RT_Vivaldi: A Symmetric & Equality RTT Prediction Algorithm in Networks

Predicting network round-trip times (RTT) between nodes is difficult because of the asymmetric and unequal network features. The phenomenon of Triangle Inequality Violation (TIV) can seriously affect the results in prediction. The traditional Vivaldi algorithm has a low accuracy and an unsteady convergence rate. In this paper, a novel Restrained TIV Vivaldi (RT_Vivaldi) algorithm is proposed. Firstly, the TIV is analyzed to set a triangle detection and suppression coefficient. The coefficient is defined to correct the asymmetric and unequal influence on the TIV. Multiple reference neighbor nodes are set based on the distances between them and then dynamically update the parameters to eliminate the effects of the links with the various lengths. Finally, simulations have proven that the proposed RT_Vivaldi algorithm has a higher accuracy and a greater stability than those traditional algorithms.