Ad-Hoc Research at PI4

Traditionally, research in Mobile Ad-Hoc Networks (MANET) focusses on network layer. A reason for that could be the that single-hop wireless communication already exists for quite some time and that the main challenges in a self-organizing network were identified as the capability of communication beyond the line-of-sight of the particular radio system by using radio neighbors as a relay.

Addressing this problem, dozens of algorithms were proposed, most of them based on randomly-moved nodes and quantitaviley evaluated by means of discrete event simulation. Usually, MANET algorithms focus on solving a special class of these networks very good (e.g., very mobile networks or energy-constrained nodes) trading in performance in other properties. In this sense, we at PI4 have identified Position-Based MANET routing algorithms as an interesting field for highly-mobile Ad-Hoc Networks. These algorithms make use of each node's knowledge about its own position (as perceived by means of a positioning-system like GPS). For networks were this kind of service is available, we have proposed and evaluated novel algorithms for general MANETs to both route a packet to an arbitrary (unicast) destination and to find out the needed geographic position of the destination node. In addition a focus of research are scalable multicast routing algorithms like SPBM, again relying on the greographic position of a node.

While the random-movement case is generally believed to be a worse-case scenario for any MANET no real network exists following the patterns used for simulation. Eventually, we were confronted with the problem of Ad-Hoc Routing in a very special but also very real kind of network: The challenge of routing data packets in a Vehicle-Based Ad-Hoc Network (or VANET). Working in this field provided us with a very clear view of what the node capabilites would be lilke and what movement pattern the nodes will most likely follow.

Part of the VANET-project FleetNet was to actually build a communication system for a vehicular Ad-Hoc Network. While this proved to be a lot of work, it also hinted to some serious flaws in simulation-based MANET research, basically due to too much abstraction in simulation models, e.g., in the model of the radio channel. In addition to proving the concept of position-based routing for a vehicular communications platform, we have also gained valuable insights in building such a system and measurement results.

The most recent branch of Ad-Hoc Networking research activities are sensor networks...

For publications, please have a look here.