Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Harmonic Mean Density Fusion in Distributed Tracking: Performance and Comparison

Nikhil Sharma, Ratnasingham Tharmarasa, Thiagalingam Kirubarajan

TL;DR

The paper tackles the challenge of fusing distributed tracks under unknown cross-correlations in multi-sensor target tracking. It introduces Harmonic Mean Density (HMD) fusion, showing it minimizes a weighted Pearson \\chi^2 divergence and yields a principled, non-inflated fusion of local densities, including Gaussian mixtures. Two implementations are developed: HMD-GA (Gaussian-Approximation) with closed-form fusion equations and HMD-S (sampling-based) for flexible, non-Gaussian cases; the work also analyses connections and tensions with inverse covariance intersection (ICI). Through consistency analyses and three simulation scenarios, HMD-GA demonstrates the tightest consistent fusion among conservative methods and often outperforms CI and ICI on RMSE/NEES, while the sampling approach offers robustness via covariance inflation when needed. Overall, HMD provides a scalable, robust framework for distributed target tracking with unknown correlations, suitable for integration with advanced trackers and anomaly-tolerant systems.

Abstract

A distributed sensor fusion architecture is preferred in a real target-tracking scenario as compared to a centralized scheme since it provides many practical advantages in terms of computation load, communication bandwidth, fault-tolerance, and scalability. In multi-sensor target-tracking literature, such systems are better known by the pseudonym - track fusion, since processed tracks are fused instead of raw measurements. A fundamental problem, however, in such systems is the presence of unknown correlations between the tracks, which renders a standard Kalman filter (naive fusion) useless. A widely accepted solution is covariance intersection (CI) which provides near-optimal estimates but at the cost of a conservative covariance. Thus, the estimates are pessimistic, which might result in a delayed error convergence. Also, fusion of Gaussian mixture densities is an active area of research where standard methods of track fusion cannot be used. In this article, harmonic mean density (HMD) based fusion is discussed, which seems to handle both of these issues. We present insights on HMD fusion and prove that the method is a result of minimizing average Pearson divergence. This article also provides an alternative and easy implementation based on an importance-sampling-like method without the requirement of a proposal density. Similarity of HMD with inverse covariance intersection is an interesting find, and has been discussed in detail. Results based on a real-world multi-target multi-sensor scenario show that the proposed approach converges quickly than existing track fusion algorithms while also being consistent, as evident from the normalized estimation-error squared (NEES) plots.

Harmonic Mean Density Fusion in Distributed Tracking: Performance and Comparison

TL;DR

The paper tackles the challenge of fusing distributed tracks under unknown cross-correlations in multi-sensor target tracking. It introduces Harmonic Mean Density (HMD) fusion, showing it minimizes a weighted Pearson \\chi^2 divergence and yields a principled, non-inflated fusion of local densities, including Gaussian mixtures. Two implementations are developed: HMD-GA (Gaussian-Approximation) with closed-form fusion equations and HMD-S (sampling-based) for flexible, non-Gaussian cases; the work also analyses connections and tensions with inverse covariance intersection (ICI). Through consistency analyses and three simulation scenarios, HMD-GA demonstrates the tightest consistent fusion among conservative methods and often outperforms CI and ICI on RMSE/NEES, while the sampling approach offers robustness via covariance inflation when needed. Overall, HMD provides a scalable, robust framework for distributed target tracking with unknown correlations, suitable for integration with advanced trackers and anomaly-tolerant systems.

Abstract

A distributed sensor fusion architecture is preferred in a real target-tracking scenario as compared to a centralized scheme since it provides many practical advantages in terms of computation load, communication bandwidth, fault-tolerance, and scalability. In multi-sensor target-tracking literature, such systems are better known by the pseudonym - track fusion, since processed tracks are fused instead of raw measurements. A fundamental problem, however, in such systems is the presence of unknown correlations between the tracks, which renders a standard Kalman filter (naive fusion) useless. A widely accepted solution is covariance intersection (CI) which provides near-optimal estimates but at the cost of a conservative covariance. Thus, the estimates are pessimistic, which might result in a delayed error convergence. Also, fusion of Gaussian mixture densities is an active area of research where standard methods of track fusion cannot be used. In this article, harmonic mean density (HMD) based fusion is discussed, which seems to handle both of these issues. We present insights on HMD fusion and prove that the method is a result of minimizing average Pearson divergence. This article also provides an alternative and easy implementation based on an importance-sampling-like method without the requirement of a proposal density. Similarity of HMD with inverse covariance intersection is an interesting find, and has been discussed in detail. Results based on a real-world multi-target multi-sensor scenario show that the proposed approach converges quickly than existing track fusion algorithms while also being consistent, as evident from the normalized estimation-error squared (NEES) plots.

Paper Structure

This paper contains 25 sections, 2 theorems, 87 equations, 27 figures, 3 tables.

Table of Contents

  1. Introduction
  2. Covariance Intersection and Variants
  3. Fusion of Gaussian Mixtures
  4. Problem Formulation
  5. Harmonic Mean Density Fusion
  6. Average Divergence Minimization
  7. Consistency Analysis
  8. Implementation using Gaussian Approximation
  9. Condition of Consistency of HMD-GA.
  10. Relation with ICI
  11. Correlation Structure Employed for ICI
  12. Contradictions in the design of ICI
  13. Generalized Mean Density Fusion using Importance Sampling.
  14. Sampling based HMD fusion of Gaussian Mixtures
  15. Which Density to Sample From ?
  16. ...and 10 more sections

Key Result

Theorem 1

The harmonic mean density between $p_1(\mathbf{x})$ and $p_2(\mathbf{x})$ minimizes the $\omega$-weighted average Pearson $\chi^2$ divergence, and thus, the reverse Neyman $\chi^2$ divergence.

Figures (27)

  • Figure 1: Uncertainty region (86.5%) for HMD and GMD fusion of Gaussian densities using proposed sampling technique.
  • Figure 2: Fusion of Gaussian mixtures using proposed sampling technique.
  • Figure 3: Comparison of fusion weights in case of scalar estimates.
  • Figure 4: Consistency-based test scenario 1.
  • Figure 5: Consistency-based test scenario 2.
  • ...and 22 more figures

Theorems & Definitions (3)

  • Theorem 1
  • Definition 1
  • Theorem 2