Message traceback systems dancing with the devil

Abstract

The research community has produced a great deal of work in recent years in the areas of IP, layer 2 and connection-chain traceback. We collectively designate these as message traceback systems which, invariably aim to locate the origin of network data, in spite of any alterations effected to that data (whether legitimately or fraudulently). This thesis provides a unifying definition of spoofing and a classification based on this which aims to encompass all streams of message traceback research. The feasibility of this classification is established through its application to our literature review of the numerous known message traceback systems. We propose two layer 2 (L2) traceback systems, switch-SPIE and COTraSE, which adopt different approaches to logging based L2 traceback for switched ethernet. Whilst message traceback in spite of spoofing is interesting and perhaps more challenging than at first seems, one might say that it is rather academic. Logging of network data is a controversial and unpopular notion and network administrators don't want the added installation and maintenance costs. However, European Parliament Directive 2006/24/EC requires that providers of publicly available electronic communications networks retain data in a form similar to mobile telephony call records, from April 2009 and for periods of up to 2 years. This thesis identifies the relevance of work in all areas of message traceback to the European data retention legislation. In the final part of this thesis we apply our experiences with L2 traceback, together with our definitions and classification of spoofing to discuss the issues that EU data retention implementations should consider. It is possible to 'do logging right' and even safeguard user privacy. However this can only occur if we fully understand the technical challenges, requiring much further work in all areas of logging based, message traceback systems. We have no choice but to dance with the devil.