Discrimination Between Child and Adult Forms Using Rader Frequency Signature Analysis

In this thesis we develop a method to discriminate between adult and child radar signatures. In particular, we examine radar data measured from behind a wall, which introduces radar signal attenuation and multipath effects. To investigate the child/adult discrimination problem in a through-wall, multipath scenario, a previously developed free-space human scattering model was expanded to incorporate multiple paths, and the effects of transmission through, and reactions from, walls and ground. The ground was modeled as a perfectly reacting surface, while the walls were modeled as homogeneous concrete slabs. Twenty-five reflection paths were identified, involving the direct paths, as well as reflected paths between the ground and an adjacent wall.

All paths included two-way transmission through an obstructing wall. In addition to the model expansion, radar scattering measurements were collected from human volunteers, both child and adult, in order to investigate the classification of child and adult through-wall radar data. To our knowledge, this thesis is the first published work that analyzes child radar scattering data. A classification accuracy of above 95% was reached in both the time and frequency domains, indicating that, with minimal processing, a high degree of certainty is achievable in characterizing a subject behind a wall, in a real-world, multipath environment.