الفهرس 
NEnlORK ARCHITECTIJRES & TOPOLCXlYOnly 14 pages are availabe for public view
 |  | from | 268 |  |  |
| | | from | 268 | | |
Abstract
A number of significant technical problems exist in sending voice
through a packet network . One of the most significant involves the
reconstruction of a continuous stream of voice from a set of packets
sent through the network.
The core of this thesis is • how to design voice reconstruction
system that reproduces acceptable quality speech from those packets
that • arrive with long end-to-end delay • varying transit delay which
depends on the packet network’s structure • and in some cases arrive
out of order if those packets are routed independently . So • the received packets ,
;.j·O’” Different reconstructi01 schemes are studied for the packet voice
receiver, A new reconstruction scheme is proposed and developed based
on integration of two studied schemes to avoid the disadvantages of
both of them ,
Simulation models packetized voice
transmission system
wer developed for the
to evaluate three types
of voice packet
reconstruction schemes , and to obtain the optimal packet length
which keeps overall paCkjt transmission delay
probability less than a certain permissible value Comparison among
and packet loss
three strategies is also stated with respect to
the mean packet
total delay and the statistical fluctuations of silence intervals ,
Key !fords : packet vo~ce communication ; voice reconstruction
packet voice receiver ; reassembly algorithms ; optimal packet
length ; dynamic priorities; service strategies
Buffering smoothing schemes
I
synchronization