Due to complex interdependencies and feedback loops between network layers and nodes, the development of adaptive applications is difficult. As networking applications respond nonlinearly to changes in the environment and adaptations, defining concrete adaptation rules is nontrivial. In this paper, we present the offline learner Fossa, which uses genetic programming to automatically learn suitable Event Condition Action (ECA) rules. Based on utility functions defined by the developer, the genetic programming learner generates a multitude of rule sets and evaluates them using simulations to obtain their utility. We show, for a concrete example scenario, how the genetic programming learner benefits from the clear model of the ECA rules, and that the methodology efficiently generates ECA rules which outperform nonadaptive and manually tuned solutions.

In distributed real-time applications, such as online games or interactive conferencing systems, timeliness of update event dissemination is a prevailing requirement. Direct connections are often the best solution. However, participants' connection bandwidths, especially uplink capacities of asymmetric end user Internet connections, impose a limit to this. We present and evaluate a lightweight local algorithm that optimizes many-to-many event dissemination in highly dynamic scenarios. The algorithm exploits the high clustering of interest networks, as observed in virtual environments based on local vision, but also in social networks of many kinds. Our approach allows reducing traffic by aggregating messages as well as balancing traffic among participants based on their capabilities.

In Networked Virtual Environments (NVE), Interest Management (IM) is a key part of the system that determines which of a participant's actions have to be communicated to which subset of the other participants. Since traditional client/server approaches have several drawbacks, a variety of alternative peer-to-peer (P2P) approaches have been presented by the community [1]. In many of these approaches, interest management is also responsible for maintaining connectivity of the P2P network. Aiming for latency-sensitive applications, systems like VON [2] and pSense [3] use mutual notification mechanisms, building the topology based on the participants' virtual world proximities. A common limitation of these, however, is their fixed dimensionality: most are only designed for two spatial dimensions, a few are capable of handling three. In this work, we discuss options for dealing with an arbitrary number of dimensions. We propose a dynamic localized peer-to-peer IM that supports and exploits any number of dimensions. In our IM, peers communicate directly within their vision range. The space outside of the vision range is divided in sectors, each guarded by a sensor node that notifies them of approaching peers. We determine criteria for an efficient sector partitioning, discuss several approaches and present a suitable algorithm.

In this demo we present a testbed environment for Peer-to-Peer (P2P) game architectures. It is based on Planet PI4, an online multiplayer game whose gameplay provides a standard workload for a set of gaming-specific network interfaces. Its pluggable architecture allows for the evaluation and comparison of existing and new P2P networking approaches. Planet PI4 can run on a real network for prototypical evaluation as well as in a discrete-event simulator providing a reproducible environment.

Different evaluation methods for distributed systems like prototyping, simulation and emulation have different tradeoffs. We present a testbed for Internet applications that supports real-network prototypes and multiple simulators with unchanged application code. To ensure maximum portability between runtimes, a compact but flexible system interface is defined.