Voice Protocols: At the Core

  1. RTP (CODEC itself converts to voice to data, RTP is the actual voice payload. Uses UDP with timestamps and sequence numbers)
  2. RTCP (Statistical protocol – How things are doing on voice call?)

Behind the scenes with signaling protocols..

  1. H.323 – Designed long ago to do voice + video over ISDN networks. Old protocol but mature 🙂 (Peer to Peer) Default on Cisco routers.
  2. SIP – Evolution of H.323. The goal is to setup calls, setup video etc… do what I do approach. Lightweight and customisable. Easy! (Peer to Peer)
  3. MGCP – Client to Server architecture. All gateways report to one central gateway. Scalable. Drawback… if gateway loses connection to central gateway, it loses all intelligence. ( SRST can mitigate these circumstances)

Call setup and teardown, call maintenance etc..

Voice Call Processing Stages

  1. Sampling
  2. Quantisation
  3. Encoding – G711a or ulaw/G729 – CODEC
  4. Codec Compression
  5. VoIP Encapsulation – Gateway realises it has to send over data network therefore add RTP/UDP headers.
  6. Network Transport – Packet transported
  7. VoIP Decapsulation
  8. Decoding
  9. Modulation – Convert from binary numbers into sounds

With a VoIP handset, DSPs are present on the phone itself. The router can offload tasks onto the DSPs of the phone itself.

Flavours and Features of RTP

RTP: Delivers video and audio streams over data networks

cRTP: Compresses the IP/UDP/RTP headers on low speed connections. Not really compression… it uses the gutting a fish analogy.. only good stuff left over! It strips out content from the header that is not required. Requires processing power!

SRTP: Provides encryption, authentication and integrity services

VAD: Provides bandwidth savings by eliminating silence. (On average 35% bandwidth savings)