Title

Multisensor data association and resource management for target tracking

Date of Completion

January 1998

Keywords

Engineering, Aerospace|Engineering, Electronics and Electrical|Engineering, System Science

Degree

Ph.D.

Abstract

The central theme of this research is how to track the motion of one or more moving targets using noisy measurements from one or more remote sensors (radar, sonar or electro-optical), while making the best possible use of available resources. In this multisensor-multitarget problem, there are four fundamental questions from a tracking perspective: (1) How to determine from which target a measurement originated (data association). (2) How to combine the measurements from different sensors (sensor fusion). (3) How to use the measurements to obtain the best possible state estimates (estimation). (4) How to minimize the radar resources required to obtain the measurements, while not sacrificing performance (resource management).^ In this dissertation, four different, but interconnected, tracking problems, each of which raises one or more of the above questions, are addressed in order to develop sensor management, data association and estimation algorithms.^ In the first part of the dissertation, the development of an estimation/sensor management algorithm for tracking highly maneuvering aircraft, while minimizing the required radar resources, is presented. The proposed algorithm, which uses the Interacting Multiple Model (IMM) estimator in conjunction with the Probabilistic Data Association (PDA) algorithm, also handles the electronic counter measures employed by the enemy aircraft. In the next part, data association in a closely-spaced target tracking problem, where the measurements from two different targets are distinguishable from one another, is considered. A new iterative optimization-based algorithm, which can handle the apparent overlapping of the measurements is developed.^ In the third part of the dissertation, the problem of tracking ground targets, whose motion is constrained by external factors like road networks and terrain conditions, is considered. The issue here is how to handle the added uncertainty (constraint) that can vary with time as well as across targets. To solve this, a variable structure multiple model estimator, where the estimator structure is adaptively modified for each track depending on the terrain topography at its predicted location, is developed. Finally, the efficient use of higher dimensional data association algorithms in dense scenarios, where the contention among tracks for measurements is high, is considered in the fourth part of this dissertation. A procedure, which guarantees maximum effectiveness for each sensor without sacrificing the fusion across sensors, is presented. ^