Body Sensor Networks: from Communication to Activity Classification

In this talk, we will present recent research results on the design and implementation of body sensor networks (BSNs). The first part of the talk will be devoted to "body networks:" on the basis of accurate experimental characterization of the the on-body propagation channels, we will discuss on its implications on network architecture design. Once deployed over the body, each node can be provided with (a lot of) "sensors," which can collect significant information on the person status. In the second part of the talk, we will focus on a specific application, given by activity classification (e.g., recognizing which type of movement a person is doing) through accelerometric and gyroscopic retrieved data. In particular, we will present the approach which lead the WASNLab team (http://wasnlab.tlc.unipr.it/) to win the first BSN Contest (http://bsncontest.org/). The first part of the talk is carried out in collaboration with the Wireless Communications Group of the Opera Department of the Université Libre de Bruxelles (Belgium).