Multi-target tracking in distributed sensor networks using particle PHD filters

Abstract

Multi-target tracking is an important problem in civilian and military applications. This paper investigates multi-target tracking in distributed sensor networks. Data association, which arises particularly in multi-object scenarios, can be tackled by various solutions. We consider sequential Monte Carlo implementations of the Probability Hypothesis Density (PHD) filter based on random finite sets. This approach circumvents the data association issue by jointly estimating all targets in the region of interest.To this end, we develop the Diffusion Particle PHD Filter (D-PPHDF) as well as a centralized version, called the Multi-Sensor Particle PHD Filter (MS-PPHDF). Their performance is evaluated in terms of the Optimal Subpattern Assignment (OSPA) metric, benchmarked against a distributed extension of the Posterior Cramér-Rao Lower Bound (PCRLB), and compared to the performance of an existing distributed PHD Particle Filter. Furthermore, the robustness of the proposed tracking algorithms against outliers and their performance with respect to different amounts of clutter is investigated.

Distributed Multitarget Tracking Based on Diffusion Strategies Over Sensor Networks

“…In [36], Battistelli et al developed a consensus CPHD filter that provided a distributed solution, where each node first calculates its local estimate with its own measurements, and then it calls for consensus iterations to achieve global fusion over the network by iterating local fusion among neighbors. More recently, Leonard and Zoubir [37] developed a distributed particle filter implementation of the PHD filter for multitarget tracking. Since a large number of weighted particles are generated at each node and communicated between neighbors for the adaptation step and the combination step in [37], this algorithm requires high computational complexity and communication load.…”

“…More recently, Leonard and Zoubir [37] developed a distributed particle filter implementation of the PHD filter for multitarget tracking. Since a large number of weighted particles are generated at each node and communicated between neighbors for the adaptation step and the combination step in [37], this algorithm requires high computational complexity and communication load.…”

“…On the other hand, compared to the distributed particle filter implementation in [37], our method has much lower computational complexity and communication load since the large number of weighted particles are not required.…”

“…R.1: Given A, Q, m, and P with appropriate dimensions, where Q and P are positive definite, then [40], [41] N (x; Aw, Q)N (w; m, P)dw = N (x; Am, Q+APA T ). (36) R.2: Given B, R, m, and P with appropriate dimensions, where R and P are positive definite, then [40], [41] N (z; Bx, R)N (x; m, P) = q(z)N (x; m, P), (37) where q(z) = N (z; Bm, R + BPB T ), m = m + K (z − Bm), P = (I − KB)P and K = PB T (BPB T + R) −1 . R.3: The power of a Gaussian component is a Gaussian component [36] [αN (x; m, P)] w = α ω k(ω, P)N (x; m, P/ω), (38) where k(ω, P)…”

Tracking of Multiple Closely Spaced Extended Targets Based on Prediction-Driven Measurement Sub-Partitioning Algorithm

“…It was evolved into the extended target PHD (ET-PHD) filter [21] by assuming that the number of measurements obeys the Poisson distribution and measurements are distributed around the target based on an inhomogeneous spatial Poisson point process [22]. In recent years, many multiple extended target tracking approaches based on the ET-PHD filter and its variants have been proposed [23][24][25][26], e.g., the extended target Gaussian mixture PHD filter [27], the extended target gamma GIW-PHD (ET-GGIW-PHD) filter [28][29][30] based on the random matrix approach [31][32][33], and the ET-GM-PHD filter based on random hypersurface model [34][35][36].…”