Objectives: In WP2 we concentrate on devising highly scalable gossiping algorithms to support data and (audio) signal aggregation aimed at social network analysis. Of particular concern will be devising solutions that are tolerant to poor and asymmetric communication. For example, we have devised a gossiping algorithm for computing averages in wired systems that relies on reliable communication links. Without such a reliability assumption, the algorithm will fail. In this work package, we will be looking at general continuous (signal and) discrete data aggregation solutions that can be proved to be correct within certain statistical bounds, among other comparable examples.

The groups at TUD-ES and TUD-MSP share a PhD researcher to work on low-level distributed aggregation protocols for continuous data (in particular, audio), with support from the hardware experts at DevLab. The PhD researchers at VUA and TUD will collaborate in adopting the gossiping protocols developed at VUA such that these can support efficient and effective high-level data aggregation, in particular statistical aggregation for social network analysis. These algorithms will be implemented on DevLab hardware for validation and demonstration purposes.