Contact Us
State Estimation in Systems with Random Coefficients

State Estimation in Systems with Random Coefficients

Monday 12/01/2015
  • Daniel Sigalov
  • Work towards Ph.D. degree under the supervision of Prof. Y. Oshman
  • Classroom 165, ground floor, Library, Aerospace Eng.
  • The Interdepartmental Program for Applied Mathematics,
  • Technion – Israel Institute of Technology
  • The talk will be given in English

State estimation in dynamical systems with randomly switching coefficients is an important problem with a variety of applications. The most commonly met problems that are formulated within the random coefficients state-space approach, are maneuvering target tracking and fault-tolerant filtering. A related class of problems is referred to as data association or data ambiguity.
In this family, the estimation process is further complicated by the fact that the acquired data has uncertain origin. Typical applications include target tracking in clutter and multiple target tracking.  In this talk we aim at deriving a unified framework for state estimation problems under data and model uncertainties. First, we present the problem of tracking a maneuvering target restricted to perform a bounded number of maneuvers and show that, despite the inherent deviation from the standard Markov switching assumption underlying most state-of-the-art algorithms, it may be reformulated and solved using a generalized version of such tools. We proceed with an overview of linear minimum mean-square error (LMMSE) algorithms for state estimation in systems with random coefficients. We show that the novel formulation allows treatment of problems that have not been addressed in the LMMSE sense in the past. These include target tracking in clutter and multiple target tracking. Finally, we present a general framework, accompanied by a variety of applications, which allows a utilization of a single IMM-like algorithm to solve a variety of state estimation problems.  These include maneuvering target tracking, clutter and data association, multiple target tracking, tracking of splitting targets and more. In some cases it will be shown that the resulting unified IMM-like filter reduces to some classical algorithms such as PDA and JPDA.

Light refreshments will be served before the lecture
For more info
Please fill in the details

"*" indicates required fields

This field is for validation purposes and should be left unchanged.
*required fields are marked with an asterisk