群智感知计算中任务调度及资源管理关键技术研究

With the proliferation of smart phone terminals and the rapid development of mobile Internet, mobile crowdsourcing has increasingly become an important paradigm for organizing and developing sensing systems. It enables normal mobile terminals to serve as basic mobile sensing nodes to collect sensing data over large-scale environments. By collecting various sensing data from smart mobile terminals, it is able to support a wide spectrum of real-world valuable applications. This proposal will focus on one of the key problems in mobile crowdsourcing, i.e., scheduling of sensing tasks onto mobile terminals and dynamic allocation of resources on mobile terminals. This project will conduct research on four key topics: (1) online incentive mechanism design for mobile crowdsourcing with online arrivals of sensing tasks and arbitrarily dynamic mobile terminals; (2) online scheduling of sensing tasks with time constraints, with the objective of maximizing both power consumption reduction and fairness among terminals; (3) online dynamic control mechanism for mobile crowdsourcing, with the objective of optimizing the system-oriented performance; (4) development of a mobile crowdsourcing platform and a demonstrative application of real-time urban noise mapping. The PI has laid a solid foundation in the area of mobile crowdsourcing, by having conducted extensive research on such topics as sensing data management, static incentive mechanism design, and scheduling of sensing tasks. Related research findings are reported in top conferences such as Infocom and MobiSys, in prestigious journals such as TMC and TPDS. The success of this project will deliver several key technologies and a running demonstrative application system, which will contribute to practical application of mobile crowdsourcing in the real world.

随着智能移动终端的普及和移动互联网的快速发展，群智感知计算逐渐成为一种重要的感知系统组织形式，利用大量普通移动终端作为基本的感知单元，实现感知任务分发与感知数据收集，支撑复杂的感知应用。本项目重点研究群智感知计算的一个核心问题：即感知任务的动态调度及资源动态分配。拟在以下方面深入研究：(1)面向动态感知任务及动态终端节点的在线激励机制；(2)面向带时限感知任务的公平高能效任务调度方法；(3)面向系统级性能优化的在线任务接入控制及资源分配方法；(4)构建城市噪音实时感知示范应用系统。申请人在群智感知计算领域有良好的研究基础，对感知数据处理、静态激励机制设计、感知任务调度等问题进行较深入的研究，研究成果发表于INFOCOM、MobiSys等重要学术会议及TMC、TPDS等重要期刊上。本项目的成功开展将形成若干关键技术及一套实际运行的示范应用系统，为群智感知计算的实用化提供理论及关键技术支持。

本项目围绕群智感知计算感知任务的动态调度及资源动态分配，重点开展了四个方面的研究：（1）面向动态感知任务及动态终端节点的在线激励机制设计与分析，（2）面向带时限感知任务的公平高能效任务调度研究，（3）面向系统级优化的在线任务接入控制及资源配置方法研究，（4）设计实现城市噪音实时感知示范应用系统。本项目的主要成果包括：（1）提出了一种面向移动车辆节点的感知数据汇聚路由算法, （2）针对不同的群智感知网络类型和场景，提出了一系列群智感知网络激励机制，并分析了其性质, （3）针对带时限的感知任务，提出了一种带时限感知任务的调度方法，实现良好的节能性及节能公平性,（4）面向系统性能优化，提出了动态感知任务接入及调度算法,（5）设计了一套群智感知数据的分析和挖掘算法,（6）设计并实现了城市噪音感知示范应用。本项目的研究团队超额完成考核指标。共发表会议论文27篇、国际期刊论文18篇，包括INFOCOM 5篇、KDD 1篇、IJCAI 1篇、IEEE TOC 1篇、IEEE TMC 5篇、IEEE TPDS 2篇等。获授权的国家发明专利13项。培养博士研究生2名、硕士研究生8名。作为TPC Co-Chair，组织了国际学术会议 The 2015 11th EAI International Conference on Collaborative Computing: Networking, Applications and Worksharing。