论文部分内容阅读
智慧城市为智能交通管理和交通网络智能应用的发展提供了巨大推动力。近来,智能交通系统(Intelligent transportation systems,ITSs)和移动位置服务(Location-based services,LBSs)也成为了研究领域的热点。交通领域数据量在快速不断增长,云计算在巨量数据的存储、接入、管理和处理方面有着巨大作用。交通领域相当比例的数据为GPS数据,此类数据具有非频繁、含噪声等特性,这使得维护基于GPS的实时交通软件的服务质量较为困难。在诸多智能交通系统应用中,地图匹配处理起着将GPS观测点准确排列于路网中的关键作用。考虑到准确性时,地图匹配策略的性能由两个连续的GPS观测点间的最短路径决定;另一方面,处理最短路径查询(Processing shortest path queries,SPQs)耗费着较高计算量。现有的地图匹配技术采用固定参数(固定的候选点数量,固定的误差圆半径)的办法,这可能导致确认线路分段时产生不确定性,也可导致低精度结果(或需进行大量SPQ处理以保证精度)。此外,由于采样错误的存在,较高采样时间(大于10 s)内的GPS数据常含有冗余数据,这也导致需要额外的SPQ处理。由于SPQ处理导致的高运算量问题,现有的地图匹配策略并不能实现实时应用。在本文中,我们提出一种实时地图匹配方法(Real-time map-matching,RT-MM)。该方法以云计算为基础,是一种全自适应地图匹配策略,能够应对实时GPS轨迹地图匹配中SPQ处理的关键问题。本研究还通过基于虚拟数据和实际数据的仿真,对所述方法与现有方法的性能进行了比较。
Smart cities provide a huge boost to intelligent traffic management and the development of intelligent transport network applications. Recently, intelligent transportation systems (ITSs) and Location-based services (LBSs) have also become a hot area of research. The volume of data in the transport sector is rapidly growing, and cloud computing has a huge role to play in the storage, access, management and processing of huge amounts of data. A considerable proportion of traffic data is GPS data, which is characterized by infrequent, noisy, etc. This makes it more difficult to maintain the quality of service of GPS-based real-time transport software. In many ITS applications, the map matching process plays a key role in accurately arranging GPS observation points on the road network. Taking into account the accuracy, the performance of the map matching strategy is determined by the shortest path between two consecutive GPS observation points; on the other hand, the processing shortest path queries (SPQs) consume a high amount of computation. Existing map matching techniques use fixed parameters (fixed number of candidate points, fixed error radius) which can lead to uncertainties in confirming line segments and can also result in low-precision results (or large amounts of SPQ Processing to ensure accuracy). In addition, GPS data in higher sampling times (> 10 s) often contains redundant data due to sampling errors, which also results in additional SPQ processing. Due to the high computational burden caused by SPQ processing, the existing map matching strategy does not achieve real-time application. In this paper, we propose a real-time map-matching method (RT-MM). This method, based on cloud computing, is a fully adaptive map matching strategy that can handle the key issues of SPQ processing in real-time GPS trajectory map matching. This study also compares the performance of the method with the existing methods by simulating the virtual data and the actual data.