C-BRAIN 2017: Distance Error Study (M. Tavares, SITI-COPELABS)

This presentation is dedicated to the presentation of field trial studies concerning distance measurement, that have been carried out with the SITI middleware NSense.

Traces available via the Proxemics Data Lab project

We have carried out a set of experiments with our middleware NSense installed on three devices and carried by three different individuals. The devices were Samsung S5 Galaxy, Android version 6.0.1. Each device was named User1, User2, and User3. Each experimental scene involved 2 devices (e.g., User1 to User2 or User3 to User1 readings) positioned in different locations, indoor or outdoor, on the same floor or in different floors, within and without LoS. For each scenario, the real distance has been measured horizontally or, if on different floors, vertically. LoS is 1 if the path was free of obstacles and 0 if there were obstacles on the way; SameFloor is 1 if the users are on the same floor and 0 otherwise; Outdoor is 1 if users are outdoors and 0 otherwise; the real distance between the devices, in meters. The experiments have been developed over two weeks, between 10 a.m. and 17 p.m. of different working days of the week. There are two different datasets: BTDataset.csv WiFiDataset.csv Each dataset has the following variables: - timestamp - own_device (identifier of the device) - rssi, RSSI in dB - Scale, RSSI levels (1 as Very Weak, 2 as Weak , 3 as Mild, 4 as Strong and 5 as Very Strong). RSSI deterministic levels are based upon the Android getLevel() method. - freq, corresponding to the sub-frequency of the channel - computed distance, in meters - scenario, label of the different scenarios - refer to ScenarioDescription.xls - realdistance in meters - line_of_sight. 1 for LoS; 0 for nLoS - same floor. 1 for users standing on the same floor; 0 for different floors - outdoor, 1 for both users being outdoors; 0 otherwise - connected_device, device from which the RSSI is being obtained.