OTFcreds: Activity 3.5 Implement monitoring to detect the effectiveness of the system

[hellais]

OONI currently uses Grafana and Jupyter Notebook to implement various monitoring workflows that allow us to detect when there are unexpected changes to various properties related to our infrastructure. For example, we have alerts in place that notify us when there are drops in measurement volumes from any region. This might be the result of a software bug or a censorship incident.

As part of this project, we plan to expand our monitoring system to also include metrics related to trust in the probes and trigger alerts if, for example, we notice an unusually high number of “bad” measurements affecting a particular country or if we see some unusual patterns in the features of the anonymous credentials. Such metrics would be used to generate alerts and allow us to quickly respond to suspicious measurement submissions.

This will also allow us to understand how effective the system is at detecting malicious measurements and identify potential issues with its operations.

This activity will mainly be carried out by the OONI backend engineer.. It will also require some amount of input from the cryptographer to collect feedback on which metrics are most relevant to look at.

Output: updates to our monitoring infrastructure and backend documentation to allow us to detect how effective the system is.