this packet excludingretransmissions
Bytes Downlink
Task_Bytes_Cum_IncRetr_DL The sum of bytes sent including retransmissions Bytes Downlink
Task_Bytes_Cum_NoRetr_DL The sum of bytes sent excluding retransmissions Bytes Downlink
Task_ThrPut_Inst_IncRetr_UL Instantaneous throughput not including retransmissions.Calculated once a second
Kilobits persecond
Uplink
Task_ThrPut_Inst_NoRetr_UL Instantaneous throughput not including retransmissions.Calculated once a second
Kilobits persecond
Uplink
Task_ThrPut_Inst_IncRetr_DL Instantaneous throughput not including retransmissions.Calculated once a second
Kilobits persecond
Downlink
Task_ThrPut_Inst_NoRetr_DL Instantaneous throughput not including retransmissions.Calculated once a second
Kilobits persecond
Downlink
Task_Perc_Packets_Cum_Retr_UL The cumulative percentage of retransmitted packets. It isgenerated each time a packet is received
Percent Uplink
Task_Perc_Packets_Cum_Retr_DL The cumulative percentage of retransmitted packets. It isgenerated each time a packet is received
Percent Downlink
Event
Task_Evt_Start_Session Generated at the start of a Task Session Event
Task_Evt_End_Session Generated at the end of a Task Session Event
Task_Evt_Abort_Session Generated when on Packet Layer abort or Application Layertimeout
Event
Task_Cause_Abort The cause of the Task Abort Event
Event Times
Task_Time_Start_Session Time of Task_Evt_Start_Session Time
Task_Time_End_Session Time of Task_Evt_End_Session Time
Packet Session
Packet_Session_ID A unique ID for this Packet Session Identifier
Packet_Direction The direction of this packet Uplink/Downlink
Key Performance Indicators
Packet_Packets_Retr_Perc_UL Percentage of packet retransmissions Percent Uplink
Packet_Packets_Retr_Perc_DL Percentage of packet retransmissions Percent Downlink
7/28/2019 02-White Paper - Analysis of GPRS Services Rev 1.1
30/34
White paper Analysis of GPRS services Page 30 of 34
Attribute Description Unit Direction
Packet_Duration_Initialization The time it takes for the Server to be ready to serve a Mobile'srequest for data transfer
Time
Packet_Duration_Close The time it takes for the Mobile or Server to close the PacketSession
Time
Measure
Packet_Packets_Cum_IncRetr_UL Packet count, including retransmissions Count Uplink
Packet_Packets_Cum_NoRetr_UL Packet count, not including retransmissions Count Uplink
Packet_Packets_Cum_Retr_Perc_UL Percentage of packet retransmissions Percent Uplink
Packet_ThrPut_Inst_IncRetr_UL Instantaneous Packet Throughput, including retransmissions.Calculated once a second
Kilobits persecond
Uplink
Packet_ThrPut_Inst_NoRetr_UL Instantaneous Packet Throughput, not includingretransmissions. Calculated once a second
Kilobits persecond
Uplink
Packet_Packets_Cum_IncRetr_DL Packet count, including retransmissions Count Downlink
Packet_Packets_Cum_NoRetr_DL Packet count, not including retransmissions Count Downlink
Packet_Packets_Cum_Retr_Perc_DL Percentage of packet retransmissions Percent Downlink
Packet_ThrPut_Inst_IncRetr_DL Instantaneous Packet Throughput, including retransmissions.Calculated once a second
Kilobits persecond
Downlink
Packet_ThrPut_Inst_NoRetr_DL Instantaneous Packet Throughput, not includingretransmissions. Calculated once a second
Kilobits persecond
Downlink
Packet_Duration_Data_Session The duration of the Packet Data Session, maximum of
Packet_Duration_Mobile_Data_Session andPacket_Duration_Server_Data_Session
Time
Packet_Duration_Mobile_Data_Session
The duration of the Mobile Packet Data Session Time
Packet_Duration_Server_Data_Session
The duration of the Server Packet Data Session Time
Packet_RTT_Server_Side Round-Trip Time it takes to send a packet to the Server andcome back to the Mobile
Millisec
Packet_RTT_Mobile_Side Round-Trip Time it takes to send a packet to the Mobile andcome back to the Server
Millisec
Packet_Bytes_Delta_IncRetr_UL The number of bytes sent in this packet includingretransmissions
Bytes Uplink
Packet_Bytes_Delta_NoRetr_UL The number of bytes sent in this packet excludingretransmissions
Bytes Uplink
Packet_Bytes_Cum_IncRetr_UL The sum of bytes sent including retransmissions Bytes Uplink
Packet_Bytes_Cum_NoRetr_UL The sum of bytes sent excluding retransmissions Bytes Uplink
Packet_Bytes_Delta_IncRetr_DL The number of bytes sent in this packet includingretransmissions
Bytes Downlink
Packet_Bytes_Delta_NoRetr_DL The number of bytes sent in this packet excludingretransmissions
Bytes Downlink
Packet_Bytes_Cum_IncRetr_DL The sum of bytes sent including retransmissions Bytes Downlink
Packet_Bytes_Cum_NoRetr_DL The sum of bytes sent excluding retransmissions Bytes Downlink
Event
Packet_Evt_Start_Session Generated on the first packet of the Packet Session Event
Packet_Evt_Incomplete_Start_Session
Generated when a start of the Packet Session was not found Event
Packet_Evt_Start_Data_Session Generated when the initialization phase of the Packet Sessionis over
Event
Packet_Evt_Mobile_End_Data_Sessi
on
Generated when the Mobile indicates the Packet Data Session
is over
Event Uplink
Packet_Evt_Server_End_Data_Session
Generated when the Server indicates the Packet Data Sessionis over
Event Downlink
Packet_Evt_End_Data_Session Generated when both the Mobile and Server has indicated thePacket Data Session is over
Event
Packet_Evt_End_Session Generated when both the Mobile or Server indicate the PacketSession is over
Event
Packet_Evt_Abort_Session Generated when the Mobile or Server abnormally terminate thePacket Session
Event
Packet_Evt_Packet_Retransmitted This packet is a retransmission of some previous packet Event
7/28/2019 02-White Paper - Analysis of GPRS Services Rev 1.1
31/34
White paper Analysis of GPRS services Page 31 of 34
Attribute Description Unit Direction
Packet_Evt_Packet_Truncated This packet is truncated Event
Packet_Evt_Packet_Corrupted This packet is corrupted Event
Event Initiator
Packet_Initiator_Open The entity initiating the opening of the Packet Session Server/Mobile
Packet_Initiator_Close The entity closing the Packet Session Server/Mobile
Packet_Initiator_Abort The entity aborting the Packet Session Server/Mobile
Event Times
Packet_Time_Start_Session Time of Packet_Evt_Start_Session Time
Packet_Time_Start_Data_Session Time of Packet_Evt_Start_Data_Session Time
Packet_Time_End_Data_Session Time of Packet_Evt_End_Data_Session Time
Packet_Time_Mobile_End_Data_Session
Time of Packet_Evt_Mobile_End_Data_Session Time Uplink
Packet_Time_Server_End_Data_Session
Time of Packet_Evt_Server_End_Data_Session Time Downlink
Packet_Time_End_Session Time of Packet_Evt_End_Session Time
7/28/2019 02-White Paper - Analysis of GPRS Services Rev 1.1
32/34
White paper Analysis of GPRS services Page 32 of 34
Appendix B
KPIs formulas
Service_Duration_Session_Initialization = essionme_Start_SService_Ti-nize_Sessiome_InitialService_Ti
Service_Perc_Packets_Retr =IncRetrckets_Cum_Service_Pa
NoRetrckets_Cum_Service_PaIncRetrckets_Cum_Service_Pa100
Service_ThrPut_Fin =_Cuma_Transferration_DatService_Du
tes_CumService_By8*
Task_Perc_Packets_Retr =Retrts_Cum_IncTask_Packe
etrts_Cum_NoRTask_PackeRetrts_Cum_IncTask_Packe100
Task_ThrPut_Fin = nion_SessioTask_Durat
_CumTask_Bytes8*
Packet_Packets_Retr_Perc =ncRetrkets_Cum_IPacket_Pac
oRetrkets_Cum_NPacket_PacncRetrkets_Cum_IPacket_Pac100
Packet_Duration_Initialization = ssione_Start_SePacket_Timta_Sessione_Start_DaPacket_Tim
Packet_Duration_Close = _Sessione_End_DataPacket_Timione_End_SessPacket_Tim
7/28/2019 02-White Paper - Analysis of GPRS Services Rev 1.1
33/34
White paper Analysis of GPRS services Page 33 of 34
Measures formulas
Service_Duration_Data_Transfer_Delta = Task_Duration_Session
Service_Duration_Data_Transfer_Cum = =
n
i
i
1
_Deltaa_Transferration_DatService_Du
Service_Duration_Session_Delta =
=
=
niii
i-i
K21.2
1
when,nEnd_SessioTask_Evt_-nEnd_SessioTask_Evt_
when,ze_Sessiont_InitialiService_EvnEnd_SessioTask_Evt_1.
Service_Duration_Session_Cum =
iont_End_SessService_Evwhen,ze_Sessiont_InitialiService_Ev-iont_End_SessService_Ev
iont_End_SessService_Evnowhen,ze_Sessiont_InitialiService_Ev-nEnd_SessioTask_Evt_1.
.2
Service_Packets_Delta = Task_Packets_Cum
Service_Packets_Cum = =
n
i
i
1
ackets_DeltService_Pa
Service_Bytes_Delta = Task_Bytes_Cum
Service_Bytes_Cum = =
n
i
i
1
tes_DeltaService_By
Task_Duration_Session = Task_Evt_End_Session Task_Evt_Start_Session
Task_Time_Delta =
=
=
niii
i-i
K21.2
1
when,TimeArrivalPacket-TimeArrivalPacket
whenion,Start_SessTask_Evt_TimeArrivalPacket1.
Task_Time_Cum = =
n
i
i
1
DeltaTask_Time_
Task_Bytes_Delta_IncRetr =
=
=
=
WAPeotocol_TypService_PrwhendLength,WSP_Payloa
ICMPeotocol_TypService_PrwhenadLength,ICMP_Paylo
FTPorHTTPeotocol_TypService_PrwhenadLength,TCP_Paylo
.
.
3
.2
1
Task_Bytes_Delta_NoRetr =
( )dtransmitte_Packet_RePacket_Evtnotif
=
=
=
WAPeotocol_TypService_PrwhendLength,WSP_Payloa
ICMPeotocol_TypService_PrwhenadLength,ICMP_Paylo
FTPorHTTPeotocol_TypService_PrwhenadLength,TCP_Paylo
.
.
3
.2
1
Task_Bytes_Cum = =
n
i
i
1
_DeltaTask_Bytes
Task_Packets_Cum_IncRetr = ( ) ++> cRetrets_Cum_InTask_PackRetr_Delta_IncTask_Bytesif 0
Task_Packets_Cum_NoRetr =( ) ( )( )
++
>
_NoRetrackets_CumTask_P
Retr_Delta_IncTask_Bytesanddtransmitte_Packet_RePacket_Evtnotif 0
Task_ThrPut_Inst =
sec1
_CumTask_Bytes=
n
i
i
1 , n = the number of the packets in current second
7/28/2019 02-White Paper - Analysis of GPRS Services Rev 1.1
34/34
Task_Perc_Packets_Cum_Retr =Retrts_Cum_IncTask_Packe
etrts_Cum_NoRTask_PackeRetrts_Cum_IncTask_Packe100
Packet_Packets_Cum_IncRetr = ( ) ++> ncRetrkets_Cum_IPacket_PacncRetres_Delta_IPacket_Bytif 0
Packet_Packets_Cum_NoRetr =( ) ( )( )
++
>
oRetrkets_Cum_NPacket_Pac
ncRetres_Delta_IPacket_Bytanddtransmitte_Packet_RePacket_Evtnotif 0
Packet_Packets_Cum_Retr_Perc =ncRetrkets_Cum_IPacket_Pac
oRetrkets_Cum_NPacket_PacncRetrkets_Cum_IPacket_Pac100
Packet_ThrPut_Inst =
sec1
es_CumPacket_Byt=
n
i
i
1 , n = the number of the packets in current second
Packet_Duration_Mobile_Data_Session = Packet_Evt_Mobile_End_Data_Session Packet_Evt_Start_Data_Session
Packet_Duration_Server_Data_Session = Packet_Evt_Server_End_Data_Session Packet_Evt_Start_Data_Session
Packet_Duration_Data_Session = ( er_Data_Sation_ServPacket_Durssion,le_Data_Seation_MobiPacket_DurMax
Packet_RTT_Server_Side =
=
=
=
WAPeotocol_TypService_PrwhenAck,olData/Contr-MobilebysentPacketolData/Contr
ICMPeotocol_TypService_Prwhenreceived,ReplyEcho-MobilebysentRequestEcho
FTPorHTTPeotocol_TypService_Prwhenreceived,ackSYNTCP-MobilebysentSYNTCP
.
.
3
.2
1
Packet_RTT_Mobile_Side =
=
=
=
WAPeotocol_TypService_PrwhenAck,olData/Contr-MobilebysentPacketolData/Contr
ICMPeotocol_TypService_Prwhenreceived,ReplyEcho-MobilebysentRequestEcho
FTPorHTTPeotocol_TypService_Prwhenreceived,ackSYNTCP-MobilebysentSYNTCP
.
.
3
.2
1
Packet_Bytes_Delta_IncRetr =
=
=
=
WAPeotocol_TypService_PrwhendLength,WSP_Payloa
ICMPeotocol_TypService_PrwhenadLength,ICMP_Paylo
FTPorHTTPeotocol_TypService_PrwhenadLength,TCP_Paylo
.
.
3
.2
1
Packet_Bytes_Delta_NoRetr =
( )dtransmitte_Packet_RePacket_Evtnotif
=
=
=
WAPeotocol_TypService_PrwhendLength,WSP_Payloa
ICMPeotocol_TypService_PrwhenadLength,ICMP_Paylo
FTPorHTTPeotocol_TypService_PrwhenadLength,TCP_Paylo
.
.
3
.2
1
Packet_Bytes_Cum = =
n
i
i
1
es_DeltaPacket_Byt