earticle

영어

This paper describes an efficient audio flow control method in the point of quantitative
performance using TDM(Tree-structural Distributed-terminal Mixing), compared to existing
P2P(Peer To Peer) method. In comparison with existing P2P method, using central mixing
and distributed terminal mixing method, we achieved advance at the point of global network
usage and each terminal's CPU load, and additionally we expect more session, more terminal
can be served by same amount of network bandwidth and computers. By using P2P method in
audio communication, speaker and listener must connect to each other. So it has the critical
defect that as the participants grows more and more, the network bandwidth usage, each
terminal's CPU load will grows rapidly. So the number of participants in same session will be
extremely restricted. In comparison with P2P method, the central mixing method has the
great advantage at the points of network usage and terminals CPU load. Regardless of the
number of speakers and listeners, all the participants can speak and listen with all other
participants by using just one stream's amount of data size and CPU load. But all the network
usages and CPU loads of "Audio decompression->Buffering->Mixing->Audio Compression"
are concentrated on central server. So the number of sessions and terminals can be
participated in one server will be highly restricted. This study solves the problems of server's
CPU load and network load by using TDM. The TDM is the method to distribute the overload
for gathering, mixing and distributing the audio data according to each terminal’s up/down
bandwidth and data processing capability using tree-structure. This method enables many
terminals participate in the internet-broadcasting by distributing the overload of server to all
the terminals.