nokia volte optimization white paper 071114

Report
Nokia Networks
Voice over LTE (VoLTE)
Optimization
Contents
1.
Introduction
3
2.
VoIP Client Options
5
3.
Radio Network Optimization
6
4.
Voice Quality Optimization
11
5.
Handset Power Consumption Optimization
17
6.
Summary
19
7.
Further Reading
19
Page 2
networks.nokia.com
1.
Introduction
Mobile operator networks carry tens of billions of minutes of voice
in 2G and 3G technologies. Most LTE operators serve voice calls via
architecture.
multimedia services. The main reasons for adopting VoLTE are
summarized in Figure 1.
Provide faster call setup time
Migrate from dual radio CDMA + LTE devices
to LTE only devices
Enhance voice quality with wideband codec
Take benefit of low band LTE for
extended coverage
Enable simultaneous voice and LTE data
Improve spectral efficiency
Provide evolution from voice to rich next
generation IMS services
Prepare evolution to LTE only deployments
Fig. 1. Motivations for VoLTE.
Page 3
networks.nokia.com
a number of aspects of VoLTE optimization in radio networks and in
power consumption. The white paper also compares VoLTE with CS
operator voice can be run over Wi-Fi radio. If evolved packet core is
handset support.
white paper.
The radio network optimization solutions are considered in this paper
on the optimization of handset power consumption. The contents
are summarized in Figure 2.
Radio network optimization
End user quality optimization
Terminal power consumption optimization
Fig. 2. Contents of this paper.
Page 4
networks.nokia.com
2.
VoIP Client Options
interworking with legacy CS networks. The VoLTE client can also
for voice packet prioritization. It is also possible to have a non-native
VoLTE client from third parties which runs on top of the application
1.
Native VoLTE client integrated in the handset chip set. The
number of commercial handsets with native VoLTE clients is
growing rapidly.
2.
Non-native VoLTE clients. Third party applications which can
and Viber.
Page 5
networks.nokia.com
3.
Radio Network Optimization
The success rate and the retainability of a VoLTE call must match and
exceed the level provided by CS connections. The key performance
rate and call completion success rate. Network optimization also aims at
The network optimization includes parameter optimization and feature
and 54% no activity. The throughput shows the impact of the codec bit
kbps with the codec rate of 23.85 kbps and 8.8 kbps with the codec rate
of 12.65 kbps.
using voice activity detection. The lower average throughput in this
of the Mean Opinion Score MOS. This same application with Enhanced
on the codec they used and the particular implementation of features
such as codec rate adaptation or voice activity detection.
45
42.8
40
35.8
35
kbps
30
25
20
17.6
17.3
15
10
10.2
8.3
8.8
5
0
Page 6
SIP EFR
SIP AMR OTT AP1 OTT AP2
WB
OTT AP3
VoLTE
VoLTE
23.85 kbps 12.65 kbps
Fig. 3. Measured throughput
networks.nokia.com
The throughput measurements shown in Figure 3 include full IP
header can be larger than the voice packet if header compression is
maximizing VoLTE capacity.
45
Header compression
40
Header size (byte)
35
30
25
20
Original header size
Uplink header size
Downlink header size
15
10
5
0
Page 7
networks.nokia.com
VoLTE uplink performance in the weak signal can be enhanced with TTI
bundling which allows the handset to repeat the same transmission
in four consecutive 1 ms TTIs. TTI bundling makes the uplink more
up substantial radio resources. TTI bundling is switched on only when
the handset hits the edge of the coverage area. TTI bundling runs
between the base station and the handset.
80.0
PUSCH BLER comparison
72.7
70.0
BLER (in %)
60.0
With TTI bundling
Without TTI bundling
50.0
40.0
30.0
20.0
10.0
9.0
0.0
Page 8
networks.nokia.com
in highly loaded cells. Figure 6 shows an example VoLTE call setup success
rate in a live network including tens of millions of VoLTE calls during the
excellent VoLTE availability can be achieved even in the loaded network.
The network carried tens of billions of packet calls at the same time.
Another important factor for VoLTE is reliable mobility. The handover
success rate can be optimized with RF planning and new features included
Optimization. This enables radio level connection re-establishment in the
event of handover failure.
100
VoLTE call setup success rate
98
96
(%)
94
92
90
Three-month period
Fig. 6. VoLTE call setup success rate.
Page 9
networks.nokia.com
can be handed over to a CS connection in a 3G or 2G network. This
leg change. The SRVCC probability also depends on the parameter
shows the probability of SRVCC in an example network. The SRVCC
probability is reduced by 7% to below 3% by network optimization.
More than a million SRVCC attempts are included in this graph.
SRVCC functionality is available for QCI1 connections but not for OTT
10%
Percentage of VoLTE Calls Using SRVCC
9%
8%
7%
6%
5%
4%
3%
2%
1%
0%
Three-month period
Fig. 7. Probability of VoLTE call using SRVCC.
Page 10
networks.nokia.com
4.
Voice Quality Optimization
acoustics may limit the maximum bandwidth provided by the speech
26.131. The bandwidth is illustrated in Figure 8. The CS connections
AMR - NB
80 - 3700 Hz
0.05 0.1
1
2
AMR - WB
50 - 7000 Hz
3
4
5
6
7 kHz
Fig. 8. Audio bandwidth of narrowband and wideband AMR.
the performance of the telephone transmission system by listening
Page 11
networks.nokia.com
codec scored only slightly lower with 2.7. The same SIP client with an
this and other third party SIP clients could be increased to 3.4 or 3.6
by tweaking some optional functionality such as deactivating voice
transmit a constant data stream regardless of whether the speaker
was talking or silent. The OTT VoIP applications scored between 4.1
to the native VoLTE client.
That codec will make it possible for VoLTE to match and beat the voice
4.5
4.0
MOS
3.5
3.0
2.5
2.0
1.5
1.0
Page 12
3G CS
AMR-NB
SIP EFR
SIP
AMR-WB
OTT AP1
OTT AP2
OTT AP3
VoLTE
networks.nokia.com
OTT VoIP as a function of network loading. VoLTE uses QoS Class
OTT VoIP decreases as a function of loading. The OTT VoIP call also
OTT VoIP-Good
OTT VoIP-Bad
VoLTE-Good
VoLTE-Bad
4.5
4.0
MOS
3.5
3.0
2.5
2.0
OTT VoIP fails
to work
1.5
1.0
2
21
192
382
762
1902
Effective number of non-GBR data users
Page 13
networks.nokia.com
Mouth-to-ear delay as a function of loading is shown in Figure 11.
the base station. The delay of OTT VoIP increases as a function of
simultaneous data loading in the cell.
OTT-Bad
VoLTE-Good
300
352
352
491
525
451
500
427
541
600
300
Mouth to Ear Delay (ms)
700
400
VoLTE-Bad
661
OTT-Good
200
OTT VoIP fails
to work
100
0
2
21
192
382
762
1902
Effective non-GBR Load (users)
Page 14
networks.nokia.com
the call establishment time compared to the legacy CS systems. The
total time for call establishment depends on many factors such as
whether the handsets were initially in RRC Idle or RRC Connected
state when the call was originated and the latency experienced by the
signaling network. The laboratory measurements show a VoLTE call
depending on the operator network and transport architecture. The
corresponding typical CS call setup time is four seconds and with CS
Call setup time
7.0
6.0
6.0
Seconds
5.0
3.8
4.0
3.0
2.4
2.2
2.0
0.9
1.0
0.0
idle - idle
(field)
idle - idle
VoLTE
connected connected
3G idle CSFB - CSFB
idle
CS voice
Fig. 12. Call setup time measurements.
Page 15
networks.nokia.com
strategy for incoming VoLTE calls. VoLTE paging logic is executed in a
procedure by marking packets related to the incoming voice calls with
Figure 13 shows an example of paging response delay distribution. A
more aggressive timer value for paging a re-try timer could be applied
Number of Samples
1000
100
10
9.5
9
8.5
8
More
Delay between Paging and Service Request
7.5
7
6.5
6
5.5
5
4.5
4
3.5
3
2.5
2
1.5
1
0.5
0
1
Fig. 13. Paging response delay distribution in live network.
Page 16
networks.nokia.com
5.
Handset Power Consumption
Optimization
users. It can be extended by using advanced features in the radio
network and by optimized handset design. The two main factors
The power optimization solutions are shown in Figure 14.
Handset architecture
optimization
• VoLTE integrated to the chip set
• Application processor can enter sleep mode
Radio features
including DRX
• Discontinuous reception (DRX)
• Modem activity from 100% to below 50%
Fig. 14. Mobile handset power saving solutions.
Page 17
networks.nokia.com
which however needs a lot of power. Power consumption can be
minimized by integrating a VoLTE client to the modem processor.
reception allows the use of power saving sleep mode even between
instantaneous handset power consumption during a VoLTE call when
the person is listening and not talking. The short power peaks happen
power consumption is considerably lower between the packets.
Current (mA)
600
Current (mA)
500
400
300
200
100
656.52
656.5
656.48
656.46
656.44
656.42
656.4
656.38
656.36
656.34
656.32
656.3
656.28
656.26
656.24
656.22
656.2
656.18
656.16
656.14
656.12
656.1
656.08
656.06
656.04
656.02
656
655.98
655.96
0
Time (s)
is shown in Figure 16. VoLTE power consumption can be reduced by
even slightly below the 125 mA measured on the same codec on 3G
CS using the same handset. The power consumption of the OTT VoIP
application is substantially higher at 248 mA.
Page 18
networks.nokia.com
300
Current (mA)
250
248
235
-50%
200
154
150
129
131
125
121
100
50
0
3G CS
AMR-NB
3G CS
AMR-WB
noDRX
DRX on with different settings
OTT VoIP
consumption for VoLTE than for OTT VoIP.
6.
Summary
setup times. This paper illustrates that optimization of radio features
and parameters is needed to provide reliable VoLTE connections with
high success rates and low drop rates. The live network results show
that excellent key performance indicators can be obtained for VoLTE in
optimized networks.
Reliability is obtained by using QoS features in the radio network and
7.
Further Reading
•
paper.pdf
•
Page 19
networks.nokia.com
Public
Nokia is a registered trademark of Nokia Corporation. Other product and company names mentioned herein may be trademarks or trade names of their
respective owners.
Nokia
Nokia Solutions and Networks Oy
Finland
Finland
networks.nokia.com

similar documents