Defining Latency, an IP telephony term

Latency is the time from mouth to ear. It is
the time it takes for a person’s voice to be sampled, packetized, sent over the IP
network, de-packetized and replayed to the other person.
Distance on the WAN circuit, by itself, can cause delay, as can lower-speed WAN
circuits. If latency is too high, it interrupts the natural conversation flow and can
cause the two parties to confuse latency for pauses in speech. Latency must not
exceed 100 milliseconds (ms) one way for toll-quality voice and must not exceed
150 ms one way for acceptable quality voice. At 150 ms, delays are noticeable, but
callers can still carry on a conversation.
Users hear jitter as degraded voice quality. Jitter is variation in latency over the LAN
and WAN, as the IP telephony packets arrive in uneven patterns at their destination.
Jitter has many sources, including network congestion, queuing methods used in
routers and switches, and routing options such as MPLS or frame relay used by carriers.

Network Requirements for Toll-Quality Voice
The fundamental requirement for achieving toll-quality voice is to deploy IP telephony
over a properly architected network infrastructure. The LAN/WAN infrastructure must
deliver sufficient throughput and meet latency, jitter and packet loss requirements.
Deliver sufficient throughput: The amount of bandwidth required for voice depends
on the number of simultaneous calls, the voice encoding scheme used in the IP
handset or softphone, and the signaling overhead.
The International Telecommunications Union (ITU) G.711 codec is commonly used
in LAN deployments where LAN bandwidth is plentiful. With G.711 and RTP
header compression, each call requires 82 Kbps.
ITU G.729 is commonly used in a WAN environment because it uses substantially
less bandwidth. With G.729 and no header compression, each call requires 26 Kbps.
With ADPCM and no RTP header compression, each call requires 52 Kbps

Packet loss requirements: Packet loss results in a metallic sound or dropouts in the
conversation. Packet loss is caused by congestion, poor line quality and geographical
distance. Since IP telephony is a real-time audio service that uses the Real Time
Protocol (RTP) running over the User Datagram Protocol (UDP), there’s no way to
recover lost packets. If even 1 or 2 percent of IP telephony packets drop, voice
quality degrades.
ShoreGear Voice Switches include a lost packet concealment capability that
reduces the impact of packet loss. When there is no voice sample to be played, the
last sample is replayed to a receiving party at a reduced level. This is repeated until
a nominal level is reached, effectively reducing the clicking and popping associated
with low levels of packet loss

source: ShoreTel White Paper:
Is Your Network Ready for IP Telephony?


Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

%d bloggers like this: