Systems and Means of Informatics

2026, Volume 36, Issue 1, pp 45-67

ADAPTIVE AND ROBUST FILTERING ALGORITHMS FOR SYSTEMS WITH RANDOM OBSERVATION DELAYS: MAIN CONCEPTUAL AND ALGORITHMIC ASPECTS

  • S. A. Bosov
  • I. V. Uryupin

Abstract

The work is motivated by a specific class of navigation problems of autonomous underwater vehicles, for which the use of acoustic measurement means encounters their sensitivity to random delays in data arrival. At long distances, this effect may lead to a significant increase in estimation error even at moderate motion speeds. The existing formal mathematical formulation reduces to the problem of state estimation for stochastic dynamic systems with random observation time delays under conditions of incomplete prior information. The practical implementation problem, on which the paper is focused, reduces to the development and software implementation of computationally efficient stochastic filtering algorithms. The method of linear pseudomeasurements adapted to an observation model with random delay is used as the basic tool. In addition to previously considered formulations with complete prior information on the parameters of the motion and observation models, the paper analyzes cases of incomplete information typical in practice. For two of them - uncertainty of measurement accuracy characteristics at the stage of target detection and unknown error distributions under changing observation conditions, methods for solving practical problems are proposed. The algorithms are described within the framework of the general objective - to form a conceptual approach to the construction of conditionally optimal, adaptive, and robust filtering algorithms for the specified classes of models.

[+] References (50)

[+] About this article