Particle Filter for Randomly Delayed Measurements with Unknown Latency Probability

Area of Science:

• Signal Processing
• Control Systems
• Estimation Theory

Background:

• Particle filters are crucial for state estimation in non-linear systems.
• Randomly delayed measurements and packet drops degrade filter performance.
• Existing methods often assume known delay probabilities or lack robustness.

Purpose of the Study:

• To develop a robust particle filter for systems with unknown random measurement delays and packet drops.
• To formulate a generalized measurement model accommodating these uncertainties.
• To propose algorithms for estimating unknown latency parameters.

Main Methods:

• A generalized measurement model incorporating random delays and packet drops.
• Formulation and derivation of the probability density function (pdf) for correlated measurement noise.
• Development of recursion equations for importance weights.
• Maximum likelihood-based offline and online algorithms for latency parameter identification.
• Analysis of filter convergence conditions.

Main Results:

• The proposed particle filter effectively handles correlated noise due to random delays and packet drops.
• Novel algorithms accurately identify unknown latency parameters.
• Simulations demonstrate superior performance compared to state-of-the-art methods.
• Convergence of the proposed filter is theoretically explored.

Conclusions:

• The developed particle filter provides a robust solution for state estimation with unknown random measurement delays and packet drops.
• The proposed latency identification algorithms enhance filter accuracy and reliability.
• The filter shows significant advantages in non-stationary growth models and target tracking scenarios.