An Open-source Software Toolkit for Professional Media over IP (ST 2110 and more) IEVGEN KOSTIUKEVYCH

Transcription

1 An Open-source Software Toolkit for Professional Media over IP (ST 2110 and more) IEVGEN KOSTIUKEVYCH IP Media Technology Architect EBU Technology & Innovation

2 UDP Multicast Massive bandwidth (min. 1.2 Gb/s)

3 TRAFFIC ON MICROSCOPIC SCALE β Bandwidth average Ideal Packet Spacing (IPS) t β Ideal Traffic Shape on microscopic scale t β Bursts Early packets Late packet Actual Sender on microscopic scale t

4 WHAT MAKES IT ALL WORK TOGETHER? SMPTE ST timing (SMPTE ST2059) and timestamping SMPTE ST buffers and packet pacing models

5 ST Constrains the packet delay variation (PDV) of a sender Describes two virtual leaky buckets

6 Sender C MAX VRX FULL C INST VRX INST Drain Rate = 1/T DRAIN (packets / second) Network compatibility model Packet j drains at time TPR J Virtual receiver buffer model

7 Tested at the output of the sender NETWORK COMPATIBILITY MODEL Measures PDV introduced by a sender ST Sender Has constant drain rate C MAX C INST Evaluates sender s compliance with network Drain Rate = 1/T DRAIN (packets / second)

8 IMPACT OF PACKET BURSTS WITH CORRELATION TO C VALUE Egress total 3.2 Tbps Cmax 4 Btotal 16 MB Ractive 96% Beta 1.1 Usage of Egress (1/Beta) and 100% of memory 1/Beta 90% 80% 70% 60% 50% 40% 30% 20% 10% 4.855E E E E E E E E E Resolution Hz Npackets Packetrate C Resolution Hz β βgap #Streams Courtesy of Willem Vermost (vermost@ebu.ch)

9 VIRTUAL RECEIVER BUFFER MODEL Similar to the Network Compatibility Model Drain rate is based on a receiver s packet read schedule VRX FULL ST Sender Evaluates sender s compliance with receiver VRX INST Packet j drains at time TPR J

10 LINEAR VS GAPPED PACKET READ SCHEDULE (PRS) N T %&'() Since Epoch T => T &7 (linear ) (N + 1) T %&'() Since Epoch Linear Packet Read Schedule TPR A TPR A N FGHIJKL T &7(TUVRPW) t TR C%%7)D TR C%%7)D T &7 (gapped) Gapped Packet Read Schedule TPR A TPR B t (N FGHIJKL MN ) T &7(OPQQRS)

11 NARROW Narrow (N) senders: Have tight PDV values C MAX = 4 VRX FULL = 8 Designed for gapped PRS Narrow Linear (NL) senders: Have tight PDV values C MAX = 4 VRX FULL = 8 Designed for linear PRS TYPES OF SENDERS WIDE Wide senders: Have loose PDV values C MAX = 16 VRX FULL = 720 Calculated values are for: 720p60, 1080i50, 1080p50

12 TYPES OF RECEIVERS Narrow constrains Narrow sender Wide constrains Wide sender Asynchronous doesn t have to be PTP locked

13 SENDER-RECEIVER COMPATIBILITY N W A N YES YES YES NL NO YES YES W NO YES YES

14 RTP PACKET ANATOMY RTP Header RTP Payload Header Flags Sequence Number Timestamp Identifier Extended Sequence number Payload Marker bit RTP Payload

15 CALCULATIONS BASED ON RTP TIMESTAMP Having timestamp, sequence number and marker bit values and PTP: Exact field/frame arrival time PTP time of the packet Index of received frame

16 N T %&'() Since Epoch T => (N + 1) T %&'() Since Epoch (N + 1) T => T &7 Assumed Packet Read Time TPR A TPR B TPR BXY TPR A t TR C%%7)D TR C%%7)D Packet Arrive Time TPR A t Δt A Δt \ Δt \XY

17 EBU SMPTE ST2110 ANALYZER Developed by Willem Vermost Offers PTP offset, C and VRX buffers analysis Available as open source software

18 LIST - Live IP Software Toolkit

19 Backend Frontend

20 list.st2110 list.libpcap list.pcap list.ptp list.rtp LIST library list.net External library list.core bimo GSL spdlog fmt boost::asio

21 LIST - Live IP Software Toolkit LIST - Live IP Software Toolkit

22 Number of streams PTP clock adjustments Preprocessor Per stream: Sample rate Stream config heuristics Packets per frame Database PCAP file Extractor Packet analysis Video and audio parsing C PEAK and VRX calculation RTP timestamp analysis

23 HIGH PRECISION PTP

24 LESS PRECISE PTP

25 EXAMPLE OF POLITE SENDER (FOR C BUFFER)

26 EXAMPLE OF BURSTY SENDER (FOR C BUFFER)

27

28 EXAMPLE OF POLITE SENDER (FOR VRX BUFFER)

29 EXAMPLE OF UNDERFLOW OF VRX BUFFER

30 EXAMPLE OF OVERFLOW OF VRX BUFFER

31 KEY TAKEAWAY Measurements of mission critical parameters in your network are. mission-critical Every parameter of the sender can be extracted from the stream capture and measured Compliance to the standards can be evaluated Risks can be lowered Better view on the devices in the network Education and training of staff The tools are there