ITR/SI+IM (CISE):Distributed Data Compression and Dissemination for Wireless Sensor Networks

该项目将开发新的方法，有效传播无线传感器网络收集的数据。随着小型、低成本微电子和微机电传感器的最新和预计的进展，很容易设想，分布在适当区域的大型传感器阵列将能够测量重要属性场的空间和时间变化，例如温度、湿度、声音、光、气体浓度等。然而，为了实现这种阵列的好处，无线通信网络必须设计成用低功率编码和传播它们产生的大量数据。以此为目标，该项目将为密集的传感器阵列开发分布式数据压缩和数据传播的新方法，也就是说，对于传感器如此接近以至于它们的测量高度相关的阵列。该项目的一个论点是，可以利用这种阵列中的相关性，以使网络基本上像稀疏传感器网络一样有效地运行，同时具有对传感器故障有弹性的额外优点，并允许自适应地测量属性字段或具有更高的空间分辨率。另一个论点是，这种网络的数据压缩和传播问题是深深交织在一起的。因此，该项目的重点是联合设计。例如，它寻求方法，剪裁分布式数据压缩方法，以partcultardissemination策略，反之亦然。在这个过程中，提出了学习传感器网络的性能如何取决于各种问题，如传感器的数量和位置.

This project will develop new methods for the efficient dissemination of the data collected by wirelesssensor networks. With recent and projected advances in small, low cost microelectronic and micro-electromechanical sensors, it is easy to envision that a large array of sensors, distributed over anappropriate region, will be able to measure the spatial and temporal variations of important attributefield such as temperature, moisture, sound, light, gas concentrations, etc.. However, to realize thebenefits of such arrays, wireless communication networks must be devised that with low power encodeand disseminate the large amounts of data they generate. With this as the goal, the project willdevelop new methods of distributed data compression and data dissemination for dense sensor arrays,that is, for arrays whose sensors are so close that their measurements are highly correlated. One thesisof this project is that the correlations in such arrays can be exploited in order to make the networkoperate essentially as efficiently as a sparse sensor network, while having the additional advantages ofbeing resilient to sensor failures and permitting the attribute field to be measured adaptively or withhigher spatial resolution. Another thesis is that the data compression and dissemination issues forsuch networks are deeply intertwined. Accordingly, the project focuses on the joint design of such.For example, it seeks methodology for tailoring distributed data compression methods to partculardissemination strategies, and vice versa. In the process, it proposed to learn how the performance ofsensor networks depends on a variety of issues such as the number and placement of sensors.