CRAWDAD dataset copelabs/usense (v. 2017-01-27)

This dataset comprises experiments carried out with the open-source middleware NSense (fomerly named as USense), available at https://github.com/COPELABS-SITI/NSense.

The data has been collected based on four sensors: bluetooth; Wi-Fi; microphone; accelerometer. NSense then relies on four different pipelines to compute aspects such as relative distance (Wi-Fi); social strength (based on bluetooth contact duration); sound activity level; motion.

We set up experiments making use of Samsung Galaxy S3 devices. For each experiment, there is the following set of data files:
- SocialProximity.dat has three columns: Timestamp, DeviceName, Encounter Duration, Average Encounter Duration, Social Strength (Per hour) and Social Strength(Per minute) towards DeviceName
- DistanceOutput.dat has three columns: Timestamp, DeviceName, and Distance towards DeviceName
- Microphone.dat has two columns: Timestamp, and Soundlevel(QUIET, NORMAL, ALERT and NOISY)
- PhysicalActivity.dat has two columns: Timestamp, and Activity as STATIONARY, WALKING and RUNNING

There are two tracesets. A first traceset has been collected relying on a first NSense version in 2015. Then, a second traceset has been collected in 2016, with a refined version of NSense.

In all tracesets, devices have been carried around by people that share the same affiliation during their individual daily routines (24 hour periods).