IEEE Trans Cybern. 2019 Feb;49(2):719-725. doi: 10.1109/TCYB.2017.2766451. Epub 2017 Nov 9.
We extend prior results on a single decision maker opportunistic sensing problem to a distributed, multidecision maker setting. The original formulation of the problem considers how to opportunistically use "in-flight" sensors to maximize target coverage. In that paper, the authors show that this problem is NP-hard with a strong polynomial heuristic for a single decision maker. This paper extends this by considering a distributed decision making scenario in which multiple independent parties attempt to simultaneously engage in opportunistic sensor assignment while managing interassignment conflict. Specifically, we develop an algorithm that: 1) produces a Pareto optimal opportunistic sensor allocation; 2) requires fewer bits of communicated information than a completely centralized deconfliction approach; and 3) runs in distributed polynomial time once the individual decision makers identify their preferred (optimal) sensor allocations. We validate these claims using appropriate simulations.
我们将单一决策者机会感知问题的先前结果扩展到分布式多决策者设置。该问题的原始表述考虑了如何利用“飞行中”的传感器来最大限度地提高目标覆盖率。在那篇论文中，作者表明对于单个决策者，这个问题具有很强的 NP 难特性，并提出了一种强多项式启发式算法。本文通过考虑分布式决策制定场景来扩展这个问题，其中多个独立的方试图同时进行机会主义传感器分配，同时管理分配之间的冲突。具体来说，我们开发了一种算法：1）产生帕累托最优的机会主义传感器分配；2）与完全集中的去冲突方法相比，需要更少的通信信息位；3）一旦各个决策者确定了他们的首选（最优）传感器分配，就可以在分布式多项式时间内运行。我们使用适当的模拟来验证这些主张。
