Imaging and tracking of targets in clutter using differential time-reversal techniques

Abstract

Two time-reversal algorithms for identifying, imaging, and tracking moving targets in clutter are introduced. The first algorithm classifies existing scatterers into stationary vs. moving targets. Multistatic data matrices (MDMs) corresponding to successive radar acquisitions (snapshots) of the scene are recorded. Singular value decomposition of the (time-)averaged MDM provides information on stationary targets, whereas singular value decomposition of the differential MDM provides information on moving targets. The second algorithm yields real-time selective tracking of each moving target by means of differential time-reversal. It requires minimal processing and memory resources, and exploits distinctive features of timereversal such as statistical stability and superresolution. Numerical simulations are used to illustrate the capabilities of the proposed algorithms in different scenarios involving clutter from discrete secondary scatterers and from inhomogeneous random medium backgrounds.

Time-Reversal Microwave Imaging Based on Random Configuration of Transmitters or Receivers

“…Since the displacement of target movements between different snapshots is assumed much shorter than the wavelength (λ = 30 cm) most of the contribution from stationary target and multiple scattering among target movements and stationary target cancel out. Thus, the eigen-spectrum of the differential MDM will mostly correspond to the target movements [13]. Here we record the waves, coming out from the antenna located at unknown locations using receiving antenna array behind the wall.…”

“…To have an accurate point-like image of the moving target in the presence of clutter object, we have to use the differential MDM approach. Traditionally [13], the differential MDM approach is implemented by subtracting the MDMs at two consecutive snapshots of the moving object as…”

Moving Target Tracking Using Time Reversal Method

“…Prior works on imaging non stationary objects using time reversal method are based on assumption of slowly varying targets in a sense that frequency distortion of transmitting signal could be neglected [25]. Therefore the time reversal operator decomposition (DORT) [8,16], Figure 4.…”

“…Furthermore, in [24] a single antenna scenario was investigated for multipath compensation in the presence of Doppler shifts due to target motion in a dense multipath environment. In [25] two time-reversal algorithms were introduced for tracking moving targets in clutter. The first algorithm classifies existing scatterers into stationary versus moving targets by means of analyzing multistatic data matrices (MDMs).…”

Threshold-Based Resampling for High-Speed Particle PHD Filter

“…In the research area of state estimation and target tracking [1][2][3][4][5][6], the technique of multiple target tracking (MTT) can estimate the target number as well as each target's state in the scene based on a sequence of uncertain measurements, where the "uncertain" mainly comes from detection uncertainty, association uncertainty and clutters in radar systems [7,8]. Traditional MTT filters, such as multiple hypothesis tracking (MHT) filters, Markov chain Monte Carlo (MCMC) filters and joint probabilistic data association (JPDA) [9], mostly adopt the strategy of measurement-data association to existing tracks and then track each target with a separate filter based on the assumption that each target moves independently.…”