Wintersemester 2017/2018

Physarum polycephalum is a giant multi-nucleated but unicellular protist. Its enormous size and highly visible internal cytoplasmic streaming make it a suitable research object for investigations on e.g. cytoskeleton and locomotion.

Despite lacking a brain or even rudimentary neuronal structures, the unassuming slime mold shows seemingly ‘intelligent’ behavior. Physarum can, for example, solve mazes and connect multiple food sources via the mathematically shortest pathway. Furthermore, the slime mold makes decisions, evaluates food sources based on their nutrient conditions, and even possesses a memory. These features are usually associated with life-forms which have a higher degree of information-processing sophistication.

Here, the underlying functions are not neuron-based, but are emergent phenomena, resulting from mechanochemical processes on and within the tubular network. Investigations of the ultrastructure and the dynamics of cell locomotion, a process which is intimately linked to the actomyosin cytoskeleton, reveal an alternative to neurological information processing: Sensory input from the environment is processed, and the information is then distributed as hydrodynamic oscillations throughout the network.

Classical decision-making tasks, which often involve abstract or game-like environments, frequently receive criticism regarding the ecological validity of the results. Therefore a different approach to researching decision making may be necessary, if the goal is to research how humans actually make decisions in their everyday lives. Naturalistic Decision Making (NDM) represents an attempt to create models for decisions made by humans using experience in realistic (or completely real) contexts. For the neurosciences, however, it is difficult to fulfill the basic principles of NDM, due to the constraints created by the methods of data acquirement. Yet approaching NDM may prove to be a remedy for the neurosciences' own problems with ecological validity. Accordingly, a quasi-realistic decision making task (QDM) was created, involving the evaluation of the weather based on two sources of information (forecast and picture of the sky) and the decision whether to take an umbrella along in the given situation.

The efficient construction and navigation of transportation networks remains a challenging task, and the analysis of real-world systems is often hindered by the lack of time-resolved data. This furthers the need for model systems such as the unicellular slime mold Physarum polycephalum, which self-organizes into a planar network of tubes optimized for transport efficiency and fault tolerance.

Remarkably, artificial and self-organized networks share important structural and functional traits, raising the question whether similar principles guide their assembly. We describe and model the time evolution of topology in P. polycephalum based on the statistics of events that modify network structure.

In opportunistic networks, nodes communicate whenever they are in the contact ranges of each other. Lack of end-to-end connectivity combined with high mobility of the nodes in these networks makes it challenging to deliver the packets to their respective destinations with minimal delays. Further, there is no central entity to make the packet forwarding decisions. In such networks, the routing protocols need be robust and resilient to make intelligent decisions about next hops and potential relay nodes for guiding the packets to their destinations.

This talk gives an overview about opportunistic networks, the factors that influence and help the routing decisions in opportunistic networks, the trade-offs involved in ensuring high data delivery and less delivery delay, the impact of network overhead and resource utility in these networks.