基于移动轨迹大数据的城市交通智能感知研究

Compared with conventional infrastructure based traffic monitoring, the traffic sensing based on mobility trajectory data has major advantages in its high coverage of the urban city, low cost in infrastructure deployment and maintenance, and timeliness in data acquisition and analysis, which together make it become one of the most active research areas in the Internet of Things and Intelligent Transportation. The mobility trajectory data of a city are usually in a large volume, with diverse sensory data types, and produced at a rapid rate. Meanwhile, such a data driven traffic sensing is inherently sparse and dynamic in temporal-spatial dimensions. Due to those challenges, however, existing works and systems thus cannot achieve complete, accurate, and timely traffic information. To tackle with these challenges, in this project we will conduct the research on exploiting big mobility trajectory data for smart sensing of urban traffic, which is built on the study of traffic correlation in the transport network, the relation between data analysis and traffic sensing, as well as some specific knowledges in the transport domain. This proposal focuses on the following three main research topics: 1) dynamic and sparse mobility trajectories based complete traffic information recovery algorithm; 2) co-designed data analysis and crowdsensing based traffic data collection mechanism; 3) high performance graph-parallel computing platform for big mobility trajectory data. This project will finally build an efficient framework on the big mobility trajectory data based traffic sensing from aspects of data collection, data processing, and data analysis. We aim to derive a number of novel research results with 5-7 high quality papers published at top-level international conferences and journals.

相较于传统基于固定传感设施的交通监测，基于移动轨迹数据的交通感知具有成本低、易部署、覆盖范围广等优点，已成为当前物联网和智能交通领域的一个重要研究方向。然而，面对海量、多样和高速产生的移动轨迹数据以及数据驱动的交通感知本身存在的数据稀疏和动态感知问题，目前已有的方法和系统仍难以获得城市范围内完整、准确和实时的交通信息。本项目在深入分析城市交通网络状态相关性的基础之上，充分考虑数据分析和交通感知的关联性和交通领域特定知识，拟开展面向大规模动态稀疏移动轨迹数据的城市交通智能感知研究。主要研究内容包括：1）面向动态稀疏移动轨迹数据的全局交通信息恢复算法；2）基于数据分析与群智感知联动的交通数据采集机制；3）面向移动轨迹大数据的高性能图并行计算框架。本课题研究将为基于移动轨迹数据的交通感知在数据采集、处理与分析等方面形成系统性的理论、技术与方法，在有重要国际影响力的会议和期刊发表学术论文5-7篇。

随着物联网和普适计算等技术的发展，智能设备日益普及并且被广泛部署，可以采集人们在城市范围内的活动轨迹，间接感知城市交通设施与系统的状态，为智能交通应用和服务提供宝贵的实时信息。然而，已有的基于移动轨迹数据的智能交通感知研究普遍存在交通感知范围有限、数据处理效率低下和数据分析精度不高等不足。本项目结合群智感知与图并行计算等技术，开展了基于移动轨迹大数据的城市交通智能感知研究。围绕研究目标和研究内容，本项目取得的主要研究成果如下：(1) 完成了基于压缩感知的全局交通信息恢复算法，可以从稀疏嘈杂的轨迹数据中准确地恢复城市范围的交通状况；(2) 完成了群智感知增强的城市交通感知机制，利用群智感知的思想主动采集高价值交通数据，进一步提高数据恢复的精度；(3) 完成了面向移动轨迹数据的图并行计算框架设计与实现，在并行计算环境下提高了交通建模与数据处理分析的效率。项目执行期间，项目组在IEEE TMC、IEEE T-ITS、IEEE IOT-J等领域权威期刊和IEEE ICDE、ACM MobiSys、DASFAA等国际顶级会议上发表(含录用)高水平论文11篇；申请国家发明专利7项(含已授权2项)；培养研究生6名(其中5人已毕业获得硕士学位)。项目组的研究成果曾获得国际知名会议IEEE ICPADS 2020的“最佳论文奖”，部分研究成果已整理形成英文专著《Mobility Data-Driven Urban Traffic Monitoring》。