The PAF-FPE Project

This is the website of the PAF-FPE project. A Marie-Curie project run by Leonardo Maccari in the period 2011-2014. I do not plan to update this website anymore, If you are interested in this project, you can find a description and a list of achievements in this short Executive Summary. In that pdf I will also update the publication list as the last papers get published.

Cooperation with the SmartUC project

The EIT project "Smart Ubiquitous Computing" is another project our lab is involved at the moment. For this project a new release of the Peerstreamer video streaming platform has been produced. Among other objectives, the group is working to have Peerstreamer run on wireless community networks. On this topic we started a cooperation that involved also the PAF-FPE project. On one side, EIT is financing the developments needed to port Peerstreamer on such a particular context, on the other, I'm working to take into account privacy-related issues coming from the analysis of the network topologies. Together with the rest of the group we are releasing some experimental code to analyse the network graph of a wireless community network and identify the group of nodes with the highest betweenness (B) and closeness centrality (C). A wireless community network is a wireless mesh network installed on the roof of a community of people. Those people will use it in order to offer internal services to the community and to access the Internet. These two centrality concepts are interesting both under a security and under a performance point of view. In particular B represents the best group an attacker can control in order to be able to intercept the highest fraction of traffic on the network. I.e, if the attacker is able hack into the nodes of the network, it's his best choice (among all the groups of the same size) to sniff the highest number of traffic flows. Imagine that Peerstreamer is installed on all the nodes in the community network (so it is embedded in the linux distribution of the wireless nodes), Peerstreamer could be programmed in order to avoid routing the video streams on untrusted nodes at least up to when some trust has been established with some out-of-band procedure. Identifying B is the first step to benchmark such a strategy since it can be seen as a measure of the robustness that the network topology offers under a security/privacy point of view. Another close concept is the closeness group centrality. In this case, imagine that peerstreamer is, again, embedded on all the nodes in the network but it is not used to convey the streams to the final users, it is used as a back-end application to replicate the streams only on some selected nodes that will then offer the video using some web-based service. This infrastructure is a possible evolution of Peerstreamer that would mix the p2p and client/server approach. A group with the highest closeness centrality is the group that guarantees the best average distance (in terms of hops, or any other quality routing metric) to any node in the network. Given a maximum distance value, identifying this group is the first step to offer a reliable web-based service on a community wireless network.The code published here is the first effort to explore those very interesting new applications for Peerstreamer.