emileaben/minrtt

minimum RTT dataset and viz from RIPE Atlas

minrtt's Issues

Data quality improvement: Networks not appearing in the dataset because they are not in RIPE Atlas traceroutes

Some networks are close to probes but we don't see them because no RIPE Atlas measurements are scheduled to or through this network.

Idea: investigate what measurements would need to be added to efficiently counter this effect.

Ability to figure out which measurement 'caused' a minRTT result

adding metadata (msm_id, ts) so one can figure out which measurement 'caused' a minRTT result.

Background: To build confidence in methodology it is important to figure out what piece of data caused a particular result in the minRTT data aggregates. So annotating each minRTT tuple (af,protoc,prb_id,asn) with data so one can find the corresponding RIPE Atlas data quickly would be really good. (msm_id, ts, and prb_id )together allow for looking up indivicual measurement results, and since we already have prb_id in the minRTT tuples, we'd only need to add msm_id and extact timestamp of the measurement.

data quality issue due to 'RTT lies'

If devices near a probe respond with a packet with src IP that is not 'their own' this results in data quality issues.

We see this in TCP (interceptor devices) more then other protocols.

We need methods to counter this type of data quality issue to increase data quality

data quality due to probe geolocation errors

If probes are not correctly geolocated this results in incorrect latency tuples.

We should find ways to exclude probes with incorrect geolocation from this dataset.

(overlaps with work Chris Amin is doing on IPMap)