الفهرس 
IntroductionOnly 14 pages are availabe for public view
 |  | from | 16 |  |  |
| | | from | 16 | | |
Abstract
A TM as a platform refers to a high-bandwidth. low delay switching and multiplexing technology that can be used in both public and private network applications. The highly efficient architecture of the ATM can handle almost any type of traffic. The main design goal of A TM is its support of a wide range of data. video. and voice applications in the same public network at very high speed The most famous teleservices are HDTV (High Definition TV). video conferencing. high-speed data transfer. videophone. video library. home education and video on demand. Each of these services will generate other requirements for the BISDN (Broadband Integrated Service Digital Network) which takes into account all possible existing and future services. This large span of requirements introduces the need for one universal network. which is flexible enough to provide all of these services in the same way
The guarantee of the QoS (Quality of Service) is one of the one of the biggest advantages of ATI\1 QoS is essential for applications such as video or audio. where all packets must arrive in a timely fashion in order for presentation to be coherent
ATM’s speed far exceeds that many other network technologies. In its current state of development. it runs at I 55Mbps. although theoretically. there is no limit. In fact. plans have been made to increase the top implemented speed specification from 622 Mbps to 2.488 Gbps
Because of the statistical nature of packet switching. buffers have to be provided in the system to store the packets whose request can not be immediatel\ serviced. Possible locations for the buffers are at the input ports or the output ports. In a switching element. mainly 3 different buffering strategies are possible (input. output or central)
Input queue element suffers from head of the line blocking (HOL) which reduce its performance compared to the output and central queue. As a result of poor performance of input queue. compared to output and central queue. we want 10 improve the performance and quality of service (QoS) for that model. There are some advantages of using input queue instead of output or central queue.
For output and central queue. to ensure that no cell is lost in the switching transfer medium before it arrives at the output queue. the cell transfer must performed at N times the speed of the inlet (where N is number of inlets). The system must be able 10 write N cells in the queue during one cell time. In case of input queue each
III
buffer (queue) may be served. So although input queue does not reach the performance level of output and central model approach. it will take the advantage of reducing the input servicing speed to IIN (i.e system must be able to write only one cell at one cell time). This is the reason. which we want to improve its performance.
Moreover. analytical model is a good approximation of reality. but it is. for simplicity, only valid within a certain assumption (e.g. N --.00 ) The simulation models are applicable under more general conditions. and give more information (e.g cell maximum waiting time. CLR. ... etc). compared to analytical model which gives a limited number of information (e.g. average queue length). The only valid analytical model is for output queue model and input queue model using FIFO selection policv But in the case of central queue analytical model evaluation leads to a too complex queuing problem to be solved in a closed form So we are simulating Output. Input (with difference cell selection policies) and Central models to get more information about their characteristics The aim of this thesis is to study the performance evaluation (mean wait time ... etc), error handling (cell loss ratio .. etc). model servicing complexity, and quality of service (jitter delay, ... etc) for the three models with different cell selection policies Moreover. as an alternative to the FIFO or Random Selection policies. to serve the cells in the input queues. three cell selection policies are proposed for input queue. which are Queue Filling Level selection. Queue Filling Level selection with Cell Delay Control. Window Service selection