Semiparametric Estimation of Target Locations

• Activity Number: 238
• Type: Contributed
• Date/Time: Monday, July 30, 2012 : 2:00 PM to 3:50 PM
• Sponsor: Section on Nonparametric Statistics
• Author(s): Nirupam Chakrabarty*+ and Moulinath Banerjee and George Michailidis
• Companies: University of Michigan and University of Michigan and University of Michigan
• Keywords: semi parametric ; kernel ; spline ; signal

Abstract

Detection,identification and tracking of spatial phenomena are important tasks in various environmental and infrastructure applications. Wireless sensor networks are widely used for monitoring natural phenomena in space and over time, as well as for target detection and tracking. Sensors acquire signals emitted from the target that are corrupted by noise,and then try to detect the location of the target. In most cases, the form of the signal generating model is assumed to be known for subsequent analysis. But the assumption of a known signal model can be restrictive, since different types of targets may emit different kinds of signals. In this project, we introduce a semi parametric model for signal propagation where both the location of the target and the signal generating function are treated as unknown. Here we exploit the fact that the signal strength received by the sensor decreases as its distance from the target increases. In the proposed model, the parametric part (target location) and the nonparametric part (signal function) are "bundled" together, which makes the problem more challenging. We develop semi parametric procedures using both spline and kernel based techniques.