Beruflich Dokumente
Kultur Dokumente
workshop
Yves Hawa
Asiacell Iraq
Table Of Contents
1- LAC/RAC Network Status and SPLIT
2- URA and FDH Background
3- URA and FDH Description
4- URA Signaling
5- URA Planning and Dimensioning
6- Feature licenses and Parameters
7- Activation and Trial
a- Radio Network KPIs
b- RNC Load KPIs
8- Conclusion
9- Counters and KPIs
Network LAC/RAC Status and Split
› CS paging/s = pmCnInitPagingToIdleUeLa/3600
› PS paging/s = pmCnInitPagingToIdleUeRa/3600
› A LAC or RAC split is considered when the paging/s is reaching more than 120 pages/sec
› Karbala during Arbaneeya event reached a maximum of approximately 300 PS pages/s while
the CS pages/s reached a maximum of 160 pages/s. A new LAC/RAC split or URA
implementation is required in preparation for next year event.
› However in Tikrit, the current number of pages/s is still way below the threshold and doesn’t
require a LAC/RAC split at the moment.
1- The UE benefits from faster access whilst still maintaining low power
consumption.
› Why URA now ? 1st the PS traffic has significantly increased over the years. However, the
biggest driver for URA_PCH came with the introduction of Smartphones.
› Smartphones introduced bigger screen sizes, more powerful processors and increased
user activity. This resulted in greater power consumption and therefore shorter battery life,
which was a problem for the end users and therefore for the UE vendors too.
› This situation was addressed by the standardization of Fast Dormancy in 3GPP Release 8
(here called Release 8 Fast Dormancy). It allows the UE to request a transition to a lower
energy state, and the network can then respond by sending the UE to CELL_FACH or
URA_PCH rather than to Idle. This significantly reduces the load on the network, as there
is much less signaling involved with transitions to and from CELL_FACH and URA_PCH
than with transitions to and from Idle. Release 8 Fast Dormancy also introduced a timer
called t323 which limits how often a UE can request Fast Dormancy.
URA DESCRIPTION (1/8)
› URA-PCH is as a “waiting state for PS services”, URA_PCH can be viewed
as the ”new Idle for PS connections”. Similar to Idle, the UE does not
transmit any data and its position is known with less granularity than when
in CELL_FACH or CELL_DCH state. The radio resources in the RNC are
released even though the signaling connections over IU are kept. Thus, the
SGSN does not know that the UE has moved to URA_PCH.
1- When the UE wants to release, it must first ask the network. To do this, it
uses an information Element (IE) which was added to the SCRI.
The content of the IE is a cause value saying “UE requested PS
session Data End”.
2- The timer t323 was added, which sets the minimum allowable time
between sending SCRI requests. t323 is set in the network and
broadcast in SIB 1. It is also sent in UTRAN Mobility Information to
cover the case when the UE has not yet read the system
information, for example because it has come from GSM.
› In the figure above, The UE decides when to switch to Idle. The state transitions for the three bursts in Figure
are identical. This is because the UE Proprietary Fast Dormancy is initiating a transition to Idle before
inactivityTimer expires. To be more precise, the UE timer is shorter than the sum of hsdDschInactivityTimer
and inactivityTimer, thus the UE is triggering Fast Dormancy. Note that the UE Proprietary Fast Dormancy
could be triggered from CELL_DCH instead. This would happen if the UE timer, which is brand specific, were
shorter than the applicable inactivity timer on CELL_DCH (e.g.hsDschInactivityTimer).
URA DESCRIPTION (8/8)
› In this figure, after the first data burst, the UE requests a downswitch and it is granted. After the second burst
the UE is inhibited from sending the request because t323 has not yet expired. Instead it has to wait till t323
expires. By the time the third burst is sent, t323 has expired again, so the UE can once again send SCRI
when it wants to. Note that if inactivityTimer on CELL_FACH expires before the UE timer it is the network
that triggers the downswitch to URA_PCH. This case is not shown in the figure. The share of network versus
UE triggered downswitches will thus depend on the relation between inactivity- and t323 timers in the RNC
and the UE timer.
URA Signaling: Paging ues in ura_pch (1/2)
› Paging messages are either of type 2 or type 1. Type 2 paging messages are targeting
users in CELL_DCH or CELL_FACH and sent to one cell only. They constitute a negligible
part of all paging messages and are not impacted by URA_PCH. Paging messages of type
1 are addressing users in Idle or URA_PCH and may be broadcasted over many cells.
When URA_PCH is introduced, there are two impacts on type 1 paging:
- A share of pages will originate from the RNC instead of the SGSN
- A share of pages will be broadcast over a UTRAN Registration Area (URA), instead of a
Location Area (LA) or a Routing Area (RA)
› The impacts have the following reasons: When the network wishes to contact a UE, a
page is required, regardless of whether the UE is in Idle or URA_PCH. However, the page
is handled differently in the two cases. The difference lies in which node originates the
page and the area size over which it is broadcast. In idle mode, CS pages are originated in
the MSC and broadcast to the LA, and PS pages are originated in the SGSN and
broadcast over the RA. With URA_PCH, both CS and PS pages are handled differently:
URA Signaling: Paging ues in ura_pch (2/2)
› A UE in Idle mode informs the MSC when it passes an LA border, and the SGSN
when it passes an RA border. It also periodically informs the MSC and SGSN
about the current LA and RA via periodic updates. The period is set by t3212 and
t3312 respectively.
› A UE in URA_PCH, on the other hand, informs the RNC when it passes a URA
border. It also sends periodic URA updates to the RNC (with periodicity t305) and
periodic LA updates to the MSC (with periodicity t3212). The reason it sends
periodic LA updates is because the MSC thinks the UE is in idle mode and is
expecting these updates.
› The URA updates do not go over Iu-CS or Iu-PS as do the LA and RA updates
respectively. This benefits the Iu-CS and Iu-PS as well as the RNC processors.
URA Signaling: multirab signaling (1/3)
› Call Setup:
› There are two ways to create a multirab of the kind “Speech + 1xPS”:
-A Speech call is ongoing when a PS Data session is started
-A PS Data session is ongoing when a speech call is started
› The first case cannot involve URA, because the setup is done on DCH. If the
speech call later ends then the PS part may drop to URA_PCH.
› In the second case the UE can be in CELL_DCH, CELL_FACH or URA_PCH
when the Speech call is to be setup. When in URA_PCH, the UE will transition to
CELL_DCH and the network will perform a RAB reconfiguration to “Speech + PS
interactive 0/0”
› From the MSC point of view there is no difference between this case and setting
up a single speech RAB. From the SGSN point of view, it does not know about the
signaling occurring and the UE state transitions, and keeps the Iu Signaling
Bearer and Iu Bearer over Iu-PS.
URA Signaling: multirab signaling (2/3)
› Call Release:
› We now examine what happens when one of the connections is released from a
“Speech plus 1xPS” multirab. There are two cases:
› Let’s take the case of the speech call being released first. If there is ongoing PS
activity at the time of the CS release then the PS part will continue until it becomes
inactive and is down-switched to CELL_FACH and then potentially to URA_PCH.
In the case where the PS session is inactive before the speech call is released,
the RAB will have been reconfigured to “Speech + PS interactive 0/0”. When the
speech call is then released the connection will also be reconfigured to
CELL_FACH and then potentially to URA_PCH
URA Signaling: multirab signaling (3/3)
› Call Release:
› Now let’s take the case of the PS data session being released first. It is
important to understand what is meant by a release here. The system will
not “release” the PS part of a multirab just because it has become inactive.
It will simply downswitch it to “Speech + PS interactive 0/0” – and this is the
case described above. A true “release” will only happen if the whole data
connection is terminated. In this case the multirab will reconfigure to
speech only, and so URA_PCH will not be involved.
URA Planning & DimensioninG (1/2)
1. The LAC, RAC and URA planning is a tradeoff between paging load on the one hand, and LA, RA
and URA updates on the other.
2. In a network facing too high paging load, it may be easier to plan new URAs than to re-plan the
LACs and RACs, new URAs doesn’t involve any Core actions.
4. Every release from URA_PCH to Idle is performed with a page. Such a release with a page is not
performed from any other state. Thus there is an increase in paging load if the URA is as big as the
LAs and RAs. It consequently has to be smaller. The impact from the extra page will be more
visible the shorter inactivityTimerPch is.
5. Each cell may belong to up to 4 URAs and the URAs can therefore be partly overlapping, which will
“smear out” the URA borders and mitigate large concentrations of URA updates on border cells, for
example when a URA border run through a busy traffic area.
URA Planning & DimensioninG (2/2)
This is the feature that enables URA_PCH. It is a prerequisite for other features Fast Dormancy
Handling and Faster Establishments, Direct upswitch from URA.
Fast Dormancy Handling ensures that Release 8 capable UEs do fast dormancy to URA_PCH rather
than to idle. This maximizes the number of users in URA_PCH and therefore maximizes the benefits of
URA_PCH.
parameters
Following table shows the recommended values for the parameters when
URA_PCH and Fast Dormancy Handling are activated.
RNCBIS953E
RRC CS success rate is maintained. 90% reduction in the PS RRC attempts after URA activation. Reduced from 99.8% to 99.4%
It has been commonly observed that when the absolute number of events decreases the success rate also decreases.
UEs in poor coverage will more often drop to Idle, Consequently higher share of UEs may end up performing RRC connection requests
from poor coverage, resulting in lower success rate.
RAB success rate
RNCBIS953E
RNCBIS953E
RNCBIS953E
RNCBIS953E
After increasing the users license in SGSN, HSDPA DCR came back to original trend.
RNCBIS953E
RNCBIS953E
RNCBIS953E
Reduction in HS/EUL users at the RNC level is observed due to users in URA_PCH state.
Channel switch fach_ura
RNCBIS953E
RNCBIS953E
RNCBIS953E
RNCBIS953E
RNCBIS953E
Number of paging type 1 attempts from SGSN to RNC reduced significantly by around 65%.
UTRAN initiated paging attempts forms the majority of paging traffic in the RNC.
All Radio KPIs are following normal trend, No KPI degradation observed
• HS drop rate=100*(pmNoSystemRbReleaseHs-
pmChSwitchAttemptHsUra+pmChSwitchSuccHsUra)/(pmNoNormalRbReleaseHs+pmNoSystemRbReleaseHs+
pmNoSuccRbReconfPsIntDch+pmPsIntHsToFachSucc+pmChSwitchSuccHsUra)