Learning to satisfy actuator networks

无线传感器和执行器网络（SANETs）由多个设备组成，每个设备配备一个低功耗微处理器、一个无线电收发器和一组传感器和/或执行器。执行器是网络中可以主动改变环境的组件，例如，分布式环境控制系统中的加热元件。SANETs代表了传感器网络的重要扩展，因为它们允许网络内的节点根据传感器测量和共享信息做出自主决策并执行动作（驱动）。传感器网络是被动的；SANETs积极地改变环境，以引起期望的反应。许多潜在的应用包括精准农业和药物的本地化交付，但这些网络还处于发展的萌芽阶段。该研究项目的总体目标是为精准农业设计一个SANET。SANET最重要的任务之一是融合感测信息，以了解其环境如何响应驱动。然后，SANET可以生成一个驱动策略，这是一组决策规则，当网络感知到特定的环境条件时，这些规则决定了网络执行的动作。农业环境中的一个例子是根据最近的温度和土壤盐度测量来决定是否使用除草剂。在进行这项研究时，我们将开发一个强大的数学框架来解决SANETs驱动策略的设计。我们将设计健壮的通信协议，允许设备共享其信息，以实现合适策略的分布式学习。

Wireless sensor and actuator networks (SANETs) consist of multiple devices each equipped with a low-power microprocessor, a radio transceiver, and a set of sensors and/or actuators. The actuators are the components of the network that can actively modify the environment - for example, heating elements in a distributed environmental control system. SANETs represent an important extension of sensor networks, as they allow nodes within the network to make autonomous decisions and perform actions (actuation) in response to sensor measurements and shared information. Sensor networks are passive; SANETs actively modify the environment in order to elicit a desired response. The many potential applications include precision farming and localized delivery of medication, but the networks are in the embryonic stage of development.The overall goal of the research project is the design of a SANET for precision farming. One of the most important tasks of a SANET is the fusion of sensed information to development an understanding of how its environment responds to actuations. The SANET can then generate an actuation strategy, a set of decision rules that determine the actions the network performs when it perceives specific environmental conditions. An example from the agricultural setting is the decision whether to administer herbicide based on recent measurements of temperature and soil salinity.  In conducting this research, we will develop a powerful mathematical framework for addressing the design of actuation strategies in SANETs. We will design robust communication protocols that allow the devices to share their information to enable distributed learning of suitable strategies.