You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
{{ message }}
Repository navigation
Come up with a definition of a session and how to track it #1623
It was brought up within the scope of the OOID (ooni/backend#173) that it would be desirable to have something like a session ID to track measurements which are related to each other.
We should first define what we mean by a session and then come up with a solution for this.
Within the scope of this I had an idea that maybe also helps us with a different story. What if we generated a random ID with which we stamp measurements, which we persist for a fixed window of time (ex. it reset every first day of month or similar) and on the probe_cc, probe_asn pair (if the probe changes network it will generate a new one).
This way we effectively have an identifier of the vantage point which reset every some amount of time.
I think this is going to help us significantly in being able to look at measurements which are pertinent to a particular "session", it is actually even broader than that as it would allow us to link measurements to the same vantage point over time.
The fact that it is reset every fixed period of time and also reset across probe_cc, probe_asn pairs, means that it cannot be used to track the movements of users, so it should be pretty privacy preserving.
We can also use this ID to be able to better estimate app usage metrics.
Given that we now have the probe_id field, I think the story I mentioned above is no longer relevant. We might still at some point care to introduce some kind of session identifier, but I suggest we file that as a separate better defined issue.
Activity
hellais commentedon Mar 16, 2020
Within the scope of this I had an idea that maybe also helps us with a different story. What if we generated a random ID with which we stamp measurements, which we persist for a fixed window of time (ex. it reset every first day of month or similar) and on the probe_cc, probe_asn pair (if the probe changes network it will generate a new one).
This way we effectively have an identifier of the vantage point which reset every some amount of time.
I think this is going to help us significantly in being able to look at measurements which are pertinent to a particular "session", it is actually even broader than that as it would allow us to link measurements to the same vantage point over time.
The fact that it is reset every fixed period of time and also reset across probe_cc, probe_asn pairs, means that it cannot be used to track the movements of users, so it should be pretty privacy preserving.
We can also use this ID to be able to better estimate app usage metrics.
bassosimone commentedon Mar 16, 2020
This is a good idea, 👍
hellais commentedon Aug 14, 2026
Given that we now have the probe_id field, I think the story I mentioned above is no longer relevant. We might still at some point care to introduce some kind of session identifier, but I suggest we file that as a separate better defined issue.