A weighting/threshold approach to sensor fusion
A weighting/threshold-based sensor fusion algorithm to decrease the false alarm rate (FAR) while maintaining a high probability of detection (PD) is being tested in the Remote Security Station (RSS). The RSS is being developed to provide temporary intrusion-detection capability on short notice. It consists of a portable, multisensor pod connected by cable to a manned control console. The pod is set up outdoors in the location that security is needed; the console and operator are located in a command bunker up to a kilometer away. The RSS software filters out alarms from low-believability sensors and also filters out alarms in low-priority areas. Each sensor's believability is proportionally enclosed as a weighing, which is continually updated as a function of the environmental conditions affecting that sensor. Area priority is proportionally encoded as a threshold value for each pie-wedge area around the pod. When an event in an area triggers one or more sensors, their weightings are summed and then compared to the area threshold value. The operator is informed of the event only if the summed weighting exceeds the threshold. Extensive field testing has not yet been done, but some results show the current sensor fusion algorithm decreases the FAR at the expense of lowering the PD. To increase the PD while retaining a low FAR, the weighting/threshold algorithm will be modified to use temporal data and pattern recognition. 4 refs., 2 figs.
- Research Organization:
- Sandia National Labs., Albuquerque, NM (USA)
- DOE Contract Number:
- AC04-76DP00789
- OSTI ID:
- 5096681
- Report Number(s):
- SAND-88-0094C; CONF-8804128-1; ON: DE88010349
- Resource Relation:
- Conference: 1. annual national symposium on sensor fusion, Orlando, FL, USA, 5 Apr 1988
- Country of Publication:
- United States
- Language:
- English
Similar Records
The Remote Security Station (RSS) final report
The Remote Security Station (RSS) final report