An optimization method of large-scale IP traffic matrix estimation

Abstract

This letter studies the large-scale IP traffic matrix (TM) estimation problem and proposes a novel method called the simulated annealing and generalized inference (SAGI). The major challenge of large-scale IP TM estimation is its highly ill-posed nature. We describe the TM estimation into a modified simulated annealing process, and then by using the generalized inference we can easily overcome its ill-posed nature. Simulation results show that SAGI is promising.