Detectability of Deep Packet Inspection in Common Provider/Consumer Relations

摘要

Payload examination using Deep Packet Inspection (DPI) offers (infrastructure) providers a whole new range of use cases, many of them with a potential to eavesdrop on non-public communication. Current research is almost exclusively concerned with raising this capabilities on a technological level. Critical voices about DPI's impact on the Internet with regard to privacy, net neutrality, and its other implications are raised, however often not within research communities but rather by politically interested groups. In fact, no definite method allowing detection of DPI is known. In this paper we present five different approaches targeting this problem. While starting points for DPI detection are given, including leakage of internal data or software errors, not all of of the presented approaches can be simulated or verified at all and none so far has been tested in real world settings.