Compression of Endpoint Identifiers in Delay Tolerant Networking

Delay and Disruption Tolerant Networking (DTN) was developed to deliver network communications to so-called "challenged environments." These include space, military, and other networks that can be described as having extremely long link delay and frequent disconnections. The DTN paradigm implements a store-and-forward network of nodes to overcome these limited environments as well as delivering "bundles" of data instead of packets. The bundles nominally contain enough data to constitute an entire atomic unit of communication. DTN introduces the Endpoint Identifier (EID) to identify bundle Agents or groups. The EID can imply naming, addressing, routing and network topology, but these features and flexibility come at the cost of verbosity and a per-packet overhead introduced by large and descriptive EIDs.In this document, we apply lossless text compression to EIDs using Zlib's DEFLATE algorithm. We develop a novel method for generating a large sample of verbose EIDs based upon Apache access logs, allowing testing over a larger, more varied, and more realistic data set than would be possible with the current DTN testing networks. Analysis of the processing overhead and compression ratio lead us to the conclusion that Zlib reduces the overhead of EIDs substantially. By compressing the dictionary, more featureful EIDs can be used without increasing overhead in the form of larger bundle dictionaries due to syntactical verbosity.