Improving efficiency of IP alias resolution based on offsets between IP addresses

Abstract

In order to get a router-level topology in Internet, IP address alias resolution techniques allow to identify IP addresses that belong to the same router. There are several proposals to make this identification, some based on active measurements and others based on inference studies. The former provides more accuracy and completeness, however efficiency is very low because of the high number of probes needed. These methods probe IP addresses in pairs. With thousands or even more IP addresses to check for aliases, the number of tests gets too high. In order to reduce the number of probes, we propose to select the pairs of IP addresses to test for aliasing using information available a priori. This selection will be based on the offset (numerical distance) between the IP addresses to test. We will show that we can improve efficiency of active alias identification with almost no loss on completeness and without generating probing traffic. The technique is also adaptable to a distributed measurement scenario.