RE: [manet] Throughput of a MANET

["John Mullen" <jomullen at nmsu.edu>]

Hi Vinod,

If you calculate the throughput as the number of data packets (or bits
contained in data packets) correctly received at the destination divided
by the duration, you will probably be OK.  However, there are a couple
of caveats.

1.  For an on-demand protocol, starting at time zero would probably, but
if you are using a proactive protocol, you may want to delay the onset
of traffic until the routing tables are established.  If so, your mean
throughput should be divided by 900 less the initial dead time.

2.  In either case, you will want to look at throughput vs. simulation
time.  You may find that the throughput is decreasing due to queueing
delays, etc.  Depending on what it is you are looking for, you may want
to exclude the initial very high throughput from your calculations.
Otherwise, there will be a dependency on the initial conditions that
will vary with offered load and simulation run time.  More critically,
if throughput is still falling at the end of the simulation, you don't
have a stable system and the calculated result would be misleading.

Since you are measuring correctly delivered packets, the impact of
things like contention, collisions, overhead, time for route searches,
etc. will all be included in your measure.  Just make sure that your
statistic actually counts only the routed data packets at the
destination and not other communications, such as RREQs, RREPs,
retransmissions, etc.

However, I usually calculate relative throughput, which is the ratio of
the number successfully delivered packets to the number of packets sent
by the originator.  If packet sizes differ, I calculate the ratio of
bits and sometimes the packet ratio by class.  I feel that calculating
relative throughput at different levels of offered load is more useful
than absolute throughput, though both are important.

John Mullen

-----Original Message-----
From: manet-bounces at ietf.org [mailto:manet-bounces at ietf.org] On Behalf
Of Vinod
Sent: Thursday, September 09, 2004 12:59 PM
To: manet at ietf.org
Subject: [manet] Throughput of a MANET


Hi all,
I was wondering how exactly is the throughput of a
MANET calculated during simulations. For a 900 second
run with n CBR flows(starting at time 0 to 900), the
number of packets correctly received at all the
destinations divided by 900 seconds will give me the throughput. Is that
correct? Will that take into account contention and stuff at the MAC
layer or is there something else i am missing?

Thanks,
Vinod

 


		
_______________________________
Do you Yahoo!?
Shop for Back-to-School deals on Yahoo! Shopping.
http://shopping.yahoo.com/backtoschool

_______________________________________________
manet mailing list
manet at ietf.org
https://www1.ietf.org/mailman/listinfo/manet


_______________________________________________
manet mailing list
manet at ietf.org
https://www1.ietf.org/mailman/listinfo/manet