Detection, identification, and tracking (DIT) problems that arise in applications such as perimeter surveillance for suspected group activities or detection/assessment of radiation and chemical dispersions, require capabilities beyond those of typical sensor and cyber network modalities taken individually. They require sensor-cyber (SC) networks: sensor networks provide information about the physical location, proximity, and movements of targets or sources, and cyber networks provide data and computational resources to implement sophisticated methods to characterize, align, and fuse the sensor information for analysis and decision making. In general, SC networks may comprise sophisticated sensor, computational, and storage nodes; they collect, store, and transfer data of different modalities such as visual, mechanical, chemical, radiation, text, and images. Such networks can be used for the rapid identification and detection of chemical or biological plumes generated by explosions or leakages. An initial detection of such plumes may be carried out by using a network of detectors or sensors. However, a more detailed analysis needed for tracking and prediction requires utilization of complex plume dynamics, which in turn requires powerful computers and databases that may be remotely located. Furthermore, on-line monitoring, tracking, and active sensing for these tasks require an interactive access by the control centers consisting of human experts/analysts. In addition to a conventional sensor network, a cyber network is necessary to connect various computational and data sources. This class of DIT problems in SC Networks (DITSCN) represents a significant challenge by requiring approaches that extend beyond the conventional disciplinary boundaries of sensor and cyber networks. In sensor networks, the underlying phenomena are typically governed by the continuous time-space principles based on constraints such as physical movements; these principles have been exploited extensively in solving identification and tracking problems. On the other hand, the computations and data processing in cyber networks are mostly governed by discrete principles, which are often made use of in solving search and identification problems. These two approaches are based on different principles and structures, which must be unified and integrated to perform DIT tasks that span an SC network.
|
