how to Calculating throughput for muti-hop routing in wireless network.

Ex. source (s) to destination (d)

path is: s -r2-r4-d (r2 and r4 Relay nodes)

[Walter Roberson]

Should the calculation be statistical or by simulation?

If you were wanting to tune wait time and frequency allocation algorithms then you should use simulation: otherwise you miss the opportunities to see how often you can arrange for "everything goes right", and you miss accurate modelling of congestion.

[K Venkatesh]

simulation

[K Venkatesh]

Thank you ,Roberson

give me an example with some values.

[Walter Roberson]

I could invent some values but you would be better off researching the wireless standards you are interested in.

[K Venkatesh]

Thank you ,Roberson

Ex. source (s) to destination (d)

path is: s -r2-r4-d (r2 and r4 Relay nodes)

my channel bandwidth 10 mbps can pass only average 12000 frames per minute with each frame carrying an average 10000 bits. What is throughput of source to destination.

actually network throughput calculate =(12000*10000)/60 in the same way how to calculate with ray nodes in network from source to destination. Here relay nodes get data from source then sent to next hop upto destination.

[Walter Roberson]

That is statistical analysis not simulation.

You need to decide on a routing protocol. Do you find all possible paths from source to destination and send data on all of them, knowing that the data will arrive out of order? Or do you find only the shortest path as if the flow was the only one in existence? Do you exchange dynamic messages about how busy a node is so that routing can try to adapt to maximize overall flow?

[K Venkatesh]

Thank you ,Roberson

Every thing done as you mention above but calculating throutghput, i am facing problem.

[Walter Roberson]

Sorry, that is very vague. There are thousands of routing protocols and millions of implementations. I cannot guess what lines have to be changed in each of your functions (which I have not seen) in order to instrument throughput.

At the level you have given so far, all we can do is talk about the general definition of throughput: throughput is the number of bits of payload that can be delivered per unit time. However, in a multi-node system, you need to decide whether you want to measure maximum throughput between two particular points, or if you want to measure some kind of average throughput over multiple conversations, or if you want to measure some kind of aggregate throughput, total amount of payload deliverable over all conversations over the entire network.

For example, you could have two users, A and B, both talking to server S, and it could be the case that communications A<->S and B<->S are both being limited by timing factors, but that S is able to talk to both A and B in some overlapped manner that does not violate the timing of either A or B, so total throughput could be A<->S throughput plus B<->S throughput. Or perhaps there is some effect on the conversations, so you might be able to get some factor alpha*(A+B) instead of max(A,B) as your throughput measurement.

You need to decide first what you might want to measure before you can figure out what code changes you need to make.