Here is what I am thinking:
Currently my poisson model takes the avg. interarrival time of packet
as lamda.
With bursty traffic I plan on generating burst at nodes with
interarrival time lamda/BURST_PARAM.
Then for each burst I generate BURST_PARAM packets with interarrival
time lamda*BURST_PARAM.
Is this an accepatable implementation?
Would appreciate comments and pointers.
Affan
>With bursty traffic I plan on generating burst at nodes with
>interarrival time lamda/BURST_PARAM.
>Then for each burst I generate BURST_PARAM packets with interarrival
>time lamda*BURST_PARAM.
There's lots of literature on burst Poisson models (just Google).
The larger issue is whether it is useful. Data traffic is (for the most
part) not poisson in its arrivals (in any way shape or form). Just go
read "The Failure of Poisson Modeling" by Paxson et al.
Craig