My problem is how to implement the time where a link is not used. Example. Let us assume a two-state markov chain (Gilbert-Elliott model). Node 1 sends a packet to node 2 and this packet is lost so that the model goes to the state "last packet was lost". Now the congestion control mechanism at node 1 drops this packet (due to deadline or other parameter). After a time where the link 12 is not used, I guess that the model should come back to the state "last packet was right".
1. Is this right? Should this time be equal to a packet transmission time? Maybe it should be related to the coherence time of the channel but I guess that these models are somehow an abstraction of the physical layer.
2. Moreover, I have never read that the packet size is taken into account. Do these models take into account that all the packets have the same size?
Hello Jesus.
This might help you.
http://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.90.1411&rep=rep1&type=pdf
Hello Joaquim,
Thanks for you answer. The paper is very intereseting.
I realized that I can simulate the packet losses taking into account a continuous transmission from the packet loss model point of view. And a discontinuous transmission from the communication side, as it is actually. I think this should work.
- Badges
- Science topic