Concepedia

Packet size and DTN transport service: Evaluation on a DTN Testbed

Nikolaos Bezirgiannidis, Vassilis Tsaoussidis

2010 · 48 citations · 9 references

TL;DR

Bundle Protocol splits data into bundles of varying size, which are then encapsulated into LTP blocks and further fragmented into segments. The study evaluates DTN transport layer performance for file transfer from a Mars relay satellite to Earth, focusing on how bundle, block, and segment sizes affect delivery time, overhead, and memory usage. Experiments were conducted on a DTN Testbed built to emulate deep‑space communications, assessing how varying bundle, block, and segment sizes influence transmission. The study found that smaller bundles and LTP blocks release memory more quickly, that segments should approach the underlying MTU limit, and that file delivery time in a steady‑bit‑error channel is largely insensitive to packet size when these dependencies are ignored.

Abstract

We present a study of DTN transport layer service, evaluating its performance during a file transmission from a Mars relay satellite towards Earth. The experiments are being held in a DTN Testbed designed and implemented for emulating and evaluating deep-space communications. Bundle Protocol truncates application data into bundles of different size and Licklider Transmission Protocol is used as the convergence layer protocol, encapsulating bundles into blocks and fragmenting blocks into segments. Bundle size, LTP block size and segment size are under consideration to achieve better transmission in terms of delivery time, overhead and memory occupancy at the sending node. Our most interesting conclusion is that memory is released faster when using small bundles and LTP blocks. We also conclude that segments should be close to the maximum boundary of underlying layer's MTU, and that ignoring these dependencies, file delivery time in a channel of steady bit error rate has little dependence from packet sizes.