SENSORS: Robust Event Services for Emergency Response in Sensor Networks

无线传感器网络由成千上万的廉价设备组成，具有显著改善地震、洪水和其他自然灾害以及化学和生物袭击的应急响应系统的潜力。虽然每个节点都必须非常便宜，因此可能具有有限的容量，但节点的总体行为应该共同确定情况，检测可能的幸存者，向最近的救援队发送必要的信息，为他们确定安全的出口路径，并监视该区域的任何后续事件。因为这种紧急情况可能会持续很长时间，并且是高度动态的；系统需要适应不断变化的环境，鲁棒性强，有效利用有限的资源，并满足任何时序要求。这项研究正在为事件和消息传递开发健壮的实时中间件服务。在无线传感器网络中，事件不是二元的，而是基于复杂环境中许多噪声传感器的融合。传感器数据可能丢失、错误或过时。因此，事件服务必须支持实时、数据融合、置信度计算和节能。事件可以发生在非常局部的区域，也可以是全球性的，或者同时发生在系统的许多部分。为了在容量有限的设备上工作，事件服务必须高度分散。该服务还必须尽量减少误报。所有这些特性使得无线传感器网络的事件服务非常具有挑战性。这项工作的最终目标是提供一种工具，允许应用程序指定所需的数据和事件，并及时交付它们。如果这项研究成功，应急响应可以更快、更准确地发生；可以挽救生命，减少损害。

Stankovic - abstractWireless sensor networks, composed of 1000s of inexpensive devices, have thepotential to significantly improve emergency response systems forearthquakes, floods and other natural disasters, as well as forchemical and biological attacks. While each node must be very cheap and,therefore, is likely to have limited capacity, the aggregate behavior of the nodes should collectively determine the situation,detect possible survivors, send necessary information to the nearest rescue teams, identify a safe exit path for them, and monitor any subsequent events of interests in the area. Since such emergencies may last a long time and are highly dynamic; the system needs to adapt to the changing environment, be robust, utilize the limited resources efficiently, and meet any timing requirements. This research is developing robust and real-time middleware servicesfor events and message delivery. In wireless sensornetworks events are not binary, but are based on sensor fusionfrom many noisy sensors in complicated environments. Sensor datamay be missing, wrong, or out of date. Consequently, the event servicemust support real-time, data fusion, confidence calculations, andconserve power. Events can occur in a very local region, be global or occursimultaneously in many parts of the system. The event service must behighly decentralized in order to work on the limited capacitydevices. The service must also minimize false alarms. All of these features makes an event service for wirelesssensor networks very challenging. The ultimate vision ofthis work is to provide a facility that allows the application to specify the desirable data and events, and deliver them in a timely manner. If this research is successful, emergency response can occurmore quickly, and be more accurate; lives can be saved and damagereduced.