Microsoft PowerPoint - Interconnected Systems [Kompatibilitätsmodus]Modern TelecommunicationModern TelecommunicationModern Telecommunication Modern Telecommunication Systems and their Impact on QoS Systems and their Impact on QoS
dd Q EQ Eand and QoEQoE H W GierlichH. W. Gierlich
HEAD acoustics GmbH
Parameters influencing QoSg
Memory/ Attitude/ Emotions
Interconnected Systems © HEAD acoustics GmbH 2
Acceptance *based on S. Möller: Assessment and Prediction of Speech Quality in Telecommunications
Speech Communication p
speech
Outline
C i ti t d• Communication systems and interconnection – overview I t f• Impact of • Delay • Coding• Coding • Noise and noise cancellation • Insufficient echo loss/improper echo cancellation • Cascading of signal processing • Missing signaling information between endpoints and
network elements
Terminal - Network typical elements found in modern connectionstypical elements found in modern connections
NLPVAD jitter buffer
NLP VADjitter buffer
BSS MSC ISCMT EC 1 EC 2 EC 3 EC 4GW 1
GW 2
PSTNPSTN IP-Network Provider C
NLP NLPVADjitter buffer
ISCISC LE CTEC 5 EC 6 EC 7 EC 83 4
Interconnected Systems © HEAD acoustics GmbH 5
Provider C
NLPVAD jitter buffer
jitter buffer
BSS MSC ISCMT EC 1 EC 2 EC 3 EC 4GW 1
GW 2
MNA (≈ 80ms)
IPB (10-400ms)
PSTN1 (1-10ms)
Delay aggregation (II)y gg g ( )
Coder 4
jitter bufferVAD
Coder 3
Coder 4
NLP NLPVADjitter buffer
ISCISC LE CTEC 5 EC 6 EC 7 EC 8GW 3
GW 4
5 600
4
500
minmin
max. delaymax. delay
minmin average average
1 0%, 0ms, 0%, 0ms 1%, 0ms 2%, 0ms 3%, 0ms 5%, 0ms 1%, 20ms, 1%, 20ms,
0
100
VAD on r=0.9 (fix delay 50ms)
r=0.5 (fix delay 50ms)
network conditions
Interconnected Systems © HEAD acoustics GmbH 8
average, min. and max. PESQ for all manufacturers Source: ETSI SQTE, G.711, 10 msSource: ETSI SQTE, G.711, 10 ms
Delay aggregation (III)y gg g ( )
toneway = tMT1 + tMNA + tIPB + tPSTN1 + tLDC + tPSTN2 + tIPC + tPSTN3
Best case (no sattelite) = (10 + 80 + 10 + 1 + 20 + 1 + 10 + 1 + 13)ms = 146ms146ms( )
Worst case = (50 + 80 + 400 + 10 + 240 + 10 + 400 + 10 + 13)ms = 1113 ms1113 ms
Interconnected Systems © HEAD acoustics GmbH 9
E-model: Influence of delayy
146 ms/55 dB TCLw 1113 ms/55 dB TCLw
R_ 87,9 MOS (4.5) 4,2
R_ 26,6 MOS (4.5) 1,5
146 ms/35 dB TCLw 1113 ms/35 dB TCLw146 ms/35 dB TCLw 1113 ms/35 dB TCLw
R_ 4,3 MOS (4.5) 1
R_ 82,1 MOS (4.5) 4,1
Interconnected Systems © HEAD acoustics GmbH 10
ITU-T G.114: Effect of transmission time
90
100
users dissatisfied
Terminal - Network
NLPVAD jitter buffer
NLP VADjitter buffer
BSS MSC ISCMT EC 1 EC 2 EC 3 EC 4GW 1
GW 2
jitter bufferVAD
PSTNPSTN IP-Network P id C
NLP NLPVADjitter buffer
ISCISC LE CTEC 5 EC 6 EC 7 EC 8GW 3
GW 4
Provider C
E-Model: The effect of codec tandeming
Type Ie (G.113) C d 1 EFR (ACELP) 5Coder 1 EFR (ACELP) 5
Coder 2 G. 729 10 (CS-ACELP)
Coder 3 G. 723 (ACELP, 5.3 kbit/s)
19 kbit/s)
7 )
=> Ie(tot) = 5+10+19+7 = 46( ) R_ 47,2 MOS (4 5) 2 43
Interconnected Systems © HEAD acoustics GmbH 13
MOS (4.5) 2,43
Instrumental measures based on hearing models -
Modeling the Results of Auditory Tests by Comparison of Reference Speech Signal with Processed Speech Signal
Perceptual models:
Typical Processing Steps (Schematic): Results of Listening Tests!
Comparison, Reference, Reduction, S indexQ
Speech Signal
Signal Valuereference signal
PSQM99 PSQM99 KPN
PACE PACE Ascom
P.OLQAP.OLQA
CDMA CDMA GSM Variable Fixed enhanced rate max. fullrate
rate
1x coding/decoding
Source: Harman/Becker
Vehicle
GSM
MOST
BT
Mi h ( h & i ) & CVSD & FR
Interconnected Systems © HEAD acoustics GmbH 16
(.wav) Microphone (speech & noise) & CVSD & FR
Echo loss in IP – DECT/CATiq configurations • Assuming… G.711 speech coder, no compression • Assumimg… no transcoding
DECT ISDN
Echo canceller “w/o echo“, echo canceller with low level
DECT terminals, delay… . Audible echoes!Audible echoes!
echos… Echoes, clipping!Echoes, clipping!
EC in IAD!... DECT delay, DECT delay, DECT distortions!DECT distortions!
Packet loss and jitter under network load PLC jitter
, y gy g conversational quality!conversational quality!
Interconnected Systems © HEAD acoustics GmbH 17
network load, PLC, jitter buffer… Artifacts, delay!Artifacts, delay!
Echo Cancellation in IAD
EC td
-50
-40
-30
-20
EC
Echo loss 31 dBEcho loss 31 dB td 30 ms
+ Network delay 100, 150, … ms !!!+ Network delay 100, 150, … ms !!!
Interconnected Systems © HEAD acoustics GmbH 18
DECT/CATiq Echoesq
DECTt DECTtd
linear non-linear Spectrum avg. FFT Size:8192 Overlap:0,0% Rectangle L/dB[Pa]
-60
-50
-40
-30
60
-50
-40
-30
60
-50
-40
-30
-20
-100
-90
-80
-70
-60
-90
-80
-70
-60
-90
-80
-70
-60
f/Hz100 200 500 1000 2000 5000
Interconnected Systems © HEAD acoustics GmbH 19
Echo loss 17.8 dB Echo loss 31.0 dB Echo loss 36.3 dB
Tandeming of signal processing: Car hands-free and networkCar hands-free and network
Mobile network
Hands-free Terminal
system
(…both…) ECs
• Information exchange
EC EC exchange between gateway and terminal notterminal not possible
Microphone array i l i
ES NR
NLP, NR,
signal processing , ,
Tandeming of echo cancellation and consequences on double talk performanceon double talk performance
Base Station
Mobile Terminal
gaingain
Double Talk – Performance Limits Characterization of HFT acc. to ITU-T P.340
MOS ≥ 4,0 4,0 – 3,5 3,5 – 3,0 3,0 – 2,5 2,5 – 2,0 ≤ 2,0
TELRDT [dB]
≥ 37 ≥ 33 ≥ 27 ≥ 21 ≥ 13 < 13 [ ] AHsdt [dB] ≤ 3 ≤ 6 ≤ 9 ≤ 12 ≤ 15 > 15
AHrdt [dB] ≤ 3 ≤ 5 ≤ 8 ≤ 10 ≤ 12 > 12
Type 2 Type 1
T 2 T 2b T 2 Type 3
Type 2a Type 2b Type 2c TELRDT [dB] ≥ 37 ≥ 33 ≥ 27 ≥ 21 < 21
AHsdt [dB] ≤ 3 ≤ 6 ≤ 9 ≤ 12 > 12Hsdt [ ]
AHrdt [dB] ≤ 3 ≤ 5 ≤ 8 ≤ 10 > 10
Interconnected Systems © HEAD acoustics GmbH 22
Background noise transmission
gaingain ( )
Testing and Qualification
The new Objective Model for Speech with Background Noise: ETSI EG 202 396ETSI EG 202 396--33
S-MOS (Speech-MOS)
N-MOS (Noise-MOS)
Interconnected Systems © HEAD acoustics GmbH 25
Summaryy
• Delay aggregation is the biggest problem in modern y gg g gg p network configurations!
• Potential problems in a connection may occur due to tandemed signal processing in terminals and networkstandemed signal processing in terminals and networks
• Network planning and quality evaluation needs to involve all components of a connection and their interaction!
• Currently no signaling is possible between terminals and networks for information exchange about signal processing active in the different devicesp g
• Related work: • G.799.2 developed in SG16: Mechanism for Dynamic
Coordination of Signal Processing FunctionsCoordination of Signal Processing Functions • G.115.x sub series (Logic and protocols for the control of signal
processing network elements and functions) may potentially be enhanced to support this capability
Interconnected Systems © HEAD acoustics GmbH 32
pp p y