We study the problem of allocating network capacity through periodic auctions, motivated primarily by a service overlay architecture. Clients bid for capacity to connect to remote servers, under the condition that resources allocated in an auction are reserved for the entire duration of the connection, not subject to future contention. We study the optimal revenue for the operator under these constraints over space and time. For the single resource case, the problem is formulated as a Markov Decision Process (MDP); we develop a receding horizon approximation to the optimal policy that is computationally tractable, and can be framed in terms of network utility maximization. This opens the door for distributed implementations in the general network case, which are validated through simulations.
Copyright. All Rights Reserved.