Managing the Latency Budget

We have to consider the need to deliver end-to-end latency (delay and delay variability
guarantees) to meet particular quality of service requirements specified in our users’
service level agreement. To do this, we need to take into account all the factors contributing
to the latency budget. In a wireless IP network, these factors include the access
latency introduced by the radio physical layer (availability of radio resources) and the
access latency introduced by the network (availability of network resources). Access
delay and access delay variability may also be caused by poor protocol performance—
the inability to allocate available radio or network resources.
The criteria for measuring access latency includes session setup success/failure and
session setup delay. If we need to make multiple attempts to establish a session, we introduce
delay and delay variability. We also absorb radio and network resources, including
signaling bandwidth resource. Once a session is established, we then have the problem
of network latency. Network congestion may mean we cannot deliver sufficient network
bandwidth to deliver continuous session support, which means the session fails.
In a wireless IP network, session failure can also be caused, as we said earlier, by a
discontinuity in the availability of radio resources. The user moves out of radio coverage
or the call drops because of a problem with hard or soft handover control. Session
continuity is therefore dependent on the continuous consistent availability of radio
and network resources.
We also need to consider application latency. If the user-to-user or device-to-device
exchange includes the need to access a server, for example, then delay and delay variability
may be determined by a lack of server bandwidth or a failure to meet the
server’s admission permission criteria. 342