Generating Integrated Interpretation of Partial Information Based on Distributed Qualitative Reasoning

Abstract

Situation assessment (SA) is regarded as a problem solving process that involves the acquiring and integrating of partial information sensed from the world in order to produce a global interpretation. A new approach to SA called distributed sensor network (DSN) has recently been proposed. DSN consists of sensor nodes, each of which has its own knowledge or model to generate a partial interpretation by matching the sensed information with the model. These nodes exchange and share their interpretations with other ~odes to i~tegrate them into a consistent global mterpretation. In this paper, a qualitative model of the world is introduced to each sensor node, since the sensor information is local and partial from the spatial and temporal point of view. Each node can generate a partial interpretation which can predict a future evolution after the observed event, by the state transition on the qualitative model. The partial interpretation thus generated is represented as an envisioning tree. In order to integrate the several partial interpretations into a global one, an integration node is introduced which connects the envisioning trees by pruning inconsistent branches under global perspectives, to achieve a spatial and temporal interpolation between the sensed information.