Protocol Normalization Using Attribute Grammars

Protocol parsing is an essential step in several networkingrelated tasks. For instance, parsing network traffic is an essential step for Intrusion Prevention Systems (IPSs). The task of developing parsers for protocols is challenging because network protocols often have features that cannot be expressed in a context-free grammar. We address the problem of parsing protocols by using attribute grammars (AGs), which allow us to factor features that are not context-free and treat them as attributes. We investigate this approach in the context of protocol normalization, which is an essential task in IPSs. Normalizers generated using systematic techniques, such as ours, are more robust and resilient to attacks. Our experience is that such normalizers incur an acceptable level of overhead (approximately 15% in the worst case) and are straightforward to implement.