event analysis tutorial

Upload: crafter202

Post on 06-Apr-2018

229 views

Category:

Documents


0 download

TRANSCRIPT

  • 8/3/2019 Event Analysis Tutorial

    1/34

    Event Analysis Tutorial

  • 8/3/2019 Event Analysis Tutorial

    2/34

    INTRODUCTION

    Event analysis is done in order to provide an explanation for relevant eventsthat happen during drives. This includes Access Failures, CS and PS Drops,Application Drops, IFHO & IRAT Events and other relevant scenarios.

    We start our analysis for a particular UE by observing key attributes whichinclude RSCP, EcNo, UARFCN and Technology modes(UMTS or GSM).

    Typically, low RSCP is due to coverage issues, event takes place far from thebase station(base stations have limited range) or due to terrain

    issues(mountains , lakes, foliage etc). Low EcNo in areas with good coverage(RSCP) could be due to missing

    neighbors. Neighbor List(NL) for the frequency band under test must becarefully reviewed and NL additions must be submitted.

    Bad UCU cards might lead to degraded throughput on the aircards.

    Layer3 messaging interface in Actix(Protocol Stack) enables us to observe themessages exchanged between the server and the UE and helps us to

    determine the causes for events that happen during the drive. Note that the L3 messages can be expanded by clicking on the show all rows

    button. This is used mainly to find out reasons for application drops.

  • 8/3/2019 Event Analysis Tutorial

    3/34

    THE INTERFACE

    Layer3

  • 8/3/2019 Event Analysis Tutorial

    4/34

    Basic steps to follow for Analyzing Events.

    1) Start with plotting the UE measurement( EcIo/RSCP depending on the event.

    2) Plot the relevant event(access failure/drop and so on).3) Open the UMTS Radio Interface and Protocol Signaling from the Protocol Stack

    4) Also open the custom form which can be used to observe the SC/RSCP/EcIo

    values for the active, monitored and detected set. It also displays the event ids,

    the layer3 message and the SCs involved in the same form.

    5) Start with the L3 message where the event occurs, and observe all messages

    before the drop to find out the reason for the specific event.6) Refer to the list of event abbreviations and what each event means, before

    working on the analysis.

    7) Note that a reason for an event might not be observed in the L3 at the point

    where it occurs. We might have to go back and see what happens before the

    event.

  • 8/3/2019 Event Analysis Tutorial

    5/34

    DL interface between Base Station ->UE

    UL interface between UE -> Base Station

    All L3 messages with a prefix of DL indicates that it is sent

    from the UE to the server. Similarly all L3 messages with aprefix of UL are messages sent from the server to the UE.

    UE sends the measurement report which contain event data

    and RSCP, EcNo for SC in active, detected and monitored set.

    Base station sends measurement control information, this

    enables in updating active set.

    Basic overview of L3 messages

  • 8/3/2019 Event Analysis Tutorial

    6/34

    Call initiation messaging

    UE in idle mode, broad cast channel

    UE initiating call

    Acknowledgement from Base Station

    UE completes connection setup

  • 8/3/2019 Event Analysis Tutorial

    7/34

    Drops and AF Analysis

  • 8/3/2019 Event Analysis Tutorial

    8/34

    CS and PS Drops

  • 8/3/2019 Event Analysis Tutorial

    9/34

    CallDrops observed during drives

    1) Call drops can be of two types CS and PS.

    2) Normally we observe CS and PS drops during drives.3) These may or may not be RF related.

    4) Most of the times, the reason for a drop can be found out from observing the L3

    messages.

    5) Reasons for the drops are pretty much the same no matter which market the

    data is from.

    6) However note that there are times when a reason for a drop cannot be foundout from the L3.

  • 8/3/2019 Event Analysis Tutorial

    10/34

    The most common reason for a call drop is poor coverage. RSCP values are

    an indication of the coverage in a region. As shown below, poor coverage

    causes the drop

    10

  • 8/3/2019 Event Analysis Tutorial

    11/34

    Call Drop due to Pilot Pollution When there are too many servers in

    the region, the UE is unable to latch on to a single server. Note that all

    SCs in the active set have comparable EcNo.

    11

  • 8/3/2019 Event Analysis Tutorial

    12/34

    PS Call Drops are sometimes a result of huge traffic volumes

    caused by Buffer payloads.

  • 8/3/2019 Event Analysis Tutorial

    13/34

    13

    Radio Link Failures Note that Radio Link failures are pegged with the

    drops and they are not causes for the drops. The reason for the radio link

    failure(i.e drop) can be found out by looking deeper in the L3

    .

  • 8/3/2019 Event Analysis Tutorial

    14/34

    MISSING NEIGHBOURS ANALYSIS

  • 8/3/2019 Event Analysis Tutorial

    15/34

    Indication ofMissing Neighbors

    1) Possibility of a Missing neighbor is indicated when

    we observe poor EcIo in regions with good RSCP.

    2) We look in the measurement reports for e1a

    events( adding SC to the active set) and see if the

    UE is able to add the SCs present in the

    measurement report.

    3) We take the SCs in the measurement report, find

    the sectors mapped to it, and find out if the sectors

    serving the UE is added as neighbors to it.

    4) However, it does not necessarily mean that e1a

    events which are not completed are as a result of

    missing neighbors.

  • 8/3/2019 Event Analysis Tutorial

    16/34

    Missing neighbor from overshooting sector MIU40063(SC=310) to MIU32172(SC=278)

    F3 Site

  • 8/3/2019 Event Analysis Tutorial

    17/34

    Access Failures

  • 8/3/2019 Event Analysis Tutorial

    18/34

    Access Failures Observed

    1) One of the most common causes of Access failures

    are location updates performed by the UE at the

    same time that Agilent tries to initiate a call from it.

    2) Many times we observe access failures in

    pairs(outgoing and incoming)

    3) Incoming setup fails are caused by GPRS PDP

    context rejects

    4) Access Failures are also observed due to poor RF

    conditions.

    5) Absence of radio bearer reconfigurations after e1d

    events lead to access failures.

    6) A sharp drop in the EcIo values is observed just

    before the e1d event.

  • 8/3/2019 Event Analysis Tutorial

    19/34

    Outgoing Setup Fails happen when the Call is trying to be setup during a

    location update. i.e Agilent tries to initiate a call from a UE while it is

    communicating with the server to perform a location update. Agilent

    interprets it as busy and pegs an access failure to it.

    19

  • 8/3/2019 Event Analysis Tutorial

    20/34

    HSDPA Outgoing Setup Fail No radio bearer reconfiguration after e1d event. The Reference cell

    does not change.( The UE fails to recognize the SC that it has latched on to)

  • 8/3/2019 Event Analysis Tutorial

    21/34

    Alternatively, before the e1d event, we see a sharp drop in the EcIo value when the call is setup.

    This results in an access failure as shown

  • 8/3/2019 Event Analysis Tutorial

    22/34

    Outgoing Setup Fail Call Setup during GPRS Service Request

  • 8/3/2019 Event Analysis Tutorial

    23/34

    Incoming Setup Fail GPRS PDP Context rejected

  • 8/3/2019 Event Analysis Tutorial

    24/34

    Application Drops and 120s get/put

  • 8/3/2019 Event Analysis Tutorial

    25/34

    Application Drops and Timeouts

    1) 120s get/put happens when the UE takes more than 120s

    for a single ftp download/upload session2) Sometimes, application drops are pegged to the 120s

    timeout.

    3) Application drops are caused by other reasons too which

    include winsock errors, socket exceptions and the tests

    being aborted by the user.

    4) Note that using the IPv4/6 test in Agilent prevents the user

    from stopping the log file while the ftp session is in

    progress.

    5) An easy way to find out a reason for an application drop isto plot Data Testing->Task Summary->Task failure cause.

    When we mouse over the plot, Actix displays the reason.

    We can also click on the dot displayed which will show the

    corresponding message in L3

  • 8/3/2019 Event Analysis Tutorial

    26/34

    26

    Task Failure Cause

    Mouse over the dot to display the cause for the application

    drop. Clicking on it will take you to the L3 message accordingly.

    Plot Task Failure Cause

  • 8/3/2019 Event Analysis Tutorial

    27/34

    Application drop pegged to 120s get. Note that when app drops are caused by 120s

    get/put, we normally plot the RSCP of the UE and overlay it with the 120s get, the app

    drop and the ftp sessions. We show how the session starts in poor coverage thereby

    resulting in the download time to exceed 120s.

    Reselection

    to F3

    Plot the event task start

    for the ftp session

  • 8/3/2019 Event Analysis Tutorial

    28/34

    Application Drop Net Socket Exception

    Reselection

    to F3

  • 8/3/2019 Event Analysis Tutorial

    29/34

    ApplicationDrop Test aborted by user

    Reselection

    to F3

  • 8/3/2019 Event Analysis Tutorial

    30/34

    30

    ApplicationDrop Winsock Error

    Application drops due to Winsock error occur due to problem in network. The

    new update has fixed this in Phoenix market.

  • 8/3/2019 Event Analysis Tutorial

    31/34

    IFHO/IRAT in L3

    d d b h h ld f

  • 8/3/2019 Event Analysis Tutorial

    32/34

    Long MO IFHO Event e2d event triggered by RSCP thresholds for

    MIU32083(SC=238) to change to MIU32089(SC=238) . We search for the

    e2d(event trigger) which triggers the IFHO event. There will also be

    measurement control messages which will specify the EcIo or RSCP levels

    which triggered the UE to perform a handoff. A radio bearer reconfiguration

    message indicates an inter frequency handoff

  • 8/3/2019 Event Analysis Tutorial

    33/34

    Long MO IRAT event 1 e2d event when RSCP thresholds are reached for

    MIU32173(SC=286) to IRAT to GSM. We search for the e2d(event trigger) which

    triggers the IRAT event. There will also be measurement control messages which will

    specify the EcIo or RSCP levels which triggered the UE to perform a handoff. A

    Handover to GSM message indicates the technology change

  • 8/3/2019 Event Analysis Tutorial

    34/34

    34

    Low throughputs are observed on the aircards when the UE

    latches on to servers from either side of the IUR Boundary