A Performance Study of ATM Multicast Switch with Different Traffics
The demand of multicast service In ATM network such as video teleconferencing, broadband telephony and large file transfer makes the multicast switch very important. The multicast switch is needed to multicast an input cell to a number of selected output or broadcast it to all outputs. Since the...
Saved in:
| Main Author: | |
|---|---|
| Format: | Thesis |
| Language: | English English |
| Published: |
1999
|
| Subjects: | |
| Online Access: | http://psasir.upm.edu.my/id/eprint/9634/1/FSKTM_1999_8_IR.pdf |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | The demand of multicast service In ATM network such as video
teleconferencing, broadband telephony and large file transfer makes the multicast
switch very important. The multicast switch is needed to multicast an input cell to
a number of selected output or broadcast it to all outputs. Since the cell loss and
delay will decrease the performance of an ATM multicast switch, it should be
designed so that the degradation is minimised. Hence, to improve the
performance, which is characterised by cell loss probability and mean cell delay,
this thesis proposes a new architecture of a multicast switch.
In this thesis, the traffic is classified into three categories with different
requirements: real-time, near real-time and non real-time. Since the real-time cell
is very sensitive to delay, it is given the first priority to be served. Near real-time
cell can tolerate a small delay while the non real-time cell is less sensitive to delay. Hence, the arriving cells of those traffic can be buffered to wait their tum
to be transmitted. Two buffering schemes which respectively realise the buffer of
near real-time and non real-time cell are implemented: First In First Out with
Ordinary Blocking (FIFO-BL) and First In First Out (FIFO). To achieve a better
performance, Largest Queue First Serve (LQFS) as a scheduling algorithm is
implemented.
The proposed architecture with the above buffering schemes and traffic
classes is studied by developing appropriate simulation models. Subsequently, the
effect of arrival rate, immediate rate, requested copy number, buffer size and
RealTC rate on the switch performance are studied quantitatively and
qualitatively. It is concluded that the proposed architecture can support the future
different multimedia traffic types, where cell loss probability and delay
requirement will be the main factors. |
|---|