Asynchronous auction for distributed nonlinear Resource Allocation

Nonlinear Resource Allocation Problems are concerned with the optimal allocation of N continuous resources to M missions/tasks with nonlinear utilities. Numerous problems from diverse fields such as search theory, statistics, finance, economics, logistics, sensor and wireless networks fit this formulation. Several centralized/synchronous algorithms are available for solving this problem including RAP Auction [1] which was proposed by the the authors. RAP Auction finds a near optimal solutions with pseudo-polynomial complexity for generalized class of monotonic convex cost functions including non-differentiable and/or non-strictly convex functions. In this paper, we present and discuss asynchronous version of RAP Auction. This algorithm supports the computation framework in which nodes behave as autonomous agents making local decisions without waiting for each other.