Work

Large deviation estimates for nonlinear filtering with discontinuity and small noise

Year: 2025

Type: article

Abstract: This paper develops large deviation estimates for nonlinear filtering with discontinuity in the drift of the signal and small noise intensities in both the signal and the observations. A variational approach related to Mortensen’s optimization problem is utilized in our analysis. The discontinuity of the drift in the signal naturally arises in many applications, including modeling communication channels with a “hard limiter”. Our results extend the work of Reddy et al. (2022), in which smooth functions were used. To address the discontinuous in the drift of the signal, relaxed controls are used to study the asymptotic fraction of time the controlled signals spend in each half-space divided by the discontinuity hyperplane. Large deviation estimates are established by the weak convergence method using the stochastic control representation for positive functionals of Brownian motions and Laplace asymptotics of the Kallianpur–Striebel formula.

Source: Stochastic Processes and their Applications

DOI: 10.1016/j.spa.2025.104662

Authors: Hongjiang Qian, Yanzhao Cao, George Yin

Institutions: Auburn University, University of Connecticut

Language: English

Topic: Target Tracking and Data Fusion in Sensor Networks

Subfield: Artificial Intelligence

Field: Computer Science

Domain: Physical Sciences

Open Access status: hybrid

Estimated APC paid: $3,520

Funders: National Science Foundation, U.S. Department of Energy