I have the problem, which I described below. I wonder if there exists a class of multi-armed bandit approaches that is related to it.
I am working on computer networking optimization.
In the simplest scenario, we model the network as a graph with a circular node topology, similar to that seen in Chord (attached photo). Each node(vertex) can have a maximum number of $X$ active links (tunnels or edge) to other nodes at any given time. Then it can open, maintain, or close links (each operation has a cost associated with it). If there isn't a direct edge, traffic must be routed through neighboring nodes. What is the best link structure(optimal set of edges in the graph connecting nodes) in the underlying graph given the predicted traffic intensity matrix between the nodes?
Note: the optimal link structure should be recalculated on a regular basis to account for history (for example, it is worthwhile to keep a connection between two nodes open even though there is no traffic at the current time because it was generally a busy link in the past and the chance of using this link is high in future).
Estimation: Can multi-armed bandit be useful here?