Adaptive Hessian Estimation Based Extremum Localization

Abstract

In this paper we study continuous time adaptive extremum localization of an arbitrary quadratic function $F(\cdot)$ based on Hessian estimation, using measured the signal intensity by a sensory agent. The function $F(\cdot)$ represents a signal field as a result of a source located at the maximum point of $F(\cdot)$ and is decreasing as moving away from the source location. Stability of the proposed adaptive estimation and localization scheme is analyzed and the Hessian parameter and location estimates are shown to asymptotically converge to the true values. Moreover, the stability and convergence properties of algorithm are shown to be robust to drift in the extremum location. Simulation test results are displayed to verify the established properties of the proposed scheme as well as robustness to signal measurement noise.