On the Applicability of Knowledge Based NAT-Traversal for Home Networks

The presence of Network Address Translation (NAT) is a hindrance when accessing services within home networks, because NAT breaks the end-to-end connectivity model of the Internet protocol suite. Communication across NATs is only possible if it is initiated from a host belonging to the internal network. Thus, services expecting a connection established from the outside fail in most situations. Existing approaches for NAT-Traversal do not cover the full range of NAT-Traversal methods and fail in certain situations, or deliver sub optimal results in others. Part of the problem of existing approaches is that they do not differentiate between different types of applications. We argue that the classification of applications into four service categories helps to determine the best matching NAT-Traversal technique. An extensive field test enables us to acquire knowledge about the success rates of promising NAT-Traversal techniques. These results will help us to develop a knowledge driven NAT-Traversal framework making its choice based on an understanding of NAT behavior, NAT-Traversal options and the service category of the application.