Dresden 2003 – wissenschaftliches Programm
Bereiche | Tage | Auswahl | Suche | Downloads | Hilfe
AKSOE: Physik sozio-ökonomischer Systeme
AKSOE 4: Mikro-ökonomische Modelle und Multi-Agenten-Systeme
AKSOE 4.6: Vortrag
Dienstag, 25. März 2003, 12:00–12:30, BAR/205
Target Assignment by Physical Modelling — •Johannes Schneider1,2 and Scott Kirkpatrick2 — 1School of Engineering and Computer Science, The Hebrew University of Jerusalem, Givat Ram, Jerusalem 91904, Israel — 2Fachbereich Physik, Universität Mainz, Staudinger Weg 7, 55099 Mainz, Germany
A system of locally connected sensors has the task of tracking moving targets which pass by. We consider a system of 100 sensors which are deployed in the field by placing them close to an underlying grid. We consider square, honeycomb and triangular lattices, each with a nearest neighbor distance of 10 units. Each sensor is displaced from this lattice randomly by up to ± 1 unit in both the N-S and E-W axes. The probability that two sensors have a communications link depends on distance. It is close to unity for nearest neighbors, roughly 50% for second neighbors, and falls off exponentially beyond 19 units.
Targets appear from west and fly a twisting path through the sensor array. If a target leaves the area, a new target appears from west to keep the number of targets constant. Sensors can only detect targets within some range. Various range values R (R=15, 20, 30, 40) are considered. No sensor can track more than one target. Each sensor has three fixed 120 degree antenna sectors. At the start of the simulation, each sensor is given a random orientation, and subsequently they do not turn.
Our job is to ensure that each target is tracked by at least three communicating sensors for long enough to do something about it. Our model works well within a broad range of parameters and scales roughly linearly to larger arrays of sensors.