An integral and differential geometric approach to behavioral information acquisition and integration via binary sensor networks

This work presents an integral and differential geometric approach to acquisition and integration of human behavioral information in their movements via distributed binary sensor networks. In order to tolerate low signal-to-noise ratio (SNR) and reduce data throughput, each sensor only detects the presence of subject motion within its field of view (FOV). By utilizing multiple sensing modalities and novel sampling geometries, behavioral information of human movements, including dynamic and static features, can be better measured from statistics of binary sensory signals. The novelty of this work lies in three aspects: (1) utilizing the invariant measure density of subject motion group to build geometric probability models that associate subject behaviors with binary measurements; (2) utilizing affine connections of sensor models to achieve behavioral parameter estimation in a sensor-independent way; (3) utilizing belief aggregation methods to integrate behavioral information measured by distributed sensors. Initial results have verified the effectiveness of the proposed method.