[ ] ITU-T ATHENS FUNCTIONS AND MONITORING OF THE QOE AND QOS IN 4G NETWORKS/LTE. Cyril Lau ITU-T, Athens, September 2015 Who we are Independent Auditor +30 patents worldwide 1 Innovative More than Leader 30 patents worldwide 2 Cyril Lau ITU-T, Athens, September 2015 Next-generation technology ITU-T ETSI VQEG Wireless network performance 3 Formative Player Agenda • LTE QoE- QoS- KPIs MAPPING RELATIONSHIPS • SOME CHALLENGES EMERGED FROM THE LTE ECOSYSTEM • ASCOM LTE QoS/QoE TESTING APPROACHES • SHARING EXPERIENCE: VoLTE Testing • CONCLUSIONS Cyril Lau ITU-T, Athens, September 2015 [ ] LTE QOE- QOS- KPIS MAPPING RELATIONSHIPS Cyril Lau ITU-T, Athens, September 2015 ASCOM APPROACH WITHIN THE ITU-T PERSPECTIVE ON QoS/QoE CYCLE (ITU-T G.1000) Network/service provider Subscriber/user QoS Requirements of Subscriber/User OR QoE Requirements Execution gap Value gap QoS Perceived by Subscriber/User OR QoE of the User QoS Targeted by Provider (Target SLA) Alignment gap QoE/QoS Cycle Perception gap KPI, QoS QoS Delivered/Achieved by Provider (Achieved SLA) QoE Subscriber-centric Quality (QoE) Cyril Lau ITU-T, Athens, September 2015 Network-centric Quality (QoS) LTE QoE- QoS - KPIs MAPPING Network Performance Customer Experience MME / S-GW MME / S-GW S1 S1 S1 S1 X2 E-UTRAN eNB eNB X2 X2 eNB Cyril Lau ITU-T, Athens, September 2015 COST EFFICIENT TOP DOWN APPROACH ENABLED BY PREDEFINED LTE QoE-QoS-KPIs MAPPINGS Increased operational efficiency Top-down customer experience centric approach Voice Service: ‘97 Auryst, ‘03 PESQ, ‘12 POLQA Video-Audio Services: ‘08 VSQI/MTQI, ’12 PEVQ, ‘12 VQmon Application Layer (integrity, accessibility, retainability) QoE Video Streaming QoE Behavior QoS / SLAs QoS/KPI Upper Layers (IP/User Data Protocol , IP/Transport Control Protocol): Throughput, Delay, Packet Loss Layer 2& Layer 3 messaging reports related to QoE dimensions e.g. Packet Data Protocol context, HO info, codec usage), 4 3 2 1 0 MOS - QoE packet Loss 0 50 Time (sec) 100 Re-buffering Network performance (KPIs) Physical layer (e.g. coverage, interference) Reduce troubleshooting size and time by using QoE centric mapping to QoS and KPIs per service type Cyril Lau ITU-T, Athens, September 2015 7 [ ] SOME CHALLENGES EMERGED FROM THE LTE ECOSYSTEM Cyril Lau ITU-T, Athens, September 2015 THE LTE ECOSYSTEM CHALLENGE – Why QoE? MOBILE VOIP (0.3% OF DATA CONSUMPTION) MOBILE VIDEO (70.5% OF DATA CONSUMPTION) Cyril Lau ITU-T, Athens, September 2015 ACCOUNT FOR ~70% OF TRAFFIC ACCOUNT FOR ~?% OF REVENUE VoLTE-ViLTE-RCS EXPERIENCE Call quality as perceived by users MOS / ITU-T P.863; J.247 & P.120x.x based (video) Codecs (KPI) On device VoLTE client (MTSI client): rebuffering/time scaling for voice; error concealment type for video Voice path delay (“mouth to ear”), echo, video – voice lip sync Devices Clients (KPI) RTP Packet loss, latency, jitter Call Experience (QoE) HOIT (LTE HO, eSRVCC) Throughput Network (QoS/KPI) Voice /video codec type and bit rates Call control performance Session set-up: SIP signaling statistics / IMS Registration, RTT Session Accessibility, Retainability QCI allocation verification LTE RRC connection and HO statistics Cyril Lau ITU-T, Athens, September 2015 VoLTE QoE AND ITS ROOT CAUSES (QoS, KPIs SOURCES) Perceived frequency Spectrum (QoE) Interruptions (QoE) (incl. time clipping) Network (QoS): Limited Bandwidth Network (QoS/KPI) Device (KPIs): Spectral shaping Reverberations Device based signal processing (KPIs) (e.g. (IP/IMS loss, jitter, RAN erroneous bits, RAN HO) NR, EC) Codec/client (KPIs): PL concealment schemes Aggressive VAD schemes Cyril Lau ITU-T, Athens, September 2015 “Mouth to ear delay” (QoE) Network (QoS) (IMS path can be key contributor) Codec/device signal processing (KPIs) Noisiness (QoE) (incl. musical noise) Network (QoS): Limited BWD with noisy speech (speech contamination) Device (KPIs): Imperfect NR (musical noise) Loudness during silent periods Non-optimal loudness levels Codec/client (KPIs) PLC interpolation based (“additive artifacts”) Perceived call session performance (QoE) IMS network (QoS/KPIs): SIP statistics, IMS registration ITU-T on going work: G.VoLTE, P.TCA (Technical Cause Analysis) MOBILE VIDEO STREAMING EXPERIENCE Codec types: Terminals/devices/clients High • Different form factors Low • Different 2D/3D displays Adaptive bit rates • Adaptive error concealment schemes Network centric: Wide variety of content & bit rates Packet loss Jitter Allocated GBR and QCI HTTP/TCP vs. RTP/RTSP Cyril Lau Service-Centric ITU-T, Athens, September 2015 MOBILE VIDEO STREAMING QOE AND ITS ROOT CAUSES (QoS, KPIs SOURCES) QoS Low throughput (limited bandwidth) MME / S-GW Video resolutions/rescaling Encoding/transcoding rates Compression schemes Packet loss , discard (RTP), late arrival (TCP) IP delays / jitter S1 S1 S1 X2 E-UTRAN Limited coverage, interference eNB eNB X2 Bandwidth estimation algorithms impacting the bit rate selection KPIs S1 Possible additional transcoding X2 KPIs MME / S-GW eNB Content server load, (competing video streams) Content complexity (e.g. low to high movement) KPIs Client under/over flow; improper buffer lengths/adaptation length Initial buffering settings Display resolutions, form factors Cyril Lau ITU-T, Athens, September 2015 Visual impariments Bockiness, Blueriness, Jerkiness, Freezing with and/or without skipping Perceived service accessibility / access time QoE [ ] ASCOM LTE QOE/QOE TESTING APPROACHES Cyril Lau ITU-T, Athens, September 2015 Measuring QoE: TEST THE NETWORKS LIKE A REAL CUSTOMER IS USING IT A CUSTOMER CENTRIC TESTING SOLUTION Enables testing of end user terminal QoS settings and IP stack characteristics Provides framework for future services to be added easily and controlled by a single client (e.g. Blixt implementation) Cyril Lau ITU-T, Athens, September 2015 Service M IP Logging Repeatable “control scripts” supported across multiple devices (repeatable testing) Service B Call Control First to test VoLTE-ViLTE as a user via on-device VoLTE client , test everything ON Device Server resides inside the terminal, managing the supported services (VoLTE, IP Logging, Call Control etc) Service A VoLTE Laptop-based drive test solution (TEMS Investigation) maintains control and coordination through an On-Device Server QoS setting per service TCP/IP (SIP, RTP, etc) On-Device Server UNDERSTAND AND USE ITU-T BASED QoE METRICS BEST FITTED TO THE TESTED SERVICE; LTE VOICE AND MOBILE VIDEO SERVICES LTE Broadcast/ Streaming server Video stream (YouTube, eMBMS) IP recordings (RTP KPIs/QoS) synchronized with RAN KPIs Encoding Cellular Voice & Data Network Decoding Conversational Video Solutions Voice: MOS P.OLQA (ITU P.863) Video: MOS PEVQ (ITU J.246) Voice Video stream Voice/Conversational Video IP transport/payload parameters Full Reference listening media quality evaluation (intrusive, perceptual) measurement Direct RTP KPIs/QoS! MOS, media based KPIs Solutions Voice: MOS - ITU P.564 based Video-audio: MOS – P.120x.x based (VQmon) Non Reference listening media quality evaluation (non-intrusive, parametric) MOS, network based KPIs Cyril Lau ITU-T, Athens, September 2015 [ ] SHARING EXPERIENCE: VOLTE TESTING Cyril Lau ITU-T, Athens, September 2015 VoLTE MOS SCORE and VoLTE KPI’s/QoS MOS Speech GSMA IR 92, IR 94 QoE Evaluation & Troubleshooting Speech & Client centric reasons Client [TEMS Discovery] [TEMS Investigation] [TEMS Pocket] POLQA; speech path delay, volume, echo Client Information (re-buffering, codec) ESM configuration (QCI, RoHC) Real time IP trace & L3 logging IP recording (RTP stats), SIP stats RAN (HOIT, Scheduling, RSRP, CINR, CQI, PMI/RI, UE category, MTU Size, Protocol stack configuration) Cyril Lau ITU-T, Athens, September 2015 Network Network centric reasons VoLTE CALL SESSION Accessibility Statistics Session Setup Delay (s) [Count] [All] [MO] [MT] 100 90 80 70 60 50 40 30 20 10 0 1 2 3 4 5 [sec] Registration Delay (ms) All Power Up Re-registration % of Total % of Total 100.0 90.0 80.0 70.0 60.0 50.0 40.0 30.0 20.0 10.0 0.0 50.0 40.0 45.0 35.0 40.0 30.0 35.0 25.0 30.0 20.0 25.0 20.0 15.0 15.0 10.0 10.0 5.0 1 2 3 4 5 6 5.0 0.0 0.0 1 Cyril Lau IMS Latency (ms) SIP Handshake Time (ms) ITU-T, Athens, September 2015 2 3 4 5 6 1 2 3 4 5 6 VoLTE CALL SESSION Retainability Statistics Session Completion Success Rate (%) Session Duration Time Session Disconnect Delay (ms) % 40 300 35 250 2 30 200 [Count] [Count] 3 150 100 50 1 2 3 4 [Duration] 90.0% 92.0% 94.0% 96.0% 98.0% 20 15 10 0 1 25 5 6 7 5 0 1 2 3 4 5 [ms] 100.0% RRC Connection Setup Time (ms) RRC Connection Setup 500 [Count] 400 300 1 200 2 100 0 1 2 [ms] Cyril Lau ITU-T, Athens, September 2015 3 VoLTE CALL SESSION Retainability Analysis Accessibility Analysis Session Setup Failure Causes RRC Connection Drop Causes 1. 1. 2. 1. 1 2 40. 3 4 Session Completion Failure Cause 5 6 8.33 25. 1 58.33 8.33 2 3 4 Cyril Lau ITU-T, Athens, September 2015 VoLTE CALL SESSION Integrity Evaluation Conversation Start Delay (ms) Speech Path Delay (ms) All MO Acceptable Call Quality MT MOS Threshold=2.7 [Distribution] 70.0 60.0 250 5. 50.0 1 40.0 150 2 30.0 100 20.0 50 10.0 43. 0.0 0 1 2 3 1 2 3 4 5 6 7 [ms] Interarrival Jitter 300000 250000 [Count] [Count] 200 200000 150000 100000 50000 0 1 2 3 4 [ms] Cyril Lau ITU-T, Athens, September 2015 5 6 VoLTE CALL SESSION Integrity Analysis Handover Interruption Time Inter/IntraeNB - Control Plane (ms) Low POLQA Category - Distribution To be correlated to user perceived interruptions (speech interruptions - to be soon released by TEMS 100 90 80 [Count] 70 60 50 40 30 20 10 0 1 2 3 4 [ms] Cyril Lau ITU-T, Athens, September 2015 5 6 3G CS to VOLTE voice call Now we know what QoE metrics we can measure, how do we locate the problem area when there is an issue? UMTS PSTN UE A UE B Sv I2 Cx Sh Mr Iu-CS C S6a Mw ISC Rx S5 S1-MME Iu-PS ISUP Gn Gm SGi Gx S11 S1-U Dh Dx Uu Gr RTP LTE-Uu More than 28 interfaces 1 VoLTE Service Monitoring Solution, © Ascom Extract them from all around the NW Check Compliance individually Correlate them all together 24 Only one party VOLTE voice call monitoring network architecture TRANSPORT layer and traffic management ISBC1 HSS2 HSS2 HSS1 MME1 PCRF1 ISBC2 ISBC1 PCRF2 PCRF1 P-CSCF1 I/SCSCF1 TAS1 MRF1 P-CSCF2 I/SCSCF2 TAS2 MRF2 P-CSCF3 I/SCSCF3 TAS3 MRF3 P-CSCF4 I/SCSCF4 TAS4 MRF4 P-CSCF5 I/SCSCF5 TAS5 MRF5 P-CSCF6 I/SCSCF6 TAS6 MRF6 SEG1 S-GW1 eNodeB1 P-GW1 SEG2 MME2 S-GW2 P-GW2 SEG3 MME2 eNodeB2 SEG4 S-GW2 P-GW3 EPC 2 IMS More than 10 network elements involved Consider a high availability Network (Network Element Failure ≠ Service Unavailability) VoLTE Service Monitoring Solution, © Ascom 25 Ascom approach: Combined Core/RAN active testing Test point 1: E2E QoE Test point 3: IMS QoS Test point 2: E2E QoS HSS Rx S6a PCRF MME Rx P-CSCF S-CSCF S1-U S5/S8 SGW Mr’ MRF PGW RTP EnodeB LTE TAS Gx Gm Uu ISC EPC IMS Emulated Interface Internal Interface Media Path VoLTE Solution 2015 © Ascom 26 Measuring QoE: E2E testing • E2E testing using a commercial UE. • The SUT effectively includes the whole of oprators’s core and radio network, plus the UE itself! • QoE is affected by UE firmware, SW app under test….etc • This is useful for problem identification, but wide coverage makes it difficult to locate the point of failure. TOOL SUT (System Under Test) HSS Rx S6a PCRF MME Test UE Rx P-CSCF S-CSCF S1-U S5/S8 SGW Mr’ MRF PGW RTP EnodeB LTE TAS Gx Gm Uu ISC EPC IMS Emulated Interface Internal Interface Media Path VoLTE Solution 2015 © Ascom 27 Measuring E2E delivery network QoS • E2E testing using a simulated device under a known, controlled environment. • Focus on testing the service delivery network (QoS), by eliminating the UE from the test. TOOL SUT (System Under Test) HSS Rx S6a UE simulation PCRF MME Rx P-CSCF S-CSCF S1-U TMM SW S5/S8 SGW Mr’ MRF PGW RTP EnodeB LTE TAS Gx Gm Uu ISC EPC IMS Emulated Interface Internal Interface Media Path VoLTE Solution 2015 © Ascom 28 Measuring IMS core network QoS • Active testing of IMS core network with simulated traffic. • Focus on the QoS delivered by the IMS core network TOOL SUT (System Under Test) Rx PCRF Rx P-CSCF S-CSCF ISC TAS Gx Mr’ Gm PGW simulation Test MRF agent RTP LTE EPC IMS Emulated Interface Internal Interface Media Path VoLTE Solution 2015 © Ascom 29 [ ] CONCLUSIONS Cyril Lau ITU-T, Athens, September 2015 ASCOM APPROACH WITHIN THE ITU-T PERSPECTIVE ON QoS/QoE CYCLE (ITU-T G.1000) Network/service provider Subscriber/user QoS Requirements of Subscriber/User OR QoE Requirements Execution gap Value gap QoS Perceived by Subscriber/User OR QoE of the User QoS Targeted by Provider (Target SLA) Alignment gap QoE/QoS Cycle Perception gap KPI, QoS TEMS QoS Delivered/Achieved by Provider (Achieved SLA) QoE Subscriber-centric Quality (QoE) Cyril Lau ITU-T, Athens, September 2015 TEMS Network-centric Quality (QoS) CONCLUSIONS PLEASE VISIT OUR WEBSITE: http://www.ascom.com/nt/en/index-nt/about-us-network-testing/nt-about-usresources.htm/ White papers: VoLTE, Video Streaming, HetNets, Carrieir Aggregation and...watch the space: eMBMS testing to come soon Informa Webinar: Advanced testing with Ascom in LTE networks Webinars: VoLTE, Video Streaming Cyril Lau ITU-T, Athens, September 2015 [ ] THANK YOU Cyril Lau ITU-T, Athens, September 2015