Monitoring CNs: Report on Experimentations on CNs (v2)

This Deliverable reports the advancements in the development and in the application of the monitoring instruments for Community Networks (CNs). In Deliverable D2.5 [1] we described a set of metrics, together with the source code needed for their computation, which let someone “feel the pulse” of a CN and understand their level of decentralization and the possible presence of single points of failure. Such metrics are applicable to the network topology graph of a CN, but also to the social network graph, which can be derived through the analysis of communication instruments used by the community such as mailing lists. With these instruments, it is possible to perform a multi-layer analysis and identify those CN nodes that are critical for the robust operation of the network as well as those people who are key for the community to thrive. Since people own nodes, the two factors are strongly correlated and must be analysed together. It is key to note that robustness is a key factor for the sustainable growth of a CN. While the other tasks in WP2 focus on defining, motivating, and embedding by design the CN sustainability, these are largely offline processes. This task complement this vision with tools that make it possible to continuously (on-line) monitor the evolution of the CN, in order to prevent failures which hinder the growth of a network. This deliverable further develops this theme in two directions. The first one is the development of new metrics and new source code. An important part of this process is the realization of scientific publications for delving deeper into the methodology and developing it further according to the feedback from the academic and research community (in Chapter 2). The analysis of two running networks suggests that CNs, albeit being part of the family of “spatial networks”, behave in a different way depending on external conditions. In spatial networks, contrary to other networks such as social networks, the nodes have a position in space. Models of spatial networks typically tend to generate networks with a strong spatial hierarchy, where the geographical areas under the “influence” of each node are disjoint. In our experimental study, one of the analysed CN confirms this, while another exhibits a different pattern. We consider this an important result since it suggests that individual CNs may bear distinct features that distinguish them from other similar networks. The second direction of progress over D2.5 has dealt with the integration of the aforementioned metrics with existing software monitoring tools of CNs. In this regard, our work has focused on two currently used platforms, NodeShot and OpenWISP2. The two pieces of software have been created by members of the ninux.org network and by a community of people including a few developers from ninux, respectively. Whereas NodeShot is used only in the ninux community, OpenWISP2 is of broader interest; thus, we have focused our developments on this platform, which we detail in Chapter 3. The deliverable details the contribution of Task 2.4 on netCommons goals, discussing the impact on ninux and other CNs. There is an ongoing discussion with ninux members to encourage the adoption of solutions studied and proposed in netCommons Task 2.4, and the integration with OpenWISP2 will help disseminating these concepts outside the ninux community. Further adoption and impact of our software in the third year of the project will be documented as dissemination activities in WP6.