Horizontal estimation and information fusion in multitarget and multisensor environments.

Abstract

In recent years, there has been a considerable increase in both the variety and number of sensors which needed to be tied together. A new distributed estimation architecture for Distributed Sensors Networks (DSN) is introduced. It is called Horizontal Estimation Architecture (HEA). The term horizontal is used to imply that the geographically dispersed nodes do not differ in rank and are peer-to-peer coupled. Each node is connected by a data link to its neighbors (where possible), thus providing a mesh network topology. The introduced HEA has four major components, the local estimator, the information fusion process (both together are called a horizontal estimator), the network access protocol, and the controller-decisionmaker. The HEA techniques are applied to the solution of Multitarget and Multisensor Tracking (MMT) problems in Track-While-Scan (TWS) systems with an emphasis towards track fusion. A mathematical framework which encompasses the components of the horizontal estimator is developed with an emphasis towards the track fusion algorithm. An artificial intelligence approach using expert systems for track fusion has been presented. Through this HEA application its main features and practical usefulness are addressed.