University of Konstanz
Graduiertenkolleg / PhD Program
Computer and Information Science

Graduation Talks

title

Reliable Message Delivery in Disruption Tolerant Networks

speaker

Arshad Islam, University Konstanz
Konstanz, Germany

date & place

Wednesday, 03.06.2009, 16:15 h
Room C 252

abstract

The presence of an increasing number of pro- grammable mobile devices has caught on the desire to use them in ad-hoc networks and first ad-hoc routing protocols were born, allowing to find an active path to a destination, which can then be used for real-time data exchange. Delay/disruption tolerant networks (DTNs) go one step further and allow messages to be passed even when the destination and/or intermediate nodes are not currently online. For many data exchanges, this additional delay is acceptable; it may even be an advantage, as the possibility to use nodes carrying the data (“data mule” or just “mule”) to a destination can provide higher bandwidth than having the data contin- uously hop from one mobile node to the next. Some of the prominent DTN applications include crisis environ- ments like emergency response in case of a catastrophe, military operations, deep space communication, vehic- ular communication, and non-interactive Internet ac- cess in rural areas. Usefulness of DTNs greatly depends on the routing efficiency of the network. Several proposals for efficient routing mechanisms have been devised. Strategies that do not employ redundancy use computationally inten- sive procedures to determine the path for the message. This path may be calculated at the source and then regularly updated on each hop depending on the net- work topology and congestion. Mechanism that involve redundancy try to exploit multiple paths for replicas to enhance the delivery of message through the net- work. Furthermore, various protocols assume different levels of knowledge available to network nodes, ranging from only the set of nodes currently directly reachable over connectivity history of the node in question and, maybe, other nodes it has seen up to accurate predic- tion of the future (oracles). Obviously, the quality of prediction depends on the accuracy of the oracle. One can safely assume that the higher the accuracy, the less likely it is to actually find such an oracle in realiy. The aim of the purposed thesis is to define and ana- lyze metrics that may be tuned optimally according to the underlying network to attain acceptable message delivery.