Efficient Large Scale Network Topology Measurement

Abstract

Studies related to the Internet topology measurement are challenging becauseof the large-scale topology of the network with limited number of vantage points.As topology mapping studies rely on the few vantage points to probe a large numberof destinations, there is considerable redundancy in the collected measurements.Large number of redundant probes during a measurement study typicallycontain significant overlaps, which increases measurement traffic and duration.To decrease the probing overhead, in this thesis, we propose an efficient subnetworkprobing technique by identifying ingress points of the autonomous systemstowards the subnetwork. To achieve this, we first probe announced prefixes of agiven autonomous system until the ingress points of the target prefixes are identified.After identifying the ingress points of autonomous system with respect tothe subnetworks, the probing is conducted based on the vantage point reachabilityof the ingresses. As shown in experimental study, this approach can considerablyreduce the probing overhead while minimizing the loss of information.