The task these days is to do a simulation in opnet.
The aim is to see weather the fixed-length cell (ethernet frame) or the variable-length cell(our EPF) have a better end-to-end delay and jitter.
I modeled a node named "src" to generate packets as poisson distribution. And with in this node, have a processor "process", when doing the VLC case, it directly forward the packets generate to the trasmitter, while when doing the FLC case, it split the packets into the fixed cell size.
And another node called "process" simulate the header process delay and queuing delay.
Finally a "sink" node destroy the generated packet and calculate the statistical variables.
Here I have some questions still confuse me:
- If our EPF is a fixed-length frame and have a length as the maximum ethernet frame, how should a packet be splitted?
- Is there any access rate to the "process" node? As the process node contains a queue and the queue have a certain process rate. I think it must have a access rate to put the packet into the queue. But now I didn't do it anyway.
- If we measure the End-to-End delay, it must related to link rate. That is the data transfer rate. Should we take this into account?
Now I have modeled this, and send the model to the boss, and waiting for his response.
And now let me think this later.