Media: Voice and Video in your SIP Environment

1
Media Voice and Video in your SIP Environment
Jitendra Shekhawat
2
Agenda
Objective Introduction of Media in the SIP
environment.

• Common Audio and Video Codecs
• Media/Codec Negotiations
• Tuning Your Network for Voice and Video
• QoS issues, metrics and user quality expectations

3
IP Audio/Video Telephony Network

• Call Control SIP
• Media RTP
• Video H263, H264, MPEG4
• Audio G711, G723, G729, G726, AMR-NB, etc.

SIP Video Endpoints
SIP Soft Phone
SIP Desk Phone
SIP
SIP
RTP
RTP
PC Email Client
Multimedia Server
SIP
SIP
Broadband Users
RTP
RTP
RTSP

• Applications
• Video Mail
• Video Portal
• Live content streaming

CNN, ESPN, Bloomberg, live feed
4
SIP Call Example
5
Audio Video Codecs and Payload Types

• RFC 3551
• Some codecs

6
Media Transport

• RTP
• Real Time Transport Protocol
• media packet transport
• One stream per direction between endpoints
• RTCP
• RTP Control Protocol
• Provides quality information
• Generate reports to the network

7
RTP Packet
RTP Datagram
RTP Datagram
RTP Datagram
IP Header 20 bytes
UDP Header 8 bytes
RTP Header 12 bytes
RTP Payload N bytes
Version 2 bits
Padding 1 bit
Extension 1 bit
CSRC count 4 bits
Marker 1 bit
Payload Type 7 bits
Sequence Number 2 bytes
Time stamp 4 bytes
Source Identifier 4 bytes
8
RTCP Packet

• Receiver of RTP stream sends periodic updates to
  the originator
• Packet count
• Byte count
• Packet loss
• Timestamps to assess round-trip delay
• Jitter

9
RTP Packet Payload size
Function of codec speed, frame-size
Frequency packets sent

• Example g.711, 20 ms frames 64000 bps X 20 msec
  / 8 160 byte payload

codec speed X frame size
Payload size
8 X 1000
bits/byte
msec / sec
10
Media Stream (RTP) Bandwidth

• Packet size Header Payload
• Header Ethernet (IP UDP RTP) 38 (20
  8 12) 38 40 bytes
• Payload depends on codec
• Example g.711, 20 ms frames (50 packets/s)
• 160 byte payload (38 40) byte header
• IP bandwidth 200 byte/packet 80,000 bps ? 160
  kbps for 2 way
• Ethernet bandwidth 238 byte/packet 95,2000 bps
  ? 190.4 kbps for 2 way
• Ethernet Preamble (8) Ethernet Header (14)
  Ethernet CRC (4) Inter-frame gap (12) 38

11
Codec Bandwidths
Coder Bitrate Encoded bandwidth
G.711 64 kbps 200-3400 Hz
G.723 5.4 or 6.3 kbps 200-3400 Hz
G.729A (20ms Packet) 8 kbps 200-3400 Hz
AMR 4.75 to 12.2 kbps 200-3400 Hz
AMR-WB Variable 6.6 up to 23.85 (non-continuous) 50 to 7000 Hz
AMR-WB Variable 6-36 kbps (mono) or 7-48 kbps (stereo) 50 Hz 7.2 kHz up to 50 Hz 19.2 kHz
iLBC 13.33 kbps for 30 ms, 15.20 kbps for 20ms 200-3400 Hz
12
Codec Bandwidths
Coder IP Bandwidth / RTP stream
G.711 (30 ms Packet) 74.6 kbps
G.711 (20ms Packet) 80 kbps
G.711 (10 ms Packet) 96 kbps
G.723.1 (30ms Packet) 15.7 kbps
G.729A (20ms Packet) 24 kbps
AMR (20 ms) 20.4 - 28 kbps
AMR-WB (20ms) 22.4 39.6 kbps
AMR-WB (20ms) 22 52 kbps
iLBC (20ms or 30ms) 31.2 kbps or 24 kbps
13
Video streams
I-frames (Key frames)
P-frames (predicted frames)
Frame Sequence
14
Video Formats (IP vs. 3G)

• High resolution for IP networks
• More bandwidth available
• SIP Video Phones are generally CIF size (352
  288 pixels)
• Recommended CIF, 15 or 30fps, up to 384kbps
• Low resolution for 3G networks
• Total bandwidth limited to 64kbps
• Generally video audio is approx 52kbps
  (12.2kbps AMR 40kbps H263)
• 3G Mobile phones are generally QCIF size (176
  144 pixels)

15
Performance Issues

• Propagation Delay
• Time required to travel end to end across the
  network
• Transport Delay
• Time required to traverse network equipment
• Packetization Delay
• Time to digitize, build frames and undo at
  destination
• Jitter Delay
• Fixed delay by receiver to hold 1 or more packets
  to damp variations in arrival times
• Packet Loss
• Packet size impacts sound quality

16
Jitter Delay

• Calculated on inter-arrival time of successive
  packets
• Average inter-arrival time
• Standard deviation
• Goal inter-arrival time inter-arrival time on
  emitted packets
• 3 phenomena effecting jitter
• Packet loss (threshold 5)
• Silence suppression
• Out of sequence packets
• Can be configured on most VoIP equipment

17
Packet Fragmentation

• Audio RTP packets
• Not generally fragmented since packet size is
  less than MTU
• Video RTP packets
• A large frame is fragmented into a series of
  packets for transmission over network
• I-Frame fragmentation
• Receiver must receive all fragments to properly
  reconstruct frame

18
Packet Loss

• Audio
• Packet Loss Concealment (PLC)
• Mask effect of lost or discarded packets
• Replay previous packet or use previous packets to
  estimate missing data
• Key method for improving voice quality
• Packet Loss Recovery (PLR)
• Packet Redundancy
• Increased bandwidth
• Video
• I-Frame
• If a fragment is lost, subsequent P-Frames will
  not be sufficient to reconstruct image at
  receiver
• Video conversion tools available to generate
  files more suitable for real-time transmission

19
G.107 to MOS mapping
20
Codec Bandwidth and Voice Quality Comparison
Codec Payload Bit Rate Voice Quality
G.711 64 Kbps Excellent (MOS 4.2)
G.723 6.4 Kbps / 5.3 Kbps Good (MOS 3.9) Fair (MOS 3.7)
G.729 8 Kbps Good (MOS 4.0)
G.726 or G.721 16/24/32/40 Kbps 2/3.2/4/4.2
iLBC 13.33/15.2 kbps Good (MOS 4.0)
AMR-WB 6-36 kbps Good (MOS near 4.0)
21
Network Issues?
22
Network Issues Now What

• Determine the source of delay
• Codecs?
• Too many hops?
• Not enough bandwidth?
• Define means to reduce delay
• Provision smaller packet sizes
• Reduce hop count
• Change routing protocols used
• Keep monitoring
• Find problems first
• Objectively identify issues

23
IP Header
24
Traffic Shaping

• DiffServ
• RSVP
• MPLS

25
Conclusion

• Reliability
• Can calls be made when needed?
• Will call setup time match current environment?
• Will calls be disconnected?
• Quality
• Is the voice quality of the calls the same?
• Can the users tell the difference?
• Cost
• What are the cost benefits of VoIP?
• What equipment will be needed?

26
Wrap-up
Q A / Quiz
27
Frame Sizes
Format Dimension (H x W, pixels) gt1 bits/pixel
Sub-QCIF (SQCIF) 128 x 96
Quarter-CIF (QCIF) 176 x 144
CIF (Common Intermediate Format) 352 x 288
4CIF (4 x CIF) 704 x 576
16CIF (16 x CIF) 1408 x 1152