Versions Compared
Key
- This line was added.
- This line was removed.
- Formatting was changed.
...
| Table of Contents | ||
|---|---|---|
|
Simulated datasets
Description
For AI/ML in general, and In the context of Energy Savings in particular, it is important that datasets offering training data exist. The SIM project proposes currently 2 data sets to be leveraged: one based on real network data, coming from Telecom Italia, and another one based on synthetic data created with the help of Viavi.
Telecom Italia dataset
Where
The TIM data set representing the “Internet” usage from the network, for 2 months (Nov and Dec 2013, timestamps every 10 minutes) is published in O-RAN SC nexus: https://nexus3.o-ran-sc.org/#browse/browse:datasets
How
Just download the data set and extract.
Details
The data is parsed, meaning we pruned the voice and SMS usage, keeping only the Internet traffic. The values themselves for the usage are normalised to a value known only to TIM (assuring some anonymisation).
It contains data for November and December 2023, with timestamps every 10 minutes. We have a different file for each of the days.
The data is from the city of Milano. As described in the references, the city was divided into 10.000 squares (roughly a grid of 100 x 100 squares). That is the SquareID (between 1 and 10000) column. The timestamp is the start time interval of the measurement. The end time will be start time + 10 minutes. Please note that the timestamp is GMT+1, Milano local time. The InternetTraffic represents “number of CDRs generated inside a given Square id during a given Time interval.”. It does not have any unit of measure, because it is somehow normalised such that it is anonymised. It has no meaning in itself, but we can see patterns of how that value increases and decreases over time.
Other references
- https://dataverse.harvard.edu/dataset.xhtml?persistentId=doi:10.7910/DVN/EGZHFV
- https://www.nature.com/articles/sdata201555
Viavi dataset
Where
TBD, it will be put somewhere in the nexus repo.
How
Download the data set. Files will be password protected. Whoever is interested in using the data can comment on this page and will receive the file password via email.
Details
To be given here.