An Improved Link Weight Change Algorithm for Traffic Matrix Estimation

Traffic matrix has the benefit of offering an over view of the whole network and computes the desired OD flow volume efficiently compared to direct measurement. Former algorithms of traffic matrix estimation mainly focus on inference techniques but lack the update information at different time intervals. Link weight change method could reflect the dynamic updates thus becomes a future direction. The original algorithm of Nucci´s consider the stability issue of a network first, so put the time efficiency aside. Our improved algorithm aim at rapid information gain, then seek the completeness to reach full rank. Result shows our improved algorithm accelerates the speed to full rank and need less candidate snapshots than the original algorithm.