Algorithm design for Bandwidth Availability

Question

I have a graph of N nodes. I have the bandwidth consumption of all the links. A link with least bandwidth available in a path from Node s to Node t is called the bottleneck of the path. To find the bandwidth availability between Node s and Node t, I am running a DFS to find N number of paths between the two nodes and then I am finding bottleneck of each path. I am then taking an average of these bottlenecks to find an average bottleneck. Can I use this as a single number to denote the bandwidth availability between Node s and Node t? What are the pros and cons? Please suggest me a right place to ask if this is not the right place.

Answer 1

It sounds like what you're looking for is Network Flow Analysis, and particularly Max flow, Min Cut.

Your current implementation ignores the fact that you might be able to send data along multiple paths at once.

One last note - you can use Djikstra's alogirithm to find the path with the highest bottleneck.