Aggregation & analysis of IPv6 prefixes at Internet-scale

Abstract

Due to its crucial importance in today's world, researchers are interested in understanding large-scale real-world usage of the internet. Concrete motivations vary, but are commonly related to security research. The internet community is in the process of exchanging the base protocol used for internet communication, introducing IPv6. This more modern protocol significantly expands address space from 2^32 to 2^128 possible addresses. While there are many benefits to this modernisation, researchers have been relying on exhaustive internet measurements for decades. Existing methods cannot directly cope with the considerable expansion of search space, and the community is interested in new ideas that restore the ability to observe the entire internet. Current solutions are still characterised by trade-offs and limitations. Structure-aware probing is an idea that addresses some of these.Instead of exhaustively and linearly inspecting every possible address, this method recursively finds interesting spaces by repeatedly splitting the measurement space, akin to binary search. A focus on high-level prefix structures allows experiments to target wide areas of the internet, while not requiring exorbitant probing rates due to re-use of previously-discovered information. The thesis proposes an algorithm that applies this method by combining a variety of existing concepts. Evaluation against a benchmark linear measurement, ground truth data, and behavioural metrics shows that structure-aware probing is a promising idea. While there still is work to be done until this particular system can reliably scan the entire internet, results are in general promising, affirming successes from existing work with related methods.