Joint estimation and identification of lidar log power returns in a switching environment.

The problem of estimating the return power in a laser integrated radar (lidar) system in the presence of multiplicative noise and partially unmodeled dynamics is explored. Several nonlinear methodologies are reviewed and compared to develop a systematic approach to signal model identification and estimation. The situations considered operate in mode-switching environments, that is, the desired unknown parameters are allowed to vary according to sudden jumps exhibiting discontinuous behavior at random times. Partitioning-based, parallel-structured techniques are shown to be significantly superior to the usual extended Kalman filter algorithm.