Resource management of task oriented distributed sensor networks

Abstract

In this paper we provide a foundation for a unified analysis of both the decision fusion and congestion avoidance aspects of a task-oriented distributed sensor network (DSN). Such a framework allows network resource management to be carried out in a manner that is sensitive to the overall objectives of the DSN rather than decoupling them via perhaps simple fairness strategies. The proposed approach associates importance measures related to the degradation and relevance of incoming data lines at each network node. These are then used to carry out intelligent resource management and avoid congestion in a manner which is implicitly related to the decision objectives of the DSN and explicitly related to the resource availability. To achieve this, we propose a 'per-flow' technique that decouples data flow among nodes at different hierarchical levels of the DSN. The resulting framework allows seamless integration of the importance measures providing resource management decisions that are sensitive to the overall DSN objectives.