面向事件检测的感知数据处理方法研究

Wireless sensor network, as the main sensing layer of Internet of Things, can be seen as a resource-constrained distributed database system, and a significant challenge is to develop reliable and energy-efficient data processing methods to extract useful event information from the distributed sensory data as soon as possible. Most of the existing sensor data processing approaches rely on using raw sensory data, while processing and transmission of these raw sensor data is costly in terms of energy, time and memory. Besides, these methods are difficult to apply in complex network environments. To this end, this research project proposes a series of event detection based sensor data processing methods, such as data presentation, storage, transmission, in-network fusion, and privacy preservation. The main content includes real-time sensor data forwarding protocol, event based kNN and Skyline query processing algorithms, event detection in complex network environments, and sensor data privacy protection approach. Our goal is to get reliable data processing results in an energy-efficient and fast-response way. Nowadays, our country is in the stage of rapid economic growth, environmental pollution and industrial accidents occurred frequently. The proposed methods in this research project have great economic and social value on reducing disaster losses and improving the intelligent management level. If this project would be funded, it brings qualitative help.

作为物联网感知层核心组成部分的无线传感器网络，可以看作为一个资源受限的分布式数据库系统，如何设计节能而可靠的感知数据处理方法，快速地从物联网大数据中获取用户需要的事件信息是一个挑战性工作。当前的感知数据处理技术大都基于不精确的原始感知数据，处理和传输这些原始感知数据需要大量的能耗、时间和内存空间，且难以适用于复杂的网络环境，为解决这些关键问题，本项目拟提出一系列面向事件检测的感知数据表示、存储、传输、网内融合和隐私保护等新方法，包括实时的感知数据传输协议，面向事件检测的kNN和Skyline查询处理算法，以及复杂网络环境下的事件检测和隐私保护策略。总的研究目标是节能而快速地得到可靠的事件检测结果。我国正处于经济高速增长阶段，环境污染事件和安全生产等事故频发，本项目拟提出的新方法，对于减少灾害损失和提升智能化管理水平具有重要的经济和社会价值。如能得到资助，将对本项目有质的帮助。

事件检测是物联网应用的一个主要任务，及时地从边缘感知节点发现正在发生或潜在事件可以有效地降低灾害损失。本项目针对事件检测问题设计了一系列分布式感知数据表示、存储、网内融合、传输和隐私保护等方法。首先研究了并提出了一种新型的物联网复杂事件检测方法，首次基于模糊集理论提出了节能实时的分布式语义信息表示和网内融合方法；其次设计了面向事件检测的kNN和Join物联网复杂查询算法；此外还对物联网IPv6数据包实时数据传输协议进行了研究，并提出了改进方法。基于真实数据集的仿真实验表明了这些方法在兼顾节能实时性和可靠性方面有良好的表现。本项目取得了良好成果，顺利地完成了预期研究目标。