Open Covid Pledge
Created with Sketch.
Open Covid Pledge
Distributed Multitarget Tracking for Passive Multireceiver Radar Systems
Multitarget tracking is of paramount importance in a wide range of (ground, air, maritime) surveillance tasks. In particular, passive multistatic radar tracking has recently attracted great interest due to the low cost and covert operation of passive radar systems. A passive multistatic radar system typically consists of multiple radar receivers and/or FM radio, or digital TV, transmitters of opportunity geographically dispersed through the surveillance area so that each receiver-transmitter pair can provide bistatic range and/or range rate measurements of the targets of interest. In order to enhance target observability, it is convenient to have multiple radar receivers which can be suitably located around the available transmitter (or transmitters) of opportunity. Further, in order to be able to change the field of view (i.e. the region where a sufficient level of target observability is ensured) the radar receivers should be mobile, e.g. mounted on vehicles. Motivated by the above considerations, the present work assumes a passive multireceiver radar system wherein the radar receiver units are mobile and connected to form a wireless network also with the aid of extra nodes which are not equipped with a radar receiver, i.e. have no sensing capabilities. The two types of nodes, characterized by having or not a radar receiver on-board, will be referred to as sensor or, respectively, communication nodes and will both be assumed to have processing and communication capabilities. The objective is to design a distributed multitarget tracking algorithm running in each (sensor or communication) node of the network and therein ensuring the achievement of a global situation awareness, with no coordination of a central unit as well as no local knowledge about the overall network topology. More specifically, we would like each node to be able, by processing local information and information from connected nodes only, to detect and track all targets of interest within the whole surveillance region. In particular, the proposed approach relies on the the so called Cardinalized Probability Hypothesis Density (CPHD) filtef for performing multitarget tracking without computationally expensive measurement-to-target association procedures, as well as on the well known consensus paradigm for fusing information over the whole network (collective fusion) by means of regional fusions over subnetworks of neighboring (1-hop distant) nodes. The peculiarity of CPHD filtering, as any other random set tracking method, is that it embeds target birth and death processes in the overall multitarget dynamics as well as missed detections and clutter in the overall multisensor dynamics, thus avoiding the resort to ad-hoc procedures for target initialization, termination and measurement-to-target association like in traditional multitarget tracking approaches. Consensus represents an effective methodology for distributed averaging of a given quantity over a network. The idea of consensus is that each node computes the collective average of such a quantity by iterative regional averages, where the terms collective and regional mean over all network nodes and, respectively, over neighboring nodes only.
Battistelli Giorgio, Chisci Luigi, Fantacci Claudio, Farina Alfonso, Graziano Antonio
Paper for Seminar/Symposium/Conference
IRS 2013 - International Radar Symposium (19-21 June 2013, Dresden, Germany)
Enter Search Text