SCUB: A Distributed, Utility-based Architecture for Closed Loop Control with Wireless Sensor Networks

This paper develops an architecture for task assignment in wireless sensor networks in support of closed-loop tactical operations. Our approach is based on a mathematical framework, which, based on multi-attribute utility theory, allows either end-users to easily express the value (utility) that they would attach to sensor information based on features associated with the data, such as the geographic scope and relevance, observation type, age, and others aspects of the data that relate to system objectives. We have applied the utility-based framework in the design of a distributed task assignment algorithm for a simple pursuit-evasion (patrol) scenario, which involves (i) a set of soldiers patrolling subregions within a larger Area of Interest (AOI), (ii) intruders that randomly appear on the boundary of and traverse across the AOI, and (iii) a network of finite-energy, wireless sensors reporting tripwire data to the soldiers, queuing them in their mission to pursue and capture the intruders. While utility framework is quite general, the algorithm we have developed is “sensor centric” in the sense that individual sensors must decide whether the detections that they have made warrant dissemination and if so to which endusers. Simulation studies based on this scenario show that in terms of mission-level objectives including both intruder capture rate and system lifetime, utility-based framework for task assignment can significantly improve upon baseline task-assignment framework, in which each sensor indiscriminately reports all data to all available end-users.