Asymptotic Efficiency of the PHD in Multitarget/Multisensor Estimation

Tracking an unknown number of objects is challenging, and often requires looking beyond classical statistical tools. When many sensors are available the estimation accuracy can reasonably be expected to improve, but there is a concomitant rise in the complexity of the inference task. Nowadays, several practical algorithms are available for multitarget/multisensor estimation and tracking. In terms of current research activity one of the most popular is the probability hypothesis density, commonly referred to as the PHD, in which the goal is estimation of object locations (unlabeled estimation) without concern for object identity (which is which). While it is relatively well understood in terms of its implementation, little is known about its performance and ultimate limits. This paper is focused on the characterization of PHD estimation performance for the static multitarget case, in the limiting regime where the number of sensors goes to infinity. It is found that the PHD asymptotically behaves as a mixture of Gaussian components, whose number is the true number of targets, and whose peaks collapse in the neighborhood of the classical maximum likelihood estimates, with a spread ruled by the Fisher information. Similar findings are obtained with reference to a naïve, two-step algorithm which first detects the number of targets, and then estimates their positions.

[1]  A. Wald Note on the Consistency of the Maximum Likelihood Estimate , 1949 .

[2]  J. Weatherwax,et al.  Solutions to Selected Problems In : Detection , Estimation , and Modulation Theory : Part I , 2014 .

[3]  Ba-Ngu Vo,et al.  The Gaussian Mixture Probability Hypothesis Density Filter , 2006, IEEE Transactions on Signal Processing.

[4]  Jeffrey D. Scargle,et al.  An Introduction to the Theory of Point Processes, Vol. I: Elementary Theory and Methods , 2004, Technometrics.

[5]  Y. Bar-Shalom,et al.  The probabilistic data association filter , 2009, IEEE Control Systems.

[6]  Ba-Ngu Vo,et al.  On performance evaluation of multi-object filters , 2008, 2008 11th International Conference on Information Fusion.

[7]  Ba-Ngu Vo,et al.  A Consistent Metric for Performance Evaluation of Multi-Object Filters , 2008, IEEE Transactions on Signal Processing.

[8]  A. V. D. Vaart,et al.  Asymptotic Statistics: Frontmatter , 1998 .

[9]  Roy L. Streit Multisensor multitarget intensity filter , 2008, 2008 11th International Conference on Information Fusion.

[10]  P. Willett,et al.  The GM-CPHD Tracker Applied to Real and Realistic Multistatic Sonar Data Sets , 2012, IEEE Journal of Oceanic Engineering.

[11]  R. Redner,et al.  Mixture densities, maximum likelihood, and the EM algorithm , 1984 .

[12]  Ronald P. S. Mahler,et al.  Statistical Multisource-Multitarget Information Fusion , 2007 .

[13]  L. L. Cam,et al.  Maximum likelihood : an introduction , 1990 .

[14]  Ba-Ngu Vo,et al.  Analytic Implementations of the Cardinalized Probability Hypothesis Density Filter , 2007, IEEE Transactions on Signal Processing.

[15]  Toby Berger,et al.  The quadratic Gaussian CEO problem , 1997, IEEE Trans. Inf. Theory.

[16]  E. L. Lehmann,et al.  Theory of point estimation , 1950 .

[17]  Yaakov Bar-Shalom,et al.  Multitarget/Multisensor Tracking: Applications and Advances -- Volume III , 2000 .

[18]  Don H. Johnson,et al.  On the asymptotics of M-hypothesis Bayesian detection , 1997, IEEE Trans. Inf. Theory.

[19]  H. V. Trees,et al.  Bayesian Bounds for Parameter Estimation and Nonlinear Filtering/Tracking , 2007 .

[20]  R. F.,et al.  Mathematical Statistics , 1944, Nature.

[21]  Paolo Braca,et al.  Multitarget-multisensor ML and PHD: Some asymptotics , 2012, 2012 15th International Conference on Information Fusion.

[22]  Peter Willett,et al.  The Bin-Occupancy Filter and Its Connection to the PHD Filters , 2009, IEEE Transactions on Signal Processing.

[23]  Ronald P. S. Mahler,et al.  Approximate multisensor CPHD and PHD filters , 2010, 2010 13th International Conference on Information Fusion.

[24]  Roy L. Streit,et al.  Poisson Point Processes: Imaging, Tracking, and Sensing , 2010 .

[25]  N. Reid Asymptotics and the theory of inference , 2003 .

[26]  J. E. Moyal The general theory of stochastic population processes , 1962 .

[27]  E. Lehmann Elements of large-sample theory , 1998 .

[28]  Dominique Heurguier,et al.  Multi-sensor PHD: Construction and implementation by space partitioning , 2011, 2011 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP).

[29]  Amir Dembo,et al.  Large Deviations Techniques and Applications , 1998 .

[30]  Harry L. Van Trees,et al.  Detection, Estimation, and Modulation Theory, Part I , 1968 .

[31]  Ba-Ngu Vo,et al.  Bayesian Filtering With Random Finite Set Observations , 2008, IEEE Transactions on Signal Processing.

[32]  A. M. Walker On the Asymptotic Behaviour of Posterior Distributions , 1969 .

[33]  Peter Willett,et al.  Matrix CRLB scaling due to measurements of uncertain origin , 2001, IEEE Trans. Signal Process..

[34]  W. Newey,et al.  Large sample estimation and hypothesis testing , 1986 .

[35]  I. R. Goodman,et al.  Mathematics of Data Fusion , 1997 .

[36]  R. Mahler Multitarget Bayes filtering via first-order multitarget moments , 2003 .

[37]  Chan‐Fu Chen On Asymptotic Normality of Limiting Density Functions with Bayesian Implications , 1985 .

[38]  Stefano Coraluppi,et al.  Aggregate surveillance: A cardinality tracking approach , 2011, 14th International Conference on Information Fusion.

[39]  S. Coraluppi,et al.  Multistatic Sonar Localization , 2006, IEEE Journal of Oceanic Engineering.