文档介绍：BARCLAY SHAW
Detection,
Classification, and
Tracking of Targets
etworks of small, densely distributed wireless the basic ideas on how the CSP algorithms interface with
sensor nodes are being envisioned and devel- working/routing algorithms being developed at
oped for a variety of applications involving Wisconsin (UW-API) [2]. We motivate the framework
Nmonitoring and manipulation of the physical via the problem of detecting and tracking a single maneu-
world in a tetherless fashion [1], [16], [17], [22], [23]. vering target. This example illustrates the essential ideas
Typically, each individual node can sense in multiple mo- behind the integration between UW-API and UW-CSP
dalities but has munication algorithms and also highlights the key
putation capabilities. Many Dan Li, Kerry D. Wong, aspects of detection and localization al-
challenges must be e before gorithms. We then build on these ideas
the concept of works be- Yu Hen Hu, to present our approach to tracking
comes a reality. In particular, there are and Akbar M. Sayeed multiple targets that necessarily re-
two critical problems underlying suc- quires classification techniques.
cessful operation of works: 1) efficient methods Tracking multiple targets via a wireless work
for exchanging information between the nodes and 2) col- is a very challenging, multifaceted problem and several re-
laborative signal processing (CSP) between the nodes to search groups have tackled various aspects of it [3]-[8],
gather useful information about the physical world. [12], [13], [15], [18], [19], [21], [23], [25]. We consider
This article describes the key ideas behind the CSP al- the signal processing aspects of this problem under the
gorithms for distributed works being developed constraints imposed by limited capabilities of the nodes as
at the University of Wisconsin (UW). We also describe well as those associated working and routing.
MARCH 2002 IEEE SIGNAL PROCESSING MAGAZINE 17
1053-5888/0