Performance of Disruption-Tolerant Network Mechanisms Applied to Airborne Networks

Bishwaroop Ganguly, Vijay Subramanian, Shiv Kalyanaraman, K.K. Ramakrishnan
2007 MILCOM 2007 - IEEE Military Communications Conference  
Recently, there has been a great deal of interest in the development of viable airborne network (AN) architectures for both defense and commercial applications. The design of an airborne network differs from traditional (e.g. Internet) networks because the dynamics of airborne networks give rise to a number of challenging conditions. In this work, we focus on link disruption caused by movement and changing orientation of network nodes/platforms. This includes wing and body blockage, terrain
more » ... kage, range limitations, atmospheric turbulence and weather effects. The focus of this paper is mitigation of disruption of links in the AN. Previous work has shown that a combination of transport layer and link layer approaches can successfully address link disruption in wireless network context. In this work, we leverage a combination of two technologies, a transport protocol LT-TCP and a hybrid link-layer FEC/ARQ protocol, LL-HARQ. LT-TCP is an enhancement to the transport layer protocol TCP, that provides loss tolerance and higher goodput on unreliable links. LL-HARQ is a link-layer protocol that provides improved reliability on a link-by-link basis. Both protocols operate under a common framework using loss estimation, proactive/reactive FEC and retransmission. These approaches have previously shown good performance on links with specified packet error rates, in this work we apply these approaches to links that suffer disruptions. We investigate the performance of these approaches, properly parameterized, on realistic link data gathered from AN flight experiments. We compare this performance to that of traditional link and transport approaches, for airborne network scenarios. Our results show that the technologies greatly enhance network goodput for airborne networks where disruptions occur frequently when compared to traditional networks.
doi:10.1109/milcom.2007.4454771 fatcat:xlkh7a66s5fkxjrj3oapimdvly