Fusion of GPS and Accelerometer Information for Anomalous Trajectories Detection
Jorge Navarro, Isaac Martín de Diego, Alberto Fernández-Isabel, Felipe Ortega · 2019
Nowadays the rise of Internet of Things (IoT) has led the use of a wide variety of sensor types able to collect multiple types of data. In order to manage this heterogeneous data and gather knowledge, it is useful to combine different sources of information. In this paper, a methodology to detect outliers combining several sources of information at similarity level is proposed. First, a similarity measure is defined for each source of information. These measures are combined using clustering results to produce a general similarity matrix. Then, this matrix is used to train a One-class Support Vector classifier in order to detect outliers. Experiments on the cattle domain have been achieved in order to illustrate the viability of the proposal. The geospatial information collected from a GPS system is combined with accelerometer information. The method detects anomalous animal trajectories and calving events in outfield environments.