REAL TIME VISIBILITY OF IPTV SUBSCRIBER EXPERIENCE AND VIEWING ACTIVITY Alan Clark CEO, Telchemy Incorporated
Outline STB centric performance management? Measuring IPTV subscriber experience how and why? Viewing acgvity ra+ngs in real +me Agents in Set Top Boxes IPTV Performance Management Architecture Summary implementagon status and benefits
STB CENTRIC PERFORMANCE MANAGEMENT Real-Time Dashboard Middleware Metrics STB Intelligent performance monitoring agents in STB s measure subscriber QoE, viewing stagsgcs. STB agents push compact reports to management middleware Management middleware collects and reports real Gme performance data
IPTV SERVICE MODEL Multicast IP network Channel GPON or DSL
CHANNEL PROBLEM Multicast IP network * Channel GPON or DSL * *
ROUTER/MSAN PROBLEM Multicast IP network * * Channel GPON or DSL * *
BANDWIDTH PROBLEM Multicast IP network * Channel GPON or DSL * *
STB CENTRIC PERFORMANCE MANAGEMENT Real-Time Dashboard Middleware Metrics STB STB agents measure quality and viewing acgvity Data incorporated into report supports real Gme correlagon Management middleware correlates arriving data in real Gme, idengfies problems and reports real Gme performance data
WHY STB BASED AGENTS? Provides near real Gme visibility of QoE data is reported within minutes ProacGvely detects problems affecgng service Middleware correlates arriving reports as they arrive Provides feedback from customer premise Network probes don t see into the customer home More scalable and less expensive than network based probes
MEASURING SUBSCRIBER EXPERIENCE Goal Measure IPTV service quality delivered to each subscriber Determine impact of significant video encoding or transmission problems on decoded video IdenGfy source of problems Advantages of STB based approach ProacGve see problems quickly CorrelaGon of STB reports can idengfy wide range common factors in real Gme Can detect outages (i.e. absence of reports)
MEASURING VIEWING ACTIVITY Goal IdenGfy channel watched by subscriber, track Gme on channel Aggregate to give real Gme visibility of viewers per channel Advantages of STB based approach Covers engre subscriber base (not sampled) Data available in near real Gme Provides data on less watched channels (which sampling approaches can t)
AGENTS IN SET TOP BOXES Low complexity IPTV performance monitoring agent (~5-10% of STB CPU) SQlive EPG NW I/F DRM Video Decoder Video Output IPTV Set Top Box
STB AGENT ARCHITECTURE MPEG Demux MPEG TR101290 Metrics SQlive I/B/P frame detecgon I/B/P frame detecgon Content Analysis Playout Buffer Emulator Blank, Frozen Image DetecGon Per Frame Quality Codec/ Bandwidth Model MoGon Blockiness Model Report GeneraGon Perceptual Quality Model
STB REPORT Interval Cross channel metrics, overall experience Correla+on tags Per Channel detailed metrics, QoE scores, Video MOS MPEG transport metrics Time spent on channel
MIDDLEWARE Service Quality Alerts Service Quality Alerts Alert Rate User Impact Alert Rate User Impact 1 2 3 Problem: Types Packet Loss Congestion Delay 4 5 7d 24hr 6hr 1h 15m 5m Top Channels Task Area 7d 24hr 6hr 1h 15m 5m Lowest Quality Channels 1 2 3 Problem: Types Packet Loss Congestion Delay 4 5 7d 24hr 6hr 1h 15m 5m Top Channels Task Area 7d 24hr 6hr 1h 15m 5m Lowest Quality Channels Web GUI Dashboards.... 7d 24hr 6hr 1h 15m 5m 7d 24hr 6hr 1h 15m 5m Problems: Correlation Problems: Correlation Core Network Core Network Access Network Access Network Customer Premise Customer Premise.. 7d 24hr 6hr 1h 15m 5m 7d 24hr 6hr 1h 15m 5m ReporGng CorrelaGon Report CollecGon Pre-correlated data in database
MIDDLEWARE IDENTIFYING PROBLEMS Head End Subscriber A Subscriber B Subscriber C Subscriber D Tags Tags Tags Tags Correlate by channel and by correlation tags to identify type and scope of problems
MIDDLEWARE VIEWING STATISTICS Current viewers per channel Subscriber A Subscriber B Subscriber C Subscriber D Tags Tags Tags Tags Correlate by channel and by correlation tags to identify type and scope of problems
MIDDLEWARE VIEWING STATISTICS Service Quality Alert Rate Alerts User Impact 1 2 3 4 5 7d 24hr 6hr 1h 15m 5m Problem: Types Top Channels Packet Loss Congestion Delay.. Top Channels 7d 24hr 6hr 1h 15m 5m Problems: Correlation Core Network Access Network Customer Premise Channel 12. Channel 47 Channel 33 Channel 71 Channel 62 Channel 15 Channel 27 Channel 05 7d 24hr 6hr 1h 15m 5m 7d 24hr 6hr 1h 15m 5m Task Area 7d 24hr 6hr 1h 15m 5m Lowest Quality Channels Lowest Quality Channels Channel 15 Channel 03 Channel 45 Channel 17 Channel 06 Channel 59 Channel 14 Channel 34 Channel 51 Channel 54 See viewing acgvity within (typically) 10 15 minutes StaGsGcs span all channels being watched (down to 1 subscriber) StaGsGcs are anonymized but retains correlagon data Provide feedback to advergsers on viewing acgvity and quality Granularity sufficient for ad tracking
IPTV PERFORMANCE MANAGEMENT ARCHITECTURE Middleware FuncGons CollecGon/ CorrelaGon Agent ConfiguraGon Channel GPON or DSL Agent in STB or ONU
RELATED STANDARDS ACTIVITIES ITU T IPTV QoE Measurement ITU T SG12 P.NAMS Performance Management ReporGng Models ITU T P.IPTV PMR ATIS IIF QOSM Task Force ATIS 0800008 QoS Metrics for Linear Broadcast Industry Forums Broadband Forum VSF
SUMMARY STB centric performance management Measuring IPTV subscriber experience Measuring viewing acgvity Agents in Set Top Boxes IPTV Performance Management Architecture ImplementaGon Status Early commercial deployment Integrated into several major STB product ranges QuesGons?