Speed data Clause Samples

Speed data. The Android location data provides multiple types of information that accompany the actual location estimation as shown in the previous subsection. One of this information is also the instantaneous speed of the user at the moment the location estimation was generated from the fused location sensor. As observed in the StudentLife dataset the speed field of the location sensor provides often zero speed for the user. This is happening most probably because the user is not moving at the point when the location estimation is generated. However, this does not allow the BehavAuth system to leverage the speed of the user in the environment in order to learn the speed patterns. For that reason, the average speed of the user in the environment was calculated. The average speed was calculated through the knowledge of the current and previous location of the user and the time required to move from one point to the other. This provided a more representative movement estimation of the user. More details about the calculation process is provided in the next paragraph where the behavioural cues are calculated. In the previous case of the location-­based pattern recognition, the smartphone has a dedicated sensor that provides location measurements. However, there is no sensor that provides a measurement of the users’ speed. So, there is a need to derive the users speed through another sensor. The majority of the smartphones today have integrated an accelerometer that provides instantaneous measurement of 3-­dimensional acceleration, from which the speed could be derived. However, in order to derive the speed from the accelerometer there is a need to compute the integral of the acceleration, which over time produces a drift due to the accumulated error. Another way that could provide an average estimation of the users’ speed over the time is to take into account the distance the user moved over a period of time. The average speed is provided by a general form velocity = distance / time. It was assumed that each user started to use the mobile data collection app from stance, where velocity was 0. So, the initial speed sample for each user is 0. Then, for each sequential pair of locations and timestamps, the distance and time were calculated. The distance was calculated similarly with the (a) point in kilometres. The time was calculated by subtracting the newest timestamp minus the previous timestamp – both timestamps were represented in Unix time – which resulted in the time...