Route optimization based on the detection of Triangle Inequality Violations

During the last decade, new services networks and distributed applications have emerged. These systems are flexible insofar as they can choose their ways of communication among so much of others. However, this choice of routing is based on a large number of measurements of times (Round Time Trip (RTT)) which are sources of overload in the network. Network Coordinate Systems (NCS) allow to reduce measurements overhead by mitigating direct measurements. However, NCS encounter inaccuracies with respect to distance prediction, when the measured distances violate the principle of the triangular inequality (TIV-Triangle Inequality Violation). Firstly, we propose a new metric, called “RPMO”, which is based on the Ratio of Prediction and the Average Oscillations of the estimated distances, to detect the potential TIVs. The obtained results show that the “RPMO” metric gives better performance compared to metrics presented in former work. Secondly, we propose to use the existence of TIVs to optimize the routing in Overlay Network. To achieve this goal, we present a new approach that enables to detect the best shortened paths offered by the existence of potential TIVs.