CRII: NeTS: Building Quality-of-Information Aware Distributed Sensing Systems

日益强大和负担得起的传感设备遍及世界的每一个角落，这导致了分布式传感系统的出现，从根本上改变了人们与物理世界互动的方式。尽管分布式传感系统带来了巨大的好处，但也带来了新的研究挑战，其中一个重要方面源于传感器节点提供的信息质量与系统和网络资源消耗之间的冲突。一方面，由于观测不完整、背景噪声、传感器质量差等各种原因，单个传感器不可靠。为了解决这个问题，一种可能的解决方案是整合来自观察相同事件的多个传感器的信息，因为这可能会抵消单个传感器的错误。另一方面，分布式传感系统通常具有有限的资源(例如，带宽、能量、存储等)。因此，由于潜在的过度资源消耗，从大量传感器收集数据通常是令人望而却步的。针对这一挑战，本项目试图开发一种资源高效的信息集成框架，该框架可以智能地集成来自分布式传感器的信息，从而在系统资源的约束下获得最高质量的信息。这项研究的成功完成将使依赖分布式传感系统收集、传输和分析传感数据的广泛应用受益。本项目旨在在这一研究领域做出几项贡献。首先，它将开发一种新颖的信息整合算法，该算法可以联合估计每个传感器的QOI，并整合传感器提供的信息。该算法对具有较高QOI的传感器赋予更多的权重，从而比对所有传感器一视同仁的直接积分方法(如平均法和投票法)能够获得更高的精度。其次，针对受限的系统资源带来的挑战，本项目将针对不同类型的分布式感知系统上的数据采集，提出一套服务质量感知的资源分配机制。对于物理传感系统，通常是部署在偏远、恶劣甚至恶劣地点的无线系统，将制定一个优化框架，以最大限度地利用网络带宽和可再生能源，以实现所提供信息的最佳综合质量。对于由人类进行数据收集的人群感知系统，将设计一种新的激励机制来补偿参与者的资源消耗和潜在的隐私泄露，不仅基于用户已经花费的努力，还基于用户可以提供的QOI。

The proliferation of increasingly capable and affordable sensing devices that pervade every corner of the world has given rise to distributed sensing systems that have fundamentally changed people's ways of interacting with the physical world. Despite their tremendous benefits, distributed sensing systems pose great new research challenges, of which one important facet stems from the conflicts between the Quality of Information (QoI) provided by the sensor nodes and the consumption of system and network resources. On one hand, individual sensors are not reliable, due to various reasons such as incomplete observations, background noise, and poor sensor quality. To address this problem, a possible solution is to integrate information from multiple sensors that observe the same events, as this will likely cancel out the errors of individual sensors. On the other hand, distributed sensing systems usually have limited resources (e.g., bandwidth, energy, storage, etc). Therefore, it is usually prohibitive to collect data from a large number of sensors due to the potential excessive resource consumption. Targeting on this challenge, this project seeks to develop a resource-efficient information integration framework that can intelligently integrate information from distributed sensors so that the highest quality of information can be achieved, under the constraint of system resources. Successful completion of the proposed research will benefit a wide spectrum of applications that rely on distributed sensing systems for the collection, transmission and analysis of sensory data.This project aims to make several contributions in this area of research. First, it will develop a novel information integration algorithm that can jointly estimate the QoI of each sensor and integrate the information provided by the sensors. This algorithm puts more weights on the sensors with high QoIs, and thus can achieve improved accuracy than the straightforward integration methods such as averaging and voting that treat all the sensors equally. Second, to address the challenge brought by the constrained system resources, this project will propose a set of QoI-aware resource allocation mechanisms for the data collection on different types of distributed sensing systems. For physical sensing systems that are usually wireless systems deployed at remote, harsh or even hostile locations, an optimization framework will be developed to maximally utilize the network bandwidth as well as renewable energy in order to achieve the optimal aggregate quality of delivered information. For crowd sensing systems where data collections are carried out by a human population, a novel incentive mechanism will be designed to compensate participants' resource consumption and potential privacy breach, based on not only the efforts a user has spent but also the QoI the user can provide.