A Novel Approximate Algorithm for Admission Control

Abstract

Admission control is the problem of deciding for a given set of requests which of them to accept and which to reject, with the goal of maximizing the profit obtained from the accepted requests. The problem is considered in a scenario with advance reservations where multiple resources exist and users can specify several resource request alternatives. Each alternative is associated with a resource capacity requirement for a time interval on one of the multiple resources and a utility. We give a novel (1 + α)-approximation admission control algorithm with respect to the maximal utility and derive the approximation ratio for different request scenarios. We also design non-guaranteed greedy heuristics. We compare the performance of our approximation algorithm and greedy heuristics in aspect of utility optimality and timing in finding solutions. Simulation results show that on average our approximation algorithm appears to offer the best trade-off between quality of solution and computation cost. And our (1 + α)-approximation algorithm shows its intrinsic stability in performance for different utility functions.