Monitoring Instruments for CNs (v1)

This deliverable describes the work done to define a set of multi-layer graph metrics that Community Networks (CNs) can use to understand the degree of decentralization of both their communication and social networks. These metrics are a fundamental step toward a proper monitoring of CNs because they allow measuring several properties in network fragility and decentralization. This is needed because the evolution of a CN is generally unplanned, and there is no guarantee that, during its growth, the network evolves spontaneously in a robust and resilient network or it evolves toward a structure that contains single points of failure at one or more levels. These points of failure may reside in nodes that become topologically critical for the network itself, in infrastructure owners who own a set of nodes that give them too much potential influence, or in people in the social network of the community who become too much important in the discussion and in the organization of the community. In the worst option, the same person can represent a point of failure both in the communication and social network. This deliverable formalises some of these concepts and introduces metrics that can detect the presence of such points of failure. The work in this deliverable is strongly based on two publications, one of which is currently available [1] and another is under preparation. To realize the publications, open source code has been released and is also partly documented in this deliverable. The metrics and the methodology applied to the ninux network revealed that ninux is a theoretically decentralized network but in practice, at the time of our analysis had one single point of failure represented by a person that owned a number of critical nodes and was also very influential in the mailing list discussion. From this observation a heuristic algorithm to re-assign the property of some nodes was designed. As part of the future agenda of Task 2.4 (T2.4), where this deliverable has been developed, we will propose the adoption of the monitoring software and of the proposed metrics directly in the web server used by the ninux community, in order to monitor the network development and avoid the future creation of more points of failure.