Levaldi: An improved network distance prediction algorithm based on network coordinate system

The Vivaldi Algorithm is a simple, adaptive and distributed algorithm for computing and predicting the hosts´ coordinates and their network distance in a network coordinate system. However, coordinate calculation of the convergence of the Vivaldi algorithm is rather slow with high noise ratio samples. A new efficient algorithm (Levaldi Algorithm) is proposed, which uses a method in which we take the adapt-step problem into consideration and modify the size of the step in Vivaldi Algorithm that reduces the affection on other nodes by a certain node. Experiments indicate that the Levaldi Algorithm significantly reduces the time of convergence in coordinate prediction. Its accuracy and the speed of the convergence increase compared to the Vivaldi Algorithm.